SWAT TO IDENTIFY WATERSHED MANAGEMENT...

121
SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONS: (ANJENI WATERSHED, BLUE NILE BASIN, ETHIOPIA) A Thesis Presented to the Faculty of the Graduate School of Cornell University in Partial Fulfillment of the Requirements for the Degree of Master of Professional Studies By Biniam Biruk Ashagre August 2009

Transcript of SWAT TO IDENTIFY WATERSHED MANAGEMENT...

Page 1: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONS:

(ANJENI WATERSHED, BLUE NILE BASIN, ETHIOPIA)

A Thesis

Presented to the Faculty of the Graduate School

of Cornell University

in Partial Fulfillment of the Requirements for the Degree of

Master of Professional Studies

By

Biniam Biruk Ashagre

August 2009

Page 2: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

© 2009 Biniam Biruk Ashagre

Page 3: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

ABSTRACT

Ethiopia is known for its wealth of natural resources. These result in part from extreme

elevation variation. However, 5,000 years of land cultivation have degraded large

areas of the natural environment. Soil erosion affects 82% of the country. The rich

highland soil, which supports 80% of the total population, only covers 45% of the

country. In these highlands the soil is becoming less fertile; droughts are more

frequent and intense; and water resources are declining, due in part to the soil erosion.

The Anjeni watershed is located in the highlands in the Blue Nile Basin with an annual

soil loss of 18.33 tons/year/ha.

The existence of soil erosion in a watershed is an indication of unsustainable land

management practices. The objective of this study was to formulate sustainable land

management options that alleviate soil erosion in the Anjeni watershed. The SWAT-

WB model that simulates saturation excess flow was applied, and the result showed

that the Anjeni watershed is dominated by saturated excess flow from the shallow soils

rather than infiltration excess flow. The conventional SWAT model uses the SCS-

curve number method which considers only infiltration excess flow. In contrast, the

SWAT-WB model simulates saturation excess flow in order to determine surface

runoff. Hence, SWAT-WB was used to investigate the flow and sediment processes in

the watershed and to compare different potential land management options to alleviate

soil erosion.

The model SWAT-WB was calibrated for flow and performed well with a coefficient

of determination (R2) of 0.92 and Nash-Sutcliffe coefficient (ENS) of 0.91. The model

also performed well in simulating soil erosion on a monthly basis with the coefficient

Page 4: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

of determination of 0.56 and the Nash-Sutcliffe coefficient of 0.55. The relatively

poorer performance of the model in simulating soil erosion can be attributed to a gully

in the watershed possibly contributing 30% of the annual soil loss from the watershed.

Model simulation suggests that the existing terraces are saving 2,046 tons/year of soil

loss. If further terraces are constructed, they could save an additional 932 tons/year.

Forestation of degraded areas and bush lands was found to reduce soil erosion by 333

tons/year. Zero-tillage technique for all fields except those covered with teff in the

watershed reduces erosion by only 45 tons/year. If gully rehabilitation work with a

90% erosion control practice is implemented in gullies, an additional 300 tons/year

would be saved. Combining foresting degraded lands and bush lands with

rehabilitation of gullies in Anjeni watershed is predicted to reduce soil loss from the

watershed by 630 tons/year. The impact of further construction of terraces on

productivity and its effect on the overall hydrological balance should be

experimentally investigated before being implemented and if it shows a significant

change, it can be practiced with some measures and innovations on the water

availability during the dry season.

Page 5: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

iii

BIOGRAPHICAL SKETCH

Personal Details Name Sex Date of Birth Nationality Email

Biniam Biruk Ashagre Male 28-Feb-09 Ethiopian [email protected]

Education

Institute Degree Duration Year

conferred Field of Study Arba Minch University B. Sc.

09/2000 - 07/2005 23/07/2005 Civil Engineering

Professional Experience

Employer Duration Position Arba Minch University

23/07/2005 up to date

Lecturer in Civil Engineering Department

Major research Interest

Water, Climate change and Environmental Sustainability

Current research Using SWAT to Identify Watershed Management Options:

(Anjeni Watershed, Blue Nile Basin, Ethiopia)

Page 6: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

iv

This study is dedicated to my beloved wife and my lovely family

Page 7: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

v

ACKNOWLEDGMENTS

I am grateful to International Water Management Institute (IWMI) for financial

support for this study and for providing with an office and facilitating the study needs.

I would like to acknowledge the Soil Conservation Research Program (SCRP) office

in Addis Ababa and the Amhara Region Agricultural Research Institute in Adit, for

their invaluable provision of information and data.

This thesis is a testimony to the professional and material help and comments from

Professor Tammo Steenhuis, Dr. Zach Easton, Dr. Matthew McCartney, and Eric D.

White. Thanks to all those people who helped see this thesis to completion, they are

too numerous to mention.

Without the help of Dr. Amy Collick, it would have been very difficult to get data

from different offices.

Last but not least, thanks go to my lovely wife for comments and encouragements. I

would also like to thank my family and my friends for their moral support.

Page 8: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

vi

TABLE OF CONTENTS

BIOGRAPHICAL SKETCH.........................................................................................iii

ACKNOWLEDGMENTS..............................................................................................v

LIST OF FIGURES.......................................................................................................ix

LIST OF TABLES .......................................................................................................xii

LIST OF ABBREVIATIONS .....................................................................................xiii

CHAPTER ONE.............................................................................................................1

INTRODUCTION..........................................................................................................1

CHAPTER TWO............................................................................................................5

STUDY AREA............................................................................................................... 5

Location .........................................................................................................................5

Climate...........................................................................................................................6

Hydrology...................................................................................................................... 6

Land use and Soil Conservation..................................................................................7

Geology and Soil ...........................................................................................................9

CHAPTER THREE ......................................................................................................11

METHODS...................................................................................................................11

SWAT Model Description..........................................................................................11

Surface Runoff ..............................................................................................................13

Soil Water .....................................................................................................................15

Lateral Flow .................................................................................................................17

Percolation and Ground Water Return Flow ...............................................................19

Sediment .......................................................................................................................22

Model Input.................................................................................................................26

Digital Elevation Model ...............................................................................................27

Page 9: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

vii

Land Use Map ..............................................................................................................28

Soil Map and Data........................................................................................................30

Weather Data................................................................................................................31

River Discharge and Sediment Yield ............................................................................34

Management Practices and Scenarios .........................................................................36

Tillage Activity..............................................................................................................37

Model Setup ................................................................................................................37

Watershed delineation ..................................................................................................37

HRU Definition.............................................................................................................38

Weather Data Definition ..............................................................................................40

Management Practices .................................................................................................40

Model Sensitivity Analysis, Calibration and Validation ..............................................41

Scenarios ......................................................................................................................47

CHAPTER FOUR ......................................................................................................50

RESULTS AND DISCUSSIONS...............................................................................50

Sensitivity Analysis.....................................................................................................50

Model Calibration and Validation ............................................................................51

Analysis of Results...................................................................................................... 59

Gully Erosion ..............................................................................................................65

Spatial Variation of Runoff and Soil Erosion ..........................................................67

Scenarios...................................................................................................................... 69

Flow..............................................................................................................................69

Sediment.......................................................................................................................72

CHAPTER FIVE ........................................................................................................77

CONCLUSIONS AND RECOMMENDATIONS ...................................................77

APPENDIX .................................................................................................................91

Page 10: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

viii

Appendix I: Parameters in SWAT database for each crops in the watershed .....91

Appendix II: Parameters in SWAT database for each soil layers in the watershed

......................................................................................................................................92

Appendix III: Parameters in SWAT database for Urban land uses in the

watershed..................................................................................................................... 95

Appendix IV: A. Sliding of sides of gullies ...............................................................96

Appendix IV: B. Soil piping in channel sides and springs in the watershed.........97

Appendix IV: C. Soil piping in gullies and side sliding...........................................98

Appendix V: Parameters used for Weather Generator in SWAT Model ............99

Appendix VI: Observed and Simulated Flow and Sediment loss in Calibration100

Appendix VII: Observed and Simulated Flow and Sediment loss in Validation103

Page 11: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

ix

LIST OF FIGURES

Figure 1: Location of Anjeni watershed.........................................................................5

Figure 2: Percentage of land use of Anjeni watershed based on the recorded land uses

on field investigation (method is provided in chapter three)..........................................7

Figure 3: Graded bund (a) and graded fanya juu bund (b) .............................................8

Figure 4: Conservation practices in the watershed......................................................... 9

Figure 5: Figure 7: Soil Water Characteristics Curve

(http://www.dpi.vic.gov.au/dpi/vro/gbbregn.nsf/pages/soil_hydraulic_pdf/$FILE/Tech

Reportch02.pdf .............................................................................................................16

Figure 6: The digital elevation model of Anjeni watershed ......................................... 28

Figure 7: Google earth image of Anjeni watershed with 20 different control points...29

Figure 8: Land use map of Anjeni using Google image and the recorded land use on

each plot........................................................................................................................29

Figure 9: Soil map of Anjeni watershed (based on FAO classification) ......................31

Figure 10: Land use as reclassified by SWAT into the four letter land use code ........38

Figure 11: Soil Map of Anjeni Watershed as reclassified into four letters soil name in

SWAT database ............................................................................................................39

Figure 12: Newly constructed Fanya Juu terrace (a) and Fanya Juu after five years of

construction (b). * Pictures taken from:

http://www.iwmi.cgiar.org/africa/west/projects/Adoption%20technolgy/rainwaterharv

estin/50-Fanya%20juu.htm...........................................................................................48

Figure 13: Coefficient of determination for simulated flow in calibration on a daily

basis ..............................................................................................................................55

Figure 14: Coefficient of determination for simulated sediment in calibration on a

daily basis .....................................................................................................................55

Page 12: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

x

Figure 15: Coefficient of determination for simulated flow in validation on a daily

basis ..............................................................................................................................56

Figure 16: Coefficient of determination for simulated sediment in validation on a daily

basis ..............................................................................................................................56

Figure 17: Coefficient of determination for simulated flow in calibration on a monthly

basis ..............................................................................................................................57

Figure 18: Coefficient of determination for simulated sediment loss in calibration on a

monthly basis................................................................................................................57

Figure 19: Coefficient of determination for simulated flow in validation on a monthly

basis ..............................................................................................................................58

Figure 20: Coefficient of determination for simulated sediment loss in validation on a

monthly basis................................................................................................................58

Figure 21: Hydrograph of the observed and simulated flow from the watershed for the

validation period on a daily basis .................................................................................61

Figure 22: Comparison of observed and simulated sediment loss from the watershed

for the validation period on a daily basis......................................................................61

Figure 23: Hydrograph for Anjeni watershed on a daily basis for the whole period of

calibration showing precipitation, flow from the watershed and sediment loss in a

daily basis for the observed and simulated values........................................................62

Figure 24: Comparison of observed and simulated sediment loss on a daily basis from

Anjeni watershed for the whole calibration period ...................................................... 62

Figure 25: Hydrograph of the observed and simulated flow from the watershed for the

calibration period on a monthly basis...........................................................................63

Figure 26: Comparison of observed and simulated sediment loss from the watershed

for the calibration period on a monthly basis ...............................................................63

Page 13: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xi

Figure 27: Hydrograph of the observed and simulated flow from the watershed for the

validation period on a monthly basis............................................................................64

Figure 28: Comparison of observed and simulated sediment loss from the watershed

for the validation period on a monthly basis ................................................................64

Figure 29: Location and comparison of the largest gully in the watershed for different

years..............................................................................................................................65

Figure 30: Map of extent of surface runoff in each HRU ............................................68

Figure 31: Map of extent of sediment loss from each HRU.........................................69

Figure 32: Comparison of average monthly flow in each scenario..............................71

Figure 33: Pattern of decrease or increase in flow in different scenarios compared to

the base scenario...........................................................................................................71

Figure 34: Annual average water yield form the watershed in each scenario..............72

Figure 35: Hydrograph in each scenario on a monthly basis (see table in Appendix VI)

......................................................................................................................................73

Figure 36: Sediment yield from the watershed in each scenario on a monthly basis (see

table in Appendix VI)...................................................................................................73

Figure 37: Comparison of Average monthly sediment loss from the watershed in each

scenario.........................................................................................................................74

Figure 38: Pattern of decrease or increase in sediment loss from the watershed in

different scenarios compared to the base Scenario.......................................................76

Figure 39: Annual average water yield form the watershed in each scenario..............76

Page 14: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xii

LIST OF TABLES

Table 1: Percentage of area of soil cover in the Anjeni watershed based on the map

prepared by (Gete Zeleke, 2000) ..................................................................................10

Table 2: Crop calendar used in the Anjeni watershed..................................................37

Table 3: Slope descritization used for creation of HRUs ............................................. 40

Table 4: Review of Calibration of Parameters by variable used by SWAT modelers .42

Table 5: Parameters used for sensitivity analysis in this study .................................... 42

Table 6: Parameters used for Calibration .....................................................................43

Table 7: The most sensitive parameters for flow and sediment ...................................51

Table 8: Output variables; simulated before calibration and observed values in mm..52

Table 9: Values of parameters used for calibration......................................................52

Table 10: Annual average output variables; simulated after calibration and observed 54

Table 11: Coefficient of determination and the Nash – Sutcliffe Coefficients for

calibration and validation both in daily basis and monthly basis .................................54

Table 12: Number of farmers interviewed for identification of the process of runoff

production.....................................................................................................................59

Table 13: Area of the gully in m2 for the three years ...................................................66

Table 14: Calculation of mass of soil loss due to gully erosion for the three different

years..............................................................................................................................67

Page 15: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xiii

LIST OF ABBREVIATIONS

ADJ_PKR: Peak Rate Adjustment Factor

AGNPS: The Agricultural Non-Point Source

AJDYCA: Anjeni-Dystric Cambrisols

AJEURE: Anjeni-Eutric Regosols

AJHAAC: Anjeni-Haplic Acrisols

AJHAAL: Anjeni-Haplic Alisols

AJHALX: Anjeni-Haplic Lixisols

AJHANI: Anjeni-Haplic Nithisols

AJHUAL: Anjeni-Humic Alisols

AJHUNI: Anjeni-Humic Noithisols

AJLILE: Anjeni-Lipthic Leptosols

AJVELU: Anjeni-Vertic Luvisols

ALFA: Alfa Alfa

ALHA_BF: Baseflow Alpha Factor

ArcGIS: Suit consisting of Geographical Information System software products

produced by ESRI

ArcSWAT: Soil and Water Assessment Tool version compataple with ArcGIS

AWC: Available Water Capacity

BARL: Spring Barley

CDE: Center for Development and Environment

CFRG: Coarse Fragment Factor

CH_CV: Channel Cover Factor

CH_EROD: Channel Erodibility

CN: Curve Number

Page 16: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xiv

CN-ASMC: Curve Number – Antecedent Soil Moisture Condition

CORN: Corn

CREAMS: Chemical Runoff, and Erosion from Agricultural Management System

CUSLE: Universal Soil Loss Equation Land Cover and Management Factor

DEM: Digital Elevation Model

EDC: Effective Depth Coefficient

ENS: Nash Sutcliffe Efficiency Coefficient

EPIC: Erosion-Productivity Impact Calculator

ESRI: Environmental Systems Research Institute

FAO: Food and Agriculture Organization

FC: Field Capacity

FLAX: Flax

FRST: Forest-Mixed

GIS: Geographical Information System

GLEAMS: Ground Water Loading Effects on Agricultural Management Systems

GW_REVAP: Groundwater "Revap" Coefficient

GWQ: Groundwater Discharge

GWQMN:

HRU: Hydrologic Response Unit

ISED_DET: Code governing calculation of daily maximum half-hour runoff

KUSLE: Universal Soil Loss Equation Soil Erodability Factor

LSUSLE: Universal Soil Loss Equation Topographic Factor

MUSLE: Modified Universal Soil Loss Equation

NRCS: Natural Resource Conservation Services

PET: Potential Evapo-Transpiration

PRF: Peak rate adjustment factor

Page 17: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xv

PRMS: Precipitation-Runoff Modeling System

PUSLE: Universal Soil Loss Equation Support Practice Factor

R2: Coefficient of Determination

RECHRG_DP: Deep Aquifer Percolation Fraction

REVAPMN: Threshold depth of water in the shallow aquifer for "revap" or

percolation to the deep aquifer to occur

RNGB: Range-Brush

RNGE: Range-Grasses

SCRP: Soil and Water Conservation Research Program

SCS: Soil Conservation Service

SLOPE: Average Slope Steepness

SLSUBBSN: Average Slope Length.

SOL_AWC: Soil Layer Available Water Capacity

SOL_BD: Soil Moist Bulk Density

SOL_K: Soil Saturated Hydraulic Conductivity

SOYB: Soya Bean

SPCON

SPEXP

SURLAG: Surface Runoff Lag Coefficient

SURQ: Surface Runoff

SWAT: Soil and Water Assessment Tool

SWAT-WB: Soil and Water Assessment Tools-Water Balance

SWC: Soil and Water Conservation

SWRRB: Simulator for Water Resources in Rural Basins

TEFF: Teff

USAID: United Sates Agency for International Development

Page 18: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

xvi

USLE: Universal Soil Loss Equation

USLE_K: Universal Soil Loss Equation Soil Factor

USLE_P: Universal Soil Loss Equation Management Factor

WEPP: Water Erosion Prediction Project

WP: Wilting point

WWHT: Winter Wheat

WYLD: Water yield

Page 19: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

1

CHAPTER ONE

INTRODUCTION

Watershed management strategies are critical to efficiently utilize the natural resource

base while maintaining environmental quality. Of the many at risk resources in the

Ethiopian highlands, soil and water are arguably the most critical. Nearly 85% of the

population depends on subsistence agriculture. One process that threatens the resource

base is soil erosion. Studies have shown that in Ethiopia billions of tons of soil are

lost annually. The average annual rate of soil loss in Ethiopia is estimated to be 12

tons/hectare/year with losses as high as 300 tons/hectare/year (USAID, 2000). In the

Ethiopian highlands, in particular, soil erosion is a major problem with an estimated

loss of 16-50 ton/hectare/year (Abegaz Gizawchew, 1995).

When compared to other regions, the Ethiopian highlands have the highest levels of

soil loss (Fournier 1960, Walling, 1984). The highland areas, which are rural and each

household is dependent on a low level of agricultural productivity, cover 45% of the

country. Due to the potentially high productivity of the region nearly 80% of the total

population lives in the highlands (Patterson, 2007). The level of agricultural

productivity in these areas is highly influenced by erratic and unpredicted rainfall in

addition to degradation of resources, such as soil. Resource degradation, particularly

soil degradation in the form of nutrient depletion, is an important factor in the decline

in the country’s agricultural production (Bekele and Holden, 1998). Therefore,

management techniques practiced to conserve soil are not only related to the

conservation of natural resources but also to the sustainable development of the

agricultural sector. The existence of high rate of soil erosion in a watershed can be

Page 20: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

2

taken as an indication of unsustainable land management practices (Herweg and

Stillhardt, 1999).

In order to formulate management options, soil erosion must be considered. Soil loss

from a watershed can be estimated based on an understanding of the underlying

hydrological process in a watershed, climatic conditions, landforms and soil factors.

One option for formulating management options is to use models to elucidate

processes controlling the hydrologic and sediment fluxes. Assessing and mitigating

soil erosion at the watershed level is complex both spatially and temporally. Soil type,

depth, and location, land cover type and management, topology and other factors make

the watershed a complex system where hydrologic and erosive process may differ

greatly over a small spatial scale. Erosion rates depend on the rainfall intensity and

the total amount of precipitation after the onset of the rainy season thus adding a level

of temporal complexity to the system. Hence, watershed models that are capable of

capturing these processes in a dynamic manner can be used to provide an enhanced

understanding of the relationship between hydrologic processes,

erosion/sedimentation, and management options. There are many models that can

continuously simulate stream flow, erosion/sedimentation, or nutrient loss from a

watershed. However, few models have been developed or tested in the monsoonal

climates of Africa. The Soil and Water Assessment Tool (SWAT) (Arnold et al.,

1998) has recently been adapted to more effectively model hydrological processes in

monsoonal climates such as Ethiopian (White et al., 2008).

Some models developed in temperate regions have been tested in Ethiopia but many

have inherent weaknesses, largely because they were not developed in the country.

The Agricultural Non-Point Source (AGNPS) model was tested on the highlands of

Page 21: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

3

the Augucho watershed but the outputs from this model did not accurately simulate

pattern of observed runoff (Haregeweyn and Yohannes, 2003). Haregeweyn and

Yohannes (2003) recommended the integration of AGNPS with Geographical

Information Systems (GIS) for increased efficiency and for the model to handle large

and varied types of data. The Water Erosion Prediction Project (WEPP) erosion model

over-predicts soil loss (Zeleke, 2000). The Precipitation-Runoff Modeling System

(PRMS) was tested by Legesse et al. (2003) for South Central Ethiopia but the model

required extensive calibration to predict monthly runoff. PRMS does not route on a

daily basis thus is limited to analyses where longer time steps are appropriate. In

SWAT, GIS and other interface tools can be used to support the input of topographic,

land use and soil data. SWAT can be easily calibrated and can run on a daily basis and

can easily incorporate changes in land use.

In order to choose a model for a particular watershed, the following factors should be

considered: the level of application, purpose, required accuracy, space and time scale,

and availability of data (Decoursey and Selly, 1988). SWAT is suitable in the

following context:

• Watersheds with no monitoring data can be modeled

• The relative impact of alternative input data (e.g. change in management

practices, climate, vegetation, etc) on water quality and other variables of

interest can be quantified

• A variety of management strategies can be modeled without excessive

investment in time or money.

• Enables user to study long-term inputs (Neitsch et al 2005)

Thus, SWAT was selected to model the hydrological processes and estimate the soil

loss from the Anjeni watershed.

Page 22: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

4

This study used SWAT to identify areas of the watershed highly affected by soil

erosion and target these areas for soil erosion control measures. SWAT was also used

to select the best management options to minimize soil loss. In order to formulate

management options, the following objectives were identified:

1. To simulate seasonal and long term sediment yields using SWAT

2. To conduct sensitivity analysis in order to identify hydrological parameters

that most influence surface runoff, base flow, and soil erosion in the watershed

3. To perform calibration and validation for flow and sediment at the outlet of the

watershed

4. To analyze spatial variation of runoff and soil erosion in the watershed. This

helps to identify the sub-watersheds or areas that contribute most sediment to

streams

5. To evaluate the effectiveness of existing Soil and Water Conservation (SWC)

structures in reducing soil erosion

6. To identify the best management options to minimize future soil erosion in the

watershed

Page 23: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

5

CHAPTER TWO

STUDY AREA

Location

The Anjeni watershed is located in the Amhara Region which is dominated by

highlands. The watershed is oriented North-South and flanked on three sides by

plateau ridges. It is located at 37o31’E and 10o40’N and lies 370 km NW of Addis

Ababa to the south of the Choke Mountains (Figure 1). Minchet, a perennial river

starts in the watershed and flows towards the Blue Nile Gorge. The lowest point in

elevation in the watershed is found at the outlet (2,406m above sea level). The highest

point in the watershed is found near the Village of Anjeni (2,505 m above sea level).

The research unit Anjeni, which is found in the watershed, was established in March

1984 near the watershed outlet. The research station contains a river station for

hydrological and sediment data collection and a climate station. The catchment area of

Anjeni is 113.4 ha. (Werner, 1986, Bosshart, 1995, SCRP Report, 2000).

Figure 1: Location of Anjeni watershed

Page 24: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

6

Climate

Anjeni is situated in the agro-climatic zone “Wet Woyna Dega” with only one rainy

season which lasts from the middle of May to the middle of October (Hurni 1982).

Maximum daily rainfall is 80mm. The mean annual rainfall is 1,690mm with a low

variability of 10%. In the watershed soil erosion is highly influenced by the erosivity

of rainfall. A single intense rainfall event can cause up to 50% of the monthly soil loss

(SCRP report, 2000). Additionally, there is high gully formation at the bottom of the

watershed. Some of these gullies have been rehabilitated by planting trees in the

gullies to reduce further widening and sliding. The daily minimum air temperature

ranges from 0oC to 20oC and the daily maximum air temperature ranges from 12oC to

33oC. The mean daily temperature ranges from 9oC to 23oC.

Hydrology

The catchment drains from the North-East to South-West (Bosshart, 1995). The upper

part of the watershed is dominated by highly compacted and degraded areas. The grass

lands at the bottom the watershed are also compact and have a very low infiltration

rate. Observations show that the degraded areas in the upper catchment and the bottom

grass land areas are those which produce runoff immediately after rainfall starts. An

interview of 50 farmers indicated that runoff in the watershed at the beginning of the

rainy season is not produced immediately after rainfall. In contrast runoff production

in the middle of the rainy season occurs immediately after the rainfall starts. The

runoff production after a rainfall event at the end of rainy season is faster than the

production of rainfall at the beginning of rainy season but not faster than that of the

mid-rainy season. The mid part of the watershed is dominated by cultivated fields.

These fields have a moderate slope which is further reduced by the terraces

constructed in the watershed since 1986.

Page 25: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

7

According to this study, the surface runoff contributes around 29% to the river

discharge at the outlet of the watershed. The lateral flow contributes around 49% and

groundwater recharge contributes about 22% of the discharge at the outlet of the

watershed. The mean annual discharge of the Minchet River at the catchment outlet is

730mm.

Land use and Soil Conservation

The watershed is mainly used for agricultural purposes. Cultivated fields cover more

than 65% of the watershed. The summary of the land use in the year 2008 is shown in

Figure 2.

Figure 2: Percentage of land use of Anjeni watershed based on the recorded land uses on field investigation (method is provided in chapter three)

Page 26: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

8

Figure 3: Graded bund (a) and graded fanya juu bund (b)

Most of the cultivated fields have small ditches, which serve as drains for the excess

runoff out of the fields. These drainage ditches generally have a depth of 10-20cm and

a width of 20-30cm. The number of ditches and spacing between them in a field

depends on the steepness of the field; the more steep the field, the more runoff

expected hence more ditches have to be made (Werner 1986). This is the traditional

practice that has been practiced in the past and also now. “Fanya Juu Bunds” and

“Graded Bunds” are the measures practiced by the farmers to conserve soil in their

field. “Fanya Juu” is a ditch-wall-combination where the wall is uphill of the ditch

which is on the downhill side. Whereas, the “Graded Bund” is the opposite: the ditch

uphill and the wall downhill, see Figures 3a and 3b. The current conservation practices

in the watershed are presented in Figure 4.

Page 27: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

9

Figure 4: Conservation practices in the watershed

Geology and Soil

The geology of the area is flood basalt resulting from a Tertiary Volcanic Eruption.

Thus Trappean Lava covers the Mesozoic Limestone and Sandstone layers below. The

soil of Anjeni developed on the accumulated basaltic lava to form a plateau with soils

varying over short distances. Based on the study made by (Zeleke, 1998) 8 major soil

units and 10 subgroups were identified. Alisols, Nitisols, and Cambisols are the major

types of soil covering more than 80% of the watershed. The bottom part of the

watershed is covered with deep Alisols. The mid-transitional, gently sloping parts of

the watershed are covered with moderately deep Nitisols. Shallow Regosols and

Leptosols cover the high, steepest part of the watershed. The hill top of the watershed

is covered with moderately deep young Dystric Cambisols. The soil in the watershed

can be classified as acidic and low in organic carbon content. The percentage soil

cover in the watershed, see table 1, is determined base on the study (Zeleke, 2000).

Page 28: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

10

Table 1: Percentage of area of soil cover in the Anjeni watershed based on the map prepared by (Zeleke, 2000)

Soil Subgroup Area (m2) Percentage Area Vertic Luvisols 42254.7 3.9% Haplic Alisols 206027.9 19.1% Dystric Cambisols 188467.5 17.4% Eutric Regosols 99836.3 9.2% Humic Nitosols 66090.3 6.1% Haplic Nitosols 172074.9 15.9% Haplic Lixisols 48049.4 4.5% Lithic Leptosols 24347.6 2.3% Haplic Acrisols 25799.2 2.4% Humic Alisols 208813.6 19.3%

Page 29: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

11

CHAPTER THREE

METHODS

SWAT Model Description

The soil and Water Assessment Tool (SWAT) is a physically-based continuous-event

hydrologic model developed to predict the impact of land management practices on

water, sediment, and agricultural chemical yields in large complex watersheds with

varying soils, land use and management conditions over long periods of time (Arnold

et al., 1998, 2000; Neitsch et al. 2001). It can also be used to simulate water and soil

loss in agriculturally dominated small watersheds (Tripathi et al. 2003).

While the model is not new, it was developed from earlier models: SWRRB

(Simulator for Water Resources in Rural Basins) model (Williams et al. 1985; Arnold

et al., 1990) which is a continuous time step model that was developed to simulate

non-point source loading from watershed, CREAMS (Chemical Runoff, and Erosion

from Agricultural Management System) (Knisel, 1980), GLEAMS (Ground Water

Loading Effects on Agricultural Management Systems) (Leonard et al. 1987), EPIC

(Erosion-Productivity Impact Calculator) (Williams, 1975).

SWAT simulates “subbasins” within a watershed. This helps spatial referencing and is

useful when considering spatiality for watersheds dominated by one land use and soil

type. Input information for each subbasin is organized as: Climate, HRUs (Hydrologic

Response Units), water storage structures, Ground water (a shallow unconfined and

deep confined aquifer), main channel and tributary channels.

Thus, HRUs, ponds, groundwater and channel routing are the components of the

hydrological process (Neitsh et al. 2005).

Page 30: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

12

The water balance is the driving force for the simulation of hydrology. SWAT uses

two steps for the simulation of hydrology, land phase and routing phase. The land

phase is the phase in which the amount of water, sediment, nutrient and pesticides

loadings in the main channel from each subbasin are calculated.

Where is the final water content in millimeters (mm), SWo is the initial soil water

content on day i (mm), Pday is the precipitation on day i (mm), Qsurf is the surface

runoff on day i (mm), AET is the actual evapo-transpiration on day i (mm), Qseep is the

water entering the unsaturated zone from the soil profile on day i (mm), and Qgw is the

return flow from the shallow aquifer and lateral flow on day i (mm).

In this study the modified SWAT model SWAT-WB (Soil and Water Assessment

Tools-Water Balance) (White et al., 2008) is used as the result of the different

mechanism of runoff production in the watershed. Liu et al. (2008) shows that the

runoff in most of the Ethiopian highlands is due to saturation excess flow. Saturation

excess flow is one mechanism for runoff generation in areas with shallow soils

characterized by a highly conductive top soil underlain by a dense top soil, and in

regions where the ground water is close to the surface. Runoff is usually generated

from areas that are saturated or become saturated during a storm. Infiltration

measurements and plot studies in the Ethiopian highlands have shown that the

infiltration rates, especially on hillsides with stone cover, can be of the same order of

magnitude or higher than the greatest rainfall intensity (McHugh, 2006). This high

infiltration rate results in the production of runoff for a more extended period for less

intense storms at the end of rainy season compared to the runoff produced

( )∑=

−−−−+=t

igwseepsurfdayot QQAETQPSWSW

1

Page 31: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

13

immediately after a large storm (Liu et al., 2008). This phenomenon shows the need to

modify SWAT, which is infiltration excess runoff based, to simulate runoff for

watersheds with saturation excess flow being the dominant runoff process. Therefore,

the SWAT-WB (Soil and Water Assessment Tools-Water Balance, White et al., 2008)

follows a saturation excess approach and uses a simplified water balance approach as

models developed with this approach typically outperform others used in Ethiopia (Liu

et al, 2008 and Collick et al, 2008).

Surface Runoff

SWAT 2005 uses the concept that surface runoff occurs whenever the rate of water

application to the ground surface exceeds the rate of infiltration. Based on this

assumption, SWAT uses two methods for estimating surface runoff: the Soil

Conservation Service Curve Number technique (USDA Soil Conservation Service,

1972) and the Green and Ampt infiltration method (Green and Ampt, 1911). In the

Soil Conservation Service (SCS) curve number method often called the Curve-

Number (CN) method, land use and soil characteristics are lumped into a single

parameter (White et al. 2009). The initial value for CN is assigned by the user for each

HRU then SWAT calculates the lower and upper limit. For this calculation, SWAT

uses a soil classification based on the Natural Resource Conservation Services

(NRCS). This classifies soil into four hydrologic groups (a soil group has similar

runoff potential under similar storm and cover condition (NRCS, 1996)) based on

infiltration characteristics of the soil (Neitsch et al. 2005). After this classification the

model defines three antecedent moisture conditions to determine the appropriate CN

for each day using the CN-AMC (Curve Number – Antecedent Soil Moisture

Condition) (USDA – NRCS, 2004) distribution based on the moisture content of the

soil calculated by the model (Neitsch et al., 2005). This daily CN is then used to

determine a theoretical capacity S (retention parameter) that can be infiltrated.

Page 32: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

14

The empirical model used to estimate direct runoff from storm is the SCS runoff

equation.

Where Qsurf is the daily surface runoff in millimeters (mm), Pday is the daily

precipitation (mm), Ia is the initial abstraction which is commonly approximated as

0.2S, and S is the retention parameter.

Thus,

SWAT calculates runoff if and only when the amount of precipitation is greater than

the initial abstractions and the rate of precipitation exceeds the rate of infiltration.

Thus, SWAT indirectly assumes only infiltration excess runoff is created. Whereas,

the daily runoff from a given event in SWAT-WB is equal to the amount of rainfall

minus the amount of water that can be stored in the soil before it is saturated. This

amount of storage is called available soil storage (White et al., 2008).

( )soilsoilD θφτ −=

Where τ is the available soil storage, D is the effective depth of soil profile, Φsoil is the

total soil porosity as expressed as a function of the total soil volume, and θsoil is the

volumetric soil moisture of HRU.

⎟⎠⎞

⎜⎝⎛ −= 1010004.25

CNS

( )( )SIP

IPQ

aday

adaysurf +−

−=

2

( )85.0

25.0 2

+−

−=

aday

daysurf IP

PQ

Page 33: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

15

The soil porosity is calculated by SWAT using the relationship between soil porosity

and soil bulk density.

Where ρb is the soil bulk density mg/m3 and ρs is the particle density mg/m3 (based on

researches a default value of 2.65 mg/m3 is used (Neitsch et al, 2005)). The volumetric

soil moisture θsoil varies on each day depending upon plant uptake of water,

evaporation and precipitation (White et al, 2008).

Runoff is produced only when the precipitation is greater than the available soil

storage. Otherwise no surface runoff is produced. In this method, rainfall intensity is

assumed to have a limited impact on runoff production but rainfall volumes are the

ultimate drivers of soil saturation and the total rainfall volume determines the amount

of runoff (White et al., 2008).

Thus, runoff is calculated as

τ−= PQsurf

Where Qsurf is the surface runoff in millimeters (mm), P is the daily precipitation in

mm, and τ is the available soil storage in mm.

Soil Water

Water that infiltrates into the soil profile has several routes to leave the soil. Soil water

may be lost as evapotranspiration, taken by plants, percolated to the bottom of the soil

profile; it may flow laterally in the soil and then contribute to runoff in the main

channel. Water percolated from the root zone ultimately becomes aquifer recharge.

s

bsoil ρ

ρφ −= 1

Page 34: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

16

The water content of a soil can range between wilting point to the soil porosity when

the soil is saturated. Between these two states there are two points important for plant-

soil interaction; field capacity and wilting point. Field capacity (FC) is the amount of

water held in the soil after excess gravitational water has drained away and after the

rate of downward movement has materially decreased. The soil water suction values

generally used for field capacity approximation range from 5 kPa for coarse-texture

soils to 10 kPa for samples that retain their original structure (McIntyre, 1974;

Marshall, 1982). Permanent wilting point (WP) is defined as the water content at

which the leaves of a growing plant reach a stage of wilting from which they do not

recover. Different plants have different values of soil water suction at wilting point.

Since the change in water content is small between 800 kPa and 3000 kPa for most

soils, a suction of 1500 kPa based on wilting studies with dwarf sunflower is generally

taken to be an approximation of permanent wilting point (Reeve and Carter, 1991).

Available water capacity (AWC) defined as the amount of water that can theoretically

be extracted by plants from a soil initially at field capacity (McIntyre, 1974).

Figure 5: Soil Water Characteristics Curve (http://www.dpi.vic.gov.au/dpi/vro/gbbregn.nsf/pages/soil_hydraulic_pdf/$FILE/TechReportch02.pdf

Page 35: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

17

The amount of water held in the soil between the field capacity and the permanent

wilting point is considered to be the available water for plant extraction, see Figure 5

above. This water available for plants is referred to as the Available Water Capacity

(AWC). It is calculated as:

WPFCAWC −=

Where FC is the field capacity and WP is the wilting point.

SWAT estimates the permanent wilting point for each soil layer as

Where mc is the percent clay content of the layer in percentage (%) and ρb is the bulk

density for the soil layer.

SWAT calculates field capacity (FC) by adding AWC which is an input by the user

and the wilting point (WP). Saturated flow occurs when the water content of a soil

layer surpasses the field capacity for the layer. The excess water from field capacity

water content is available for percolation, lateral flow and surface runoff.

Lateral Flow

This flow is significant in watersheds with soils having high hydraulic conductivities

in surface layers and an impermeable or semi-permeable layer at a shallow depth. The

water collects above the impermeable layer is the source of water for lateral

subsurface flow (Neitsch et al., 2005).

⎟⎠⎞

⎜⎝⎛=

10040.0 bc xm

xWPρ

( )hilld

satexcesslylat Lx

SlpxKxSWxxQ

φ,2

024.0=

Page 36: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

18

Where Qlat is the lateral flow, SWly,excess is the drainable volume of water stored in the

saturated layer in millimeters (mm) (A soil is considered to be saturated whenever the

water content of the layer exceeds the layer’s field capacity water content), Ksat is the

saturated hydraulic conductivity (mm/hr), Slp is the increase in elevation per unit

distance equivalent to tanαhill where tanαhill is the slope of the hill slope segment. Slp is

a value which is an input for SWAT (m/m), Φd is the drainable porosity of the soil

layer (mm/mm), and Lhill is the hill slope length (m).

Where SWly is the layer water content on a given day in millimeters (mm) and FCly is

the water content of soil layer at field capacity in mm.

The drainable porosity of the soil layer Φd is calculated as follows.

fcsoild φφφ −=

Where Φsoil is the total porosity of soil layer in mm/mm and Φfc is the porosity of soil

layer filled with water when the layer is at field capacity in mm/mm.

The amount of lateral flow calculated using the above procedures may not reach the

stream on the same day. The lag on lateral flow results in only a fraction of the lateral

flow from each HRU reaching the stream. Thus the daily amount of lateral flow reach

the steam is calculated as (Neitsch et al., 2005).

lylyexcessly

lylylylyexcessly

FCSWifSW

FCSWifFCSWSW

≤=

>−=

0,

,

( )⎟⎟

⎜⎜

⎛−+=

⎟⎠⎞

⎜⎝⎛ −

=lagTT

ilatstorlatlat xQQQ1

1, exp1

Page 37: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

19

Where is the amount of later flow discharged to the main stream in a day in mm,

Qlat is the amount of lateral flow generated in the subbasin on a given day in mm, and

TTlag is the lateral flow travel time (days).

The lateral flow travel time TTlag can be calculated by SWAT or the user can define it.

, if HRUs are without drainage tiles.

Where Lhill is the hill slope length in meters and Ksat,mx is the highest layer saturated

hydraulic conductivity in the soil profile (mm/hr) (one of the soil property taken from

previous studies).

, if drainage tiles are present in the HRUs.

Where tilelag is the drain tile lag time in HRUs (hours). Drainage tiles are subsurface

structures for draining water from the soil surface. They usually installed at 90mm

below the soil surface.

Percolation and Ground Water Return Flow

SWAT calculates percolation for each layer in the profile and this process occurs only

when the water content of the soil is more than field capacity.

Where Qper,ly is the amount of water percolating to the underlying soil in mm, SWly,excess

is the drainable volume of water in the soil layer, Δt is the length of the time step in

hours, and TTperc is the travel time for percolation in hours.

mxsat

hilllag K

LTT,

4.10=

24tilelagTTlag =

⎟⎟⎟

⎜⎜⎜

⎛−= ⎥

⎥⎦

⎢⎢⎣

⎡ Δ−

percTTt

excesslylyper xSWQ exp1,,

Page 38: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

20

sat

lylyperc K

FCSATTT

−=

Where TTperc is the travel time for percolation in hours, SATly is the amount of water in

the soil layer when completely saturated in mm, FCly is the water content of the layer at

the field capacity in mm, and Ksat is the saturated hydraulic conductivity for the layer in

mm/hr.

Recharge to an unconfined aquifer occurs via percolation to the water table from a

significant portion of the land surface. The fraction of the total daily recharge routed to

the deep aquifer is given by.

rchrgdeepdeep QxQ β=

Where Qdeep is the amount of water moving into the deep aquifer on day ‘I’ in mm,

βdeep is the aquifer percolation coefficient (this parameter is defined by the user as

RCHRG_DP), and Qrchrg is the amount of water entering both aquifers on day ‘i'.

Recharge to shallow aquifer can be calculated as follows.

deeprchrgshrchrg QQQ −=,

Where Qrchrg,sh is the amount of water entering the shallow aquifer on day ‘i’.

Depending on the water table height in the shallow aquifer there is a base flow

contribution to the main channel. This flow occurs only when the water stored in the

shallow aquifer is greater than the threshold water level in the shallow aquifer for

ground water contribution to the main channel to occur. This value is defined by the

Page 39: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

21

user (aqshthr,q) or in the SWAT interface the variable is presented as GWQMN (Neitsch

et al., 2005).

Thus

Where Qgw,i is the ground water flow into the main channel on day ‘i’ in mm, Qgw,--i-1

is the ground water flow into the main channel on day ‘i – 1’ in mm, and αgw is the

base flow recession constant which is a direct index of ground water flow response to

change in recharge (Smedema and Rycroft, 1983), Δt is the time step (1 day), Qrchrg,sh

is the amount of recharge to the shallow aquifer on day ‘i’ in mm, aqsh is the amount

of water stored in shallow aquifer at the beginning of day ‘i’ in mm, and aqshthr,q is the

threshold water level in shallow aquifer in mm.

When the shallow aquifer receives no recharge the above equation is simplified to;

Where Qgw,i is the ground water flow into the main channel at time t in mm, Qgw,0 is

the ground water flow into the main channel at the beginning of the recession (time =

0) in mm, αgw is the base flow recession constant defined by the user (ALPHA_BF),

and t is the time elapsed since the beginning of the recession (days).

[ ] [ ]( )qsthrshigw

qshthrshgwshrchrggwigwigw

aqaqifQ

aqaqiftxQtxxQQ

,,

,,1,,

0

exp1exp

≤=

>Δ−−+Δ−= − αα

[ ]qsthrshigw

qshthrshgwgwigw

aqaqifQ

aqaqiftxQQ

,,

,0,,

0

exp

≤=

>−= α

Page 40: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

22

Sediment

SWAT uses the Modified Universal Soil Loss Equation (MUSLE) to estimate the soil

loss from each HRU.

( ) 56.056.0,8.11 CFRGxLSxPxCxKxhruareaxqxQxSed USLEUSLEUSLEUSLEpeaksurf=

Where Sed is the sediment yield on a given day in metric tons, Qsurf is the surface

runoff from the watershed in mm/ha, qpeak is the peak runoff rate in cubic meter per

second, area,hru is the area of HRU, KUSLE is the USLE soil erodability factor, CUSLE

is the USLE land cover and management factor, PUSLE is the USLE support practice

factor, LSUSLE is the USLE topographic factor, and CFRG is the coarse fragment

factor.

The peak runoff rate is the maximum runoff rate that occurs with a given rainfall

event. SWAT calculates the peak runoff rate with a modified rational method (Neitsch

et al., 2005). A brief description of sediment routing components of SWAT is given

below (Neitsch et al., 2005).

Where qpeak is the peak runoff rate in cubic meters per second, C is the runoff

coefficient is ratio of inflow rate to peak discharge rate, i is the rainfall intensity in

mm/hr, and Area is the subbasin area in square kilometers.

Where Qsurf is the surface runoff in mm and Pday is the rainfall for the day in mm.

6.3AreaxixCq peak =

day

surf

PQ

C =

Page 41: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

23

Rainfall intensity is the average rainfall rate during the time of concentration (Neitsch

et al., 2005).

Where i is the rainfall intensity in mm/hr, Ptc is the amount of precipitation during the

time of concentration in mm, and tconc is the time of concentration for a subbasin in

hours.

daytctc PxaP =

Where atc is the fraction of daily precipitation that occurs during the time of

concentration and Pday is the daily precipitation in mm. SWAT estimates the fraction

of daily precipitation during the time of concentration (atc) as a function of the fraction

of daily rain falling in the half-hour of highest intensity rainfall.

( )[ ]5.01ln2exp1 α−−= xtxa conctc

Where α0.5 is the fraction of daily rain falling in the half-hour highest intensity rainfall

and tconc is the time of concentration in hours is the amount of time from the beginning

of a rainfall event until the entire subbasin area is contributing to flow at the outlet

(Neitsch et al., 2005).

chovconc txtt =

Where tov is the time of concentration for overland flow in hours and tch is the time of

concentration for channel flow in hours.

3.0

6.06.0

18 SlpxnxL

t Slpov =

conc

tc

tPt =

Page 42: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

24

Where Lslp is the slope length of subbasin in meters, n is the Manning’s coefficient,

and Slp is the average slope in the subbasin in m/m.

Where L is the channel length in meters, n is the Manning’s coefficient for the

channel, Area is the subbasin area in square kilometers, and Slpch is the channel slope

in m/m.

The factors KUSLE, CUSLE, PUSLE, LSUSLE, and CFRG are taken and used based on

previous studies on the watershed and the definition and calculations of the parameters

presented in the SWAT documentation (Neitsch et al., 2005).

After estimating the amount of soil contributed by each subbasin to the main channel

the next step is routing soil loss which has two components; deposition and

degradation. To determine the deposition and degradation maximum concentration of

sediment (Concsed,ch,mx) is compared to the concentration of sediment in the reach at

the beginning of the time step (Neitsch et al., 2005).

( ) exp,,,

sppkchspmxchsed VxCConc =

Where Concsed,ch,mx is the maximum concentration of sediment that can be transported

by the water in kg/lit, Csp is the coefficient defined by the user (SPCON), and Vch,pk is

the peak channel velocity in m/s, and spexp is the exponent defined by user (SPEXP).

The peak channel velocity Vch,pk is calculated as:

375.0125.0

75.062.0

chch SlpxArea

nxLxt =

Page 43: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

25

Where qch,pk is the peak flow rate in cubic meters per second (m3/s) and Ach is the

cross-sectional area of flow in the channel in square meter.

The peak flow rate qch,pk is defined as

chpkch qxprfq =,

Where prf is the peak rate adjustment factor (PRF) and qch is the average rate of flow

in m3/s.

If Concsed,ch,i is greater than Concsed,ch,mx, then deposition is the dominant process.

Thus, deposition is calculated as

( ) chmxchsedichseddep VxConcConcSed ,,,, −=

Where Seddep is the amount of sediment deposited in the reach segment in metric tons,

Concsed,ch,i is the initial sediment concentration in the reach in kg/lit or ton/m3,

Concsed,ch,mx is the maximum concentration of sediment that can be transported by the

water in kg/lit or ton/m3, and Vch is the volume of water in the reach segment in m3. If

Concsed,ch,i < Concsed,ch,mx then degradation is the dominant process in the reach

segment. The amount of sediment re-entrained is calculated as

( ) chchchmxchsedichseddep CxKxVxConcConcSed ,,,, −=

Where Seddeg is the amount of sediment re-entrained the reach segment (metric tons)

is the volume of water in the reach segment (m3), Kch is the channel erodability

ch

pkchpkch A

qV ,

, =

Page 44: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

26

factor (conceptually similar to the soil erodability factor used in MUSLE), and Cch is

the channel cover factor (defined as the ration of degradation of a channel with a

specified vegetation cover to the corresponding degradation from channel with no

vegetation cover.

Once the deposition and degradation in the reach is determined the next step is to

calculate the suspended sediment in the reach.

deg, SedSedSedSed depichch −−=

Where Sedch is the amount of suspended sediment in the reach (metric tons) and Sedch,i

is the amount of suspended sediment in the reach at the beginning of the time period

(metric tons).

The amount of sediment transported out of the reach Sedout is calculated as

Where Sedout is the amount of sediment transported out of the reach in m3, Sedch is the

amount of suspended sediment in the reach in m3, Vout is the volume of water leaving

the reach segment during the time step in m3, and Vch is the volume of water in the

reach segment in m3.

Model Input

SWAT is a comprehensive model that requires information provided by the user to

simulate runoff and soil erosion. The first step in initializing a watershed simulation is

ch

outchout V

VxSedSed =

Page 45: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

27

to partition the watershed into subbasins. The user has the option of allowing SWAT

to automatically delineate the watershed and subbasins using the Digital Elevation

Model (DEM) or the user can provide predefined subbasins. The land area in a

subbasin is divided into hydrologic response units (HRUs). Hydrologic response units

(HRUs) are portions of a subbasin and possess unique land use, slope range, and soil

attributes (Neitsch et al., 2004).

SWAT has different components. Hydrologic components of the model work on the

water balance equation, which is based on surface runoff, precipitation, percolation,

evapotranspiration, and return flow data; Weather is one of the model component that

needs data on precipitation, air temperature, solar radiation, wind speed, and relative

humidity data; Sedimentation is another component of the model that needs

information on surface runoff, peak rate flow, soil erodability, crop management,

erosion practices, slope length, and steepness; Soil temperature, crop growth, nutrient

pesticides and agricultural management are also components of SWAT. Thus, the data

required for the model are DEM, soil data, land use data, precipitation and other

weather data. For calibrating the model and also for validation purposes, river

discharge and sediment yield are required at the outlet of the watershed.

Digital Elevation Model

To delineate the watershed and subbasins and to determine drainage networks SWAT

uses the digital representation of the topographic surface. DEM is the digital

representation of the topographic surface. A 2m by 2m resolution DEM was used from

the Center for Development and Environment (CDE), Institute of Geography,

University of Berne, Switzerland, see Figure 6 below. Subbasin parameters such as

slope gradient, slope length of terrain and the stream network characteristics such as

channel length, width and slope were calculated and used by the model.

Page 46: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

28

Figure 6: The digital elevation model of Anjeni watershed

Land Use Map

A map of land use for 2008 was created by recording the crop type on each plot in the

watershed and by identifying the land cover on areas other than cultivated fields. A

Google image of the watershed in 2008 (Figure 7) exactly fits and represents all fields

in the watershed. Hence, the image was used for recording. The digital Google image

was geo-referenced by taking 15 control points around and inside the watershed. The

shape file representing each plot and other land covers was created using the digitizing

tools provided in ArcGIS, ArcMap. A total of 16 different land uses were identified

and mapped as shown in Figure 8.

Page 47: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

29

Figure 7: Google earth image of Anjeni watershed with 20 different control points

Figure 8: Land use map of Anjeni using Google image and the recorded land use on each plot

Page 48: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

30

Soil Map and Data

Soils in the watershed should be categorized and prepared as a map in a shape file

format and then linked to a customized soil database designed by the user if the soils

are not included in the existing SWAT soil database. The soil map used in this

research was taken from the study made by Zeleke (2000). The soils were classified

based on the FAO (Food and Agriculture Organization) method of soil classification

(SCRP, 2000). Based on this there are ten different soil types in the watershed.

Basic physical properties (percentage sand, clay, and silt; soil texture class; soil

texture class; the percentage of carbon and profile thickness), derived soil properties

(hydraulic conductivity, bulk density, available water capacity, and soil organic matter

content) and the basic properties of each profiles of the ten different soils in the

watershed were obtained from Zeleke (2000), Kejela (1987), and Setegn et al. (2008)

(Figure 9).

Page 49: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

31

Figure 9: Soil map of Anjeni watershed (based on FAO classification)

Weather Data

SWAT requires daily or sub-daily meteorological data. The model can either read

these meteorological data from previously measured data stored in tables or can be

generate it using a weather generator model. In this study, measured meteorological

data were used and the weather generator model was set up to estimate any missing

data. The meteorological data used were daily precipitation, daily maximum and

minimum air temperature, daily solar radiation, wind speed, and relative humidity on a

daily basis.

Daily precipitation data from 1984 – 2005 were obtained from the Soil Conservation

Research Program (SCRP), Ethiopia. There were missing precipitation data in some

months in the years 1997 and 1999. In these years the model uses values generated by

Page 50: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

32

the weather generator model. SWAT uses WXGEN weather generator model

(Sharpley and Williams, 1990). The model generates precipitation using Markov

chain-skewed (Nicks, 1974) or Markov chain-exponential model (Williams, 1995). A

first-order Markov chain is used to define the day as wet or dry. When a wet day is

generated a skewed distribution or exponential distribution is used to generate the

precipitation amount (Neitsch et al, 2005). Maximum half-hour rainfall is required by

SWAT to calculate the peak flow rate for runoff. This value can be calculated from a

triangular distribution using maximum half-hour rainfall data or using monthly

maximum half-hour rainfall for all days in the month (Neitsch et al, 2005). The

procedure used to generate daily values for maximum temperature, minimum

temperature and solar radiation (Richardson, 1981; Richardson and Wright, 1984) is

based on the weakly stationary generating process presented by Matalas (1967)

(Neitsch et al, 2005). The daily maximum and minimum temperature from the year

1984 to 2005 were obtained from SCRP, Ethiopia. For the month of June and July in

the year 1996, for the month December 1997 and for four days in 1999 data were not

available. So, the weather generator was used to provide values for these dates. Daily

average monthly relative humidity values are calculated form a triangular distribution

using average monthly relative humidity. This method was developed by Williams for

the EPIC model (Sharpley and Williams, 1990; Neitsch et al, 2005). Mean daily wind

speed is generated in SWAT using a modified exponential equation.

( )( )3110 ln rndxmonwndm −= μμ

Where μ10m is the mean wind speed for the day (m/s), μwndmon is the average wind

speed for the month (m/s), and rnd1 is a random member between 0.0 and 1.0

Page 51: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

33

SWAT used solar radiation, wind speed and relative humidity to calculate the potential

evapotranspiration (PET) as the model for this particular project used Penman-

Monteith approach for PET calculation. The meteorological station found in Anjeni

watershed has no data on solar radiation, wind speed and relative humidity. Hence,

Daily sunshine hours from the 1997 to 2006, wind speed, and relative humidity from

1994 to 2006 were taken from the nearby station Debremarkos meteorological station.

This station is located 46kms away from the watershed. Daily solar radiation was

calculated from the daily sunshine hour data using the Angstorm-Prescott equation

which is a simple empirical formulae that relates short-wave radiation with other

physical factors, such as extraterrestrial radiation, optical air mass, and turbidity, water

vapor content of the air, amount and type of cloud cover (Njau, 1996; Revfeim, 1997;

Persuad et al., 1997),

Where Qs is the solar radiation in W/m2, Qext is the daily total extraterrestrial radiation

in W/m2, a and b are constants which depend on the location, season, and state of the

atmosphere, n is the actual number of hours of bright sunshine (sunshine hour), and N

is the number of day light hours (since Ethiopia is near the Equator, N is assumed to

be 12).

Wind speed data and the relative humidity were available only from 1994 to 2006 on a

daily basis. So, the weather generator was used to simulate values for the rest years for

the PET calculation to be performed.

⎥⎦

⎤⎢⎣

⎡⎟⎠⎞

⎜⎝⎛+=

NnbaQQ exts

Page 52: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

34

River Discharge and Sediment Yield

The river discharge of the outlet Anjeni watershed was obtained from SCRP, Ethiopia.

Data from 1986 to 1993 were available as well as the years 1995, 1996, and 2000.

SWAT does not use these data values in calculations but instead they are used for

comparing observed and simulated values in calibration and validation periods. Data

from 1986 to 1993 were used for calibration and validation and the data for the years

1995, 1996, and 2000 were used for validation.

There was uncertainty surrounding the measured discharge and the sediment yield in

the data from SCRP. For this reason, the river discharge was calculated using the true

water depth measured for those years. The discharge rate at the watershed outlet, in

liters per second, was calculated from the water level height in the main channel which

is measured in each 10minutes during the rainy season. To do this the stage-discharge

relation rating equation for the Minchet River at the Anjeni watershed outlet was used

(Bosshart, 1997):

Where q(H<60) is the discharge of the River Minchet (in liters/second) at the watershed

outlet before the water height gets 60cm, q(60≤H<120) is the discharge of the River

Minchet in liters/second at the watershed outlet when the water level height is between

60cm and 120cm, q(120≤H<360) is the discharge of the River Minchet in liters/second at

the watershed outlet when the water level height is between 120cm and 360cm, is

the water level height or stage in centimeters (cm).

( )

( )

( ) 0.36200.700.14000.567.0

2.1

360120

212060

5.160

−=

+−=

=

<≤

<≤

<

HqHHq

Hq

H

H

H

Page 53: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

35

Then the amount of discharge leaving the watershed over a certain time is calculated

as

Where Qi is the amount of water in m3 leaving the watershed in the time interval (tf-ti),

qi is the discharge rate measured using the rating equation when the stage height is Hi,

ti is the initial time in seconds for the river water level height or stage comes to Hi, and

tf is the final time in seconds for the river water level to change from Hi. The amount

of water leaves the watershed in a day is calculated as

∑=

=n

iiday QQ

1

Where Qday is the amount of water in m3 leaving the watershed in a day (24hours) and

Qi is the amount of water in m3 leaving the watershed in the time interval (tf-ti).

The daily discharge is the sum of all the discharge calculated for a particular time in a

particular day. The discharge in this study is considered in mm depth of water in order

to compare the result of the SWAT simulation to the observed values. Thus, the

observed water yield from the watershed is calculated as

Where Qobs is the daily observed water yield from the watershed in mm depth

(observed discharge in mm), Qday is the amount of water in metric cube leaving the

watershed in a day, and A is the area of the watershed in hectares.

The concentration of sediment in grams/liter in the river was obtained from the SCRP

office, Addis Ababa, Ethiopia. The sediment concentration in the river discharge was

( )1000

ifii

ttxqQ

−=

( )AxQ

Q dayobs 10

=

Page 54: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

36

measured by SCRP whenever the water level is measured. Thus, the soil loss from the

watershed is calculated as

Where SEDi is the sediment loss in kg from the watershed while the water level height

is Hi, Qi is the amount of water in m3 leaving the watershed while the water level

height is Hi, and Sedi is the measured sediment concentration in gram/liter while the

water level height is Hi.

The daily soil loss from the watershed is the sum of all the calculated soil losses for all

the different water level heights in a particular day.

∑=

=n

iiday SEDSED

1

Where SEDday is the soil loss from the watershed in a day in kg, SEDi is the sediment

loss in kg from the watershed while the water level height is Hi.

Where SEDobs is the observed sediment yield from the watershed in tons/hectare,

SEDday is the soil loss from the watershed in a day in kg, and is the watershed area in

hectares.

Management Practices and Scenarios

SWAT gives options for the user to consider different management practices and the

crop calendar in a watershed. A crop calendar was prepared for this watershed after

interviewing 50 farmers in the watershed (Table 2).

iii SedxQSED =

AxSED

SED dayobs 1000

=

Page 55: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

37

Table 2: Crop calendar used in the Anjeni watershed Growing Season Harvesting Season Plowing Crop Starts Ends Starts Ends Start End

Barley 23-May 5-Sep 5-Sep 9-Dec 9-Apr 23-May Teff 27-Jul 19-Dec 19-Dec 9-Mar 7-Feb 27-Jul Wheat 6-Aug 30-Dec 30-Dec 7-Feb 7-Feb 6-Aug Corn 23-May 9-Dec 9-Dec 9-Dec 18-Feb 9-May Soy Bean 6-Aug 9-Dec 9-Dec 8-Jan 8-Jun 6-Aug Nug/FLAX 7-Jun 30-Dec 30-Dec 10-Jan 9-May 7-Jun Sinar/ALFA 27-Jul 29-Dec 29-Dec 7-Feb 8-Jun 27-Jul

Tillage Activity

The farmers make drainage ditches across furrow in their field. After the SCRP start

its activities in the watershed the farmers started using parallel terraces (Fanya juu)

and contour plowing in the year 1986. These tillage practices were considered as

factors and parameters in calibrating the model. The existence of terraces in the

watershed is included in the model setup by considering the resulting slope and slope

length change on the fields due to the construction of terraces. Contour plowing is a

management option to alleviate soil erosion which is taken into consideration during

the model calibration by changing the Universal Soil Loss Equation support practice

factor.

Model Setup

Watershed delineation

SWAT allows the user to delineate the watershed and subbasins using the Digital

Elevation Model (DEM). This tool uses and expands ArcGIS and Spatial Analyst

extension functions to perform watershed delineation (Neitsch et al., 2002). The DEM

of the area is loaded into an ESRI (Environmental System Research Institute) grid

format. Stream network was defined for the whole DEM by SWAT using the concept

of flow direction and flow accumulation. Before defining the stream network, the

Page 56: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

38

model processes the DEM map grid to remove all the non-draining zones (sinks). To

define the origin of streams a threshold area was defined. The threshold area defines

the minimum drainage area required to from the origin of a stream. The size and

number of subbasins and details of stream network depends on this threshold area

(Winchell et al., 2007). The threshold area was taken to be 3.4ha, suggested by

ArcSWAT. The threshold area, or critical source area, defines the minimum drainage

area required to form the origin of a stream. The watershed outlet is manually added

and selected for finalizing the watershed delineation. With this information the model

automatically delineated a watershed of 106.5 ha and 13 subbasins were produced.

HRU Definition

The Hydrologic Response Units (HRUs) Analysis tool in ArcSWAT helps to load land

use and soil layers to the project. The delineated watershed by ArcSWAT and the

prepared land use overlapped 100%. The 16 classes land use map was reclassified into

11 classes in order to correspond with the land use in the SWAT interface except teff

which was created in SWAT based on the study Setegn et al. (2008) (Figure 10).

Figure 10: Land use as reclassified by SWAT into the four letter land use code

a) b)

Page 57: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

39

The delineated watershed and soil map have an overlap of 99.99%. The soil classes in

the input soil map were decoded using a lookup table (a table created by the user in

.txt format according to SWAT input file format) so that the parameters corresponding

to each soil type could be accessed from the ArcSWAT database (Figure 11).

Figure 11: Soil Map of Anjeni Watershed as reclassified into four letters soil name in SWAT database

HRU analysis in ArcSWAT includes divisions of HRUs by slope classes in addition to

land use and soils. The multiple slope option (an option for considering different slope

classes for HRU definition) was selected for this study. Slope discritization was done

based on the study of Setegn et al., 2008 (Table 3). The slope discritization (0-1, 1-3,

3-5, >5) which accounts for the lower slope ranges is the best discritization option in

considering deposition of soil materials during sediment transportation (Setegn et al.,

2008).

Page 58: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

40

Table 3: Slope descritization used for creation of HRUs Classes Slope Range

1 2 3 4

0% - 1% 1% - 3% 3% - 5%

> 5%

Multiple HRUs were defined within a subbasin by ignoring land uses less than 2% of

the subbasin and also ignoring soil types in a subbasin covering less than 5% of the

subbasin. A total of 465 HRUs for 13 subbasins were created.

Weather Data Definition

The WXGEN weather generator model included in SWAT was used to fill in gaps in

measured records. This weather generator was developed for U.S. Since Anjeni

watershed is located outside U.S. the WXGEN weather generator was provided with

all the necessary statistical information from the meteorological records of the

watershed. The parameters needed for the weather generator are listed in Appendix V

(for definition of each parameters listed look at Neitsch et al (2005). These statistical

values were calculated from the meteorological data available in the Anjeni watershed

and Debremarkos station. The number of years for calculating the statistical values

depends on the availability of data in the stations. Other meteorological data (daily

precipitation, daily minimum and maximum air temperature, daily relative humidity,

daily solar radiation and daily wind speed) including the corresponding location table

were prepared according to the SWAT format and integrated into the model using the

weather data input wizard.

Management Practices

ArcSWAT provides two options for defining the management operations in the

watershed; scheduling by date and scheduling by heat units. In this study, scheduling

the management practices by date was preferred because of the lack of information on

Page 59: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

41

heat units of crops in the watershed. The basic operations are tillage, growing and

harvest and kill operation. Planting operation or the beginning of growing season used

to designate the time of planting for agricultural crops and initiation of plant growth

for a land cover that requires several years to reach maturity. The tillage operation

redistributes residue, nutrients, pesticides and bacteria. Harvest and kill operation

stops plant growth in a way that the fraction of biomass is removed from the HRU as a

residue on the soil surface. These operations were provided to the model for each crop

type based on the crop calendar prepared beforehand. The tillage practice was done

using the traditional plowing system (conventional agriculture) with a depth of plough

varying from 10cm to 15cm.

Model Sensitivity Analysis, Calibration and Validation

After the model was set up the next step was to run the model. The results from the

simulation cannot be directly used for further analysis but instead the ability of the

model to sufficiently predict the constituent sediment yield and stream flow should be

evaluated through sensitivity analysis, model calibration and model validation (White

and Chaubey, 2005).

The aim of the sensitivity analysis is to estimate the rate of change in the output of a

model with respect to changes in watersheds that result in a clear difference in

hydrologic sensitivity (Reungsang et al., 2005). Sensitivity analyses were conducted

for the Anjeni watershed hydrology to determine the parameters needed to improve

simulation results and thus to better understand the behavior of the hydrologic system

and to evaluate the applicability of the model.

Parameters for sensitivity analysis were selected by reviewing previously used

calibration parameters and documentation from the SWAT manuals (e.g., Neitsch et

Page 60: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

42

al., 2005; Werner, 1986; Zeleke, 2000; Bosshart, 1997; Setegn et al., 2008; White and

Chaubey 2005; Kirsch et al., 2002; Arnold et al., 1999 and 2000) as illustrated in

Table 4. The sensitivity parameters selected for this study are shown in Table 5.

Table 4: Review of Calibration of Parameters by variable used by SWAT modelers Output Variables Calibration Parameters

Flow CANMX4 /Crop Growth Routine5 /Curve Number1,2,3,4,5,6,7 /ESCO3,5,6 /Revap Coefficients2,3,4 /Soil Bulk Density5 /Soil Properties1 /AWC6,7 /EPCO3 /Ground Water Parameters5 /Soil Hydraulic Conductivity5

Sediment AMP4 /Channel Cover5 /Channel Erosion5 /CH_N24

/MUSLE Parameters5 /PRF4 /SLSUBBSN4 /SPCON3,4 /SPEXP3,4 /USLE_K(1)4 /SLOPE4 /CH_N14

1Arnold and Allen, 1996; 2Srinivasan et al., 1998; 3Santhi et al., 2001b; 4Cotter, 2002; 5Kirsch et al., 2002; 6Arnold et al., 2000; and 7Arnold et al., 1999

Table 5: Parameters used for sensitivity analysis in this study

Flow Sediment GW_REVAP ALPHA_BF SLSUBBSN USLE_C

GWQMN REVAPMN SPCON USLE_K EPCO ESCO SPEXP USLE_P

SOL_K CANMX Ch_N Blai SOL_AWC GW_DELAY Ch_EROD GWQMN

SOL_Z Blai Ch_K2 SLOPE

Para

met

ers

SURLAG Biomix Ch_COV SURLAG

Model calibration is the modification of parameter values and comparison of predicted

output of interest to measured data until a defined objective function is achieved

(James and Burges, 1982). Parameters for modification are selected from those

identified by the sensitivity analysis. Additional parameters, other than those identified

during sensitivity analysis, are used primarily for calibration due to the hydrological

processes naturally occurring in the watershed. Sometimes it is necessary to change

Page 61: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

43

parameters in the calibration process other than those identified during sensitivity

analysis because of the type of miss match of the observed variables and the predicted

variables (White and Chaubey, 2005), as illustrated in Table 6.

Table 6: Parameters used for Calibration

Output Variables Parameters Selected for Calibration

Flow REVAPMN GW_REVAP GW_DELAY ESCO SOL_K SOL_AWC ALPHA_BF RCHRG_DP SURLAG EPCO SOL_BD EDC Sediment SLSUBBSN SLOPE USLE_C USLE_P SPEXP ADJ_PKR Ch_EROD Ch_COV USLE_K PRF SPCON ISED_DET

In this study the calibration was divided into two steps:

i) water balance and stream flow, and

ii) sediment

The calibration for water balance and stream flow was first done for average annual

conditions. Next, calibration was done for the surface flow, groundwater flow, and

lateral flow for average annual conditions. Thus, before starting the calibration the

measured total water yield from the watershed should be sub-divided into base flow

and surface flow. The surface runoff was separated from the total flow using the

following equations.

If ( ) 11 925.09625.0 −− ≥− iSURii QxQQx

Then,

Otherwise,

Once the surface runoff is known the base flow can be calculated as (Hewlett and

Hibbert, 1967; Arnold et al, 1995; Arnold and Allen, 1999).

iSURiiBASE QQQ −=

( ) 11 925.09625.0 −− +−= iSURiiiSUR QxQQxQ

0=iSURQ

Page 62: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

44

Where QSUR i is the surface runoff on day i in mm, QSUR i-1 is the surface runoff on day

(i-1) the day before day i in mm, Qi is the total water yield from the watershed on day i

in mm, Qi-1 is the total water yield from the watershed on day (i-1) in mm, and QBASE i

is the base flow contribution to the stream on day i in mm.

For increased accuracy, ground water and lateral flow estimations were considered.

The values GWQ (Groundwater discharge) and SURQ (Surface runoff) in the SWAT

output files cannot be used directly because in stream precipitation, evaporation,

transmission losses etc. will alter the net water yield from that predicted by the WYLD

(water yield) variable. Nevertheless, the Anjeni watershed is a micro watershed and

the effect of precipitation and evaporation from the river is assumed not to have a big

influence. Thus, groundwater and lateral flow were calibrated using GWQ and SURQ

values in the SWAT output files for comparison. While this particular process was not

very accurate, but it is very helpful in a way that the simulated groundwater flow,

lateral flow, and surface water to have similar pattern over time and also the same in

amount to that of the observed one.

The base flow in the watershed was analyzed as described below. The physical

observation of the watershed shows that the low lands are wet until the end of

October. Observation of the base flow from the total water yield using the above

equation for eight years of data (1986 – 1993) showed a rapid decrease of base flow

from 1.02 mm on daily basis (end of October) to 0.73 mm of daily contribution (mid

of November) and then to 0.48 mm of daily contribution (end of November). Once the

base flow becomes 0.48 mm the change in flow decreases, for example, this value

0.48 mm continues without changing for the next three months and then drops to 0.26

mm. These conditions continue for the next three months without change. Finally, this

Page 63: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

45

amount changes because of the next rainy season. With this brief analysis of base flow

and with the assumption that the groundwater contribution to streams takes a longer

time than lateral flow, the minimum contribution of groundwater and the maximum

contribution of lateral flow can be calculated. That is, the rapid change in the base

flow contribution is due to the rapid decrease in the lateral flow. Thus, if the daily

base-flow contribution is equal to or less than 0.48 mm then the contribution is totally

from ground water and the contribution of lateral flow is zero. If the daily contribution

of base-flow is greater than 0.48 mm then the groundwater contribution to the base-

flow will be 0.48 mm with the rest coming from the lateral flow. Here, the increase in

groundwater contribution during rainy season is assumed to be zero, which may not be

correct but is a necessary assumption to estimate the minimum contribution of

groundwater. This is an area for further research.

Therefore;

mmQIfQQ iBASEiBASEiGW 48.0>=

And, 0=iLATQ

mmQIfmmOQ iBASEiGW 48.048. >=

And, mmQQ iBASEiLAT 48.0−=

Where QBASE i is the base flow contribution to the stream on day i in mm, QGW i is the

groundwater contribution to the stream on day i in mm, and QLAT i is the lateral flow

contribution to the stream on day i in mm.

By using these observed values the model was calibrated for annual average values.

The fine tuning was achieved by comparing the simulated and the observed values on

a monthly basis followed by comparisons on a daily basis.

Page 64: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

46

Sediment calibration was first done using parameters which affect the amount of soil

loss from each HRU. This includes subbasin and management parameters. After this

the sediment loss from the watershed was fine tuned by adjusting parameters

describing the characteristics of reaches.

Model calibration generally consists of statistical tests like Optimization of the Nash-

Sutcliffe Coefficient (ENS) (Santhi et al., 2001a; Cotter, 2002; Grizzetti et al., 2003).

The coefficient of determination (R2) suits model evaluation and is very sensitive to

extreme values (Harmel and Smith 2007). In this study, the coefficient of

determination (R2) and the Nash-Sutcliffe coefficient (ENS) were used.

The coefficient of determination (R2) is the square of the Pearson’s product-moment

correlation coefficient and describes the proportion of the total variance in the

observed data that can be explained by the model. R2 ranges from 0.0 to 1.0 with

higher values indicating better agreement (Legate and McCabe, 1999).

The second objective function used in this study was the Nash-Sutcliffe coefficient of

efficiency ENS which has been widely used to evaluate the performance of

hydrological models (Leavesly et al., 1983; Wilox et al., 1990; Arnold et al., 1999;

Kirsch et al., 2002). This coefficient ranges from minus infinity to 1.0, with higher

values indicating better agreement (Legate and McCabe, 1999). Nash and Sutcliffe

(1970) defined the coefficient as:

( )

⎥⎥⎥⎥⎥

⎢⎢⎢⎢⎢

⎟⎠⎞

⎜⎝⎛ −

−−=

=

=

N

i i

N

i iiNS

OO

POE

1

_

12

20.1

Page 65: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

47

Where Oi is the measured (observed) data, _O is the mean of measured data, Pi is the

modeled (predicted) data.

Model validation is similar to model calibration in that the predicted and measured

values are compared to determine the model fit to the observed data. For validation,

the data set of measured runoff and sediment should be different from that used for

calibration (White and Chaubey, 2005). In this study the data set for validating the

model was taken from the years 1995, 1996 and 2000.

Scenarios

Base Scenario (Scenario I)

This scenario presents the actual condition observed in the watershed. The watershed

is managed by using parallel terraces starting from the year 1986. The Fanya Juu were

adopted in this watershed starting from this time because of the efficiency of Fanya

Juu in decreasing soil loss and its acceptance by the farmers (Werner, 1986). To

include this management practice in the model the slope was assumed to be reduced

by 37.5% and the slope length is reduced by 50% for areas with slope greater than 5%.

Zero-Tillage (Scenario II)

In this scenario the disturbance of soil by tillage was assumed to be zero. The farmers

till their plot more than six times before planting teff (12% of the land in the

watershed grows this crop). In addition to this, the farmers also had a strong belief on

plowing their lands again and again for a better yield. The practicability of zero-tillage

for fields growing teff is doubtful. Some research (Habtegebriel et al., 2007) has

shown the impact of tillage on teff fields. It showed that the yield of grain decreased

by 4.2% to 6.9% when transforming from conventional agriculture to minimum

tillage. Thus, in this scenario zero-tillage is assumed to be practiced for cultivated

fields except in fields growing teff.

Page 66: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

48

Parallel Terraces (Scenario III)

In this scenario it was assumed that the farmers reduce the width of their existing and

terraced plot by 50%. Management practices other than this in the watershed are not

changed. Farmers still practice contour plowing, conventional plowing and use the

same crop calendar. To incorporate this, the slope length is further reduced by 50% on

agricultural fields with slope greater than 5% (see illustration in Figure 12a and 12b).

As time goes on the slope of the fields is reduced due to the deposition of soil at the

foot of each terrace. To incorporate this, in this scenario, the slope of each agricultural

plot is further reduced by 25% for areas where the slope is greater than 5%.

a)

b)

Figure 12: Newly constructed Fanya Juu terrace (a) and Fanya Juu after five years of construction (b). * Pictures taken from: http://www.iwmi.cgiar.org/africa/west/projects/Adoption%20technolgy/rainwaterharvestin/50-Fanya%20juu.htm

Page 67: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

49

Forestation (Scenario IV)

Foresting all cultivated fields is impractical and impossible for many reasons. Instead,

foresting bush lands and degraded agricultural fields is more feasible. The degraded

agricultural fields in the watershed can easily being identified because the farmers

plant nigger seed on these fields. In some cases these fields have been converted and

used for Eucalyptus planting. This shows that the farmers have already started

changing degraded agricultural fields on the top of the watershed into forest. Thus, in

this study, this scenario was established by replacing the bush lands and degraded

agricultural fields by forest this covers 9% of the watershed in area.

No-Terraces (Scenario V)

In this scenario the farmers in the watershed are assumed not to practice terracing at

all. This provides a way of showing how much soil is lost or conserved and also the

effect of terracing on river discharge. The scenario was developed by ignoring the

slope decrease made in the base scenario and the decrease in the slope length.

Page 68: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

50

CHAPTER FOUR

RESULTS AND DISCUSSIONS

Sensitivity Analysis

The groundwater parameters were found to be the parameters to which the flow was

most sensitive, in particular: the base flow alpha factor (ALPHA_BF) in days;

Threshold depth of water in the shallow aquifer required for return flow to occur

(GWQMN) in mm; Threshold depth of water in the shallow aquifer for "revap" or

percolation to the deep aquifer to occur (REVAPMIN) in mm and Groundwater

"revap" coefficient (GW_REVAP). The flow was also found to be sensitive to soil

properties: soil evaporation compensation factor (ESCO); depth from soil surface to

bottom of layer (SOL_Z) in mm and available water capacity of the soil layer

(SOL_AWC) in mm /mm of soil depth. The flow was also sensitive to crop

parameters: maximum potential leaf area index (BLAI) which is a parameter to

quantify the density of the plant and maximum canopy storage (CANMX) in mm H2O.

The most sensitive parameters for the sediment prediction were those used for

calculating the maximum amount of sediment that can be entrained during channel

routing, which includes the exponent, factors and channel properties. The coefficient

SPCON and the exponent SPEXP are defined by the user for calculation of the

maximum sediment concentration in channels. The sediment yield from the watershed

was very sensitive to these values, which affect deposition in channels. The channel

properties, Manning’s ‘n’ value for tributary channels (CH_N) affects the time of

concentration and indirectly the peak discharge in the channel. Factors like the channel

cover CH_COV and the channel erodibility CH_EROD linearly influence the soil loss

from channels. Sediment yield was also very sensitive to effective hydraulic

conductivity in the main channel alluvium (CH_K) in mm/hr. Finally, the sediment

Page 69: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

51

predictions were also found to be sensitive to the management practices in the

watershed, which is represented by USLE_P.

The ranking of variables used in the sensitivity analyses for flow and sediment

parameters are listed below (Table 7).

Table 7: The most sensitive parameters for flow and sediment

Flow Parameter Sediment Parameters Ranking Alpha_Bf Spcon 1 Gwqmn Ch_N 2

Esco Spexp 3 Sol_Z Alpha_Bf 4 Blai Ch_Cov 5

Sol_Awc Ch_Erod 6 Revapmin Ch_K2 7

Canmx Blai 8 Gw_Revap Usle_P 9

Sol_K Gwqmn 10

Model Calibration and Validation

Model calibration followed sensitivity analysis. Flow and sediment calibration for the

Anjeni watershed was conducted for the years 1986 to 1993. Two years, 1984 and

1985, were used for model initialization. Likewise, flow and sediment validation for

the Anjeni watershed was carried out for the years 1995, 1996 and 2000. These years

were selected based on the availability of data.

Initially, the model was calibrated on an annual basis. The surface runoff, base flow

and the total water yield were calibrated first. The simulated values for these variables

before calibration are shown below in Table 8.

Page 70: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

52

Table 8: Output variables; simulated before calibration and observed values in mm Total Water Yield Base Flow Surface Flow

Actual 700.6 503.2 197.4 SWAT 1137.6 801.9 337.4

The model over estimated the flow; thus parameters (Table 9) were adjusted in order

for the simulated output values to meet the actual annual averages. The parameters

were adjusted further to fine tune the simulation and the changes in parameters are

shown in Table 9.

Table 9: Values of parameters used for calibration

Parameters for flow calibration Parameters for Sediment calibration ALPHA_BF 0.2 USLE_P 0.8 GW_REVAP 0.2 USLE_K Reduced by 0.02 REVAPMN 0.001 CH_COV 0.8 RECHRG_DP 0.65 CH_EROD 0.5 SURLAG 1 PRF 1.2 ESCO 0.8 ADJ_PKR 0.5 EPCO 0.9 SPCON 0.001 EDC 0.43 SPEXP 1 SOL_AWC 10% Increase SLOPE 37.5% Decrease SOL_BD 5% Increase SLSUBBSN 50% Decrease SOL_K 10% Decrease ISED_DET Monthly maximum

ALHA_BF= the baseflow alpha factor; GW_REVAP=groundwater "revap" coefficient; REVAPMN= Threshold depth of water in the shallow aquifer for "revap" or percolation to the deep aquifer to occur; RECHRG_DP= Deep aquifer percolation fraction; SURLAG=surface runoff lag coefficient; ESCO=soil evaporation coefficient; EPCO=plant uptake compensation factor; EDC=Effective Depth Coefficient in the .bee file; SOL_AWC=soil layer available water capacity; SOL_BD=soil moist bulk density; SOL_K=soil saturated hydraulic conductivity; USLE_P=Universal soil loss equation management factor; USLE_K=universal soil loss equation soil factor; CH_CV=channel cover factor; CH_EROD=channel erodibility; PRF= Peak rate adjustment factor for sediment routing in the main channel; ADJ_PKR=peak rate adjustment factor; SPCON= Linear parameter for calculating the maximum amount of sediment that can be re-entrained during channel sediment routing; SPEXP= Exponent parameter for calculating sediment re-entrained in channel sediment routing; SLOPE=average slope steepness (m/m); SLSUBBSN=average slope length.

Page 71: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

53

The flow was calibrated using the above parameters to improve the objective functions

(R2 and NSE). The baseflow recession constant (the baseflow alpha factor

(ALPHA_BF)) which is a direct index of groundwater flow response to changes in

recharge was adjusted to 0.2. The groundwater "revap" coefficient (GW_REVAP)

which controls the rate of transfer of water from the shallow aquifer to the root zones

was adjusted to 0.2. The threshold depth of water in mm in the shallow aquifer for

"revap" or percolation to the deep aquifer to occur (REVAPMN), the soil evaporation

coefficient (ESCO), and the plant uptake compensation factor (EPCO) were adjusted

to 0.001, 0.8 and 0.9 respectively. The effective depth coefficient (EDC) parameters

needed to calibrate the water balance. To calibrate flow, the EDC values adjusted

between 0 and 1. The EDC partitions excess moisture between percolation and runoff,

so an EDC of 1 would partition excess to runoff and EDC of 0 to percolation. These

EDC values for each soil type in the watershed were adjusted to be 0.43. These

parameters effectively modify the depth distribution used to meet the soil evaporative

demand to account for the effect of capillary action, crusting and cracks.

Calibration of sediment yield was achieved by decreasing slope (SLOPE) by 37.5%

only for agricultural areas with slope greater than 5% when by considering the terrace

practiced in the watershed since 1986 (SCRP Report, 2000). The slope length of each

plot in agricultural fields with slopes greater than 5% was reduced by 50% to consider

the break in slope length due to the provision of terraces. The reduction of slope

lengths makes the SLSUBBSN for the agricultural areas 30m to 40m which is the

average terrace interval in the watershed. For small plots and micro-watersheds in

particular, the variability of triangular distribution is unrealistic (Neitsch et al., 2005).

So, ISED_DET (a code governing the calculation of daily maximum half-hour runoff)

was assigned to work with monthly maximum half-hour rainfall value. The average

Page 72: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

54

annual total flow, the base flow and the surface runoff after calibration are shown

below (Table 10).

Table 10: Annual average output variables; simulated after calibration and observed

Total Water Yield Base Flow Surface Flow Actual 700.6 503.2 197.4 SWAT 712.8 522.7 196.1

The model goodness-of-fit was evaluated both on a monthly and on a daily basis as

shown in Table 11. The linear graphs for the measured and simulated values both for

flow and sediment on daily basis for calibration and validation are shown in Figure 13

to Figure 16. The linear graphs for the measured and simulated values both for flow

and sediment on a monthly basis for calibration and validation are shown in Figures

17 through 20.

Table 11: Coefficient of determination and the Nash – Sutcliffe Coefficients for calibration and validation both in daily basis and monthly basis

Calibration Validation R2 NSE R2 NSE

Flow 0.54 0.47 0.57 0.41 Daily Sediment 0.03 -0.23 0.05 -0.17

Flow 0.92 0.91 0.97 0.93 Monthly Sediment 0.56 0.55 0.89 0.82

Page 73: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

55

Figure 13: Coefficient of determination for simulated flow in calibration on a daily basis

Figure 14: Coefficient of determination for simulated sediment in calibration on a daily basis

Page 74: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

56

Figure 15: Coefficient of determination for simulated flow in validation on a daily basis

Figure 16: Coefficient of determination for simulated sediment in validation on a daily basis

Page 75: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

57

Figure 17: Coefficient of determination for simulated flow in calibration on a monthly basis

Figure 18: Coefficient of determination for simulated sediment loss in calibration on a monthly basis

Page 76: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

58

Figure 19: Coefficient of determination for simulated flow in validation on a monthly basis

Figure 20: Coefficient of determination for simulated sediment loss in validation on a monthly basis

Page 77: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

59

Analysis of Results

The model underestimated peak flow from the watershed in some years and over

predicted in some years. The flow in the dry season which is determined by the

groundwater is also under estimated; see Figures 21 - 24 for comparison of measured

and simulated values on a daily basis. The daily simulated flow is greater than the

observed flow at the beginning of the rainy season. In these times even high rainfall

cannot produce more than 1mm of runoff in a day. This indicates that surface runoff is

produced when the soil is saturated and there is a delay in the lateral flow.

The runoff in the dry season (December to May) is mainly from ground water and

surface runoff if there is rain on that particular day; see Figures 25 - 28 for comparison

of measured and simulated values on a monthly basis. So, the contribution of lateral

flow is zero in this season. The runoff produced after a heavy storm in the dry season

simulated by the model is less than the observed runoff. This shows that there are

some portions in the watershed which produce surface runoff immediately after rain

starts falling because of the very low infiltration rate (based on physical observation in

the watershed). This idea is reinforced by interviewing farmers in the watershed. The

farmers were asked at which time of the rainy season is runoff produced immediately

after rain starts to fall. The result is as shown in Table 12.

Table 12: Number of farmers interviewed for identification of the process of runoff production

Number of Farmers Interviewed Immediate1 Medium2 Slowest3

Beginning of Rainy Season 1 18 31 Middle of Rainy season 43 7 0 End of rainy season 6 25 19

1Number of farmers explaining that the surface runoff production is the most immediate and fastest of all the other two cases 2Number of farmers expressing that the surface runoff production after rain storm is faster (more immediate) than one of the two cases and slower than the other case 3Number of farmers explaining that the surface runoff production is after a storm is the slowest of all the three cases

Page 78: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

60

The daily runoff production gradually increases, as the rainy season progresses but the

rate of increase in simulated runoff at the beginning of the rainy season is greater than

is actually observed. This is influenced by the initial soil moisture in the soil of the

watershed. Thus, the model over predicts the soil moisture in the watershed. During

August, the month of highest rainfall, the model slightly under predicts the water yield

from the watershed. At the end of the rainy season, the water yield from the watershed

takes longer to come to recession that is the decrease in flow through time is less than

that of actually observed.

The surface runoff production from each HRU was analyzed. Areas in the watershed

with Haplic Lixisols soils were contributed the least surface runoff to the reach. Haplic

Lixisols has the lowest least clay content and the highest sand content of all the soils

in the watershed and are characterized by high saturated hydraulic conductivity,

7mm/hr on in the top layer and 25mm/hr on the lower layers. Most of the areas having

Humic Alisols and Eutric Regosols produce large amount of surface runoff. Humic

Alisols have the lowest hydraulic conductivity (1mm/hr) of all the soils found in the

watershed and high clay content only in the top layer and high percentage of sand in

the lower layers. These soil properties make areas covered with this soil type produce

large amounts of surface runoff as infiltration excess flow.

The model over predicts the soil loss at the beginning of the rainy season as observed

in the whole of the calibration period. The over prediction in runoff at the beginning of

the rainy season resulted in over estimation of soil loss from the watershed at the

beginning of the rainy season. SWAT fails to estimate the peak soil loss in most of the

years. The coefficient of determination (R2) for the daily simulation shows that the

model fails to estimate daily soil loss. However, on a monthly basis the soil loss

Page 79: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

61

simulated by the model had a similar pattern as that of the measured soil loss from the

watershed, even though it fails to estimate the peak soil loss in each year. The

coefficient of determination (R2) and the Nash-Sutcliffe Coefficient, 0.56 and 0.55

respectively, were satisfactory. So the model can be used for further analysis on soil

loss for different scenarios.

Figure 21: Hydrograph of the observed and simulated flow from the watershed for the validation period on a daily basis

Figure 22: Comparison of observed and simulated sediment loss from the watershed for the validation period on a daily basis

Page 80: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

62

Figure 23: Hydrograph for Anjeni watershed on a daily basis for the whole period of calibration showing precipitation, flow from the watershed and sediment loss in a daily basis for the observed and simulated values

Figure 24: Comparison of observed and simulated sediment loss on a daily basis from Anjeni watershed for the whole calibration period

Page 81: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

63

Figure 25: Hydrograph of the observed and simulated flow from the watershed for the calibration period on a monthly basis

Figure 26: Comparison of observed and simulated sediment loss from the watershed for the calibration period on a monthly basis

Page 82: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

64

Figure 27: Hydrograph of the observed and simulated flow from the watershed for the validation period on a monthly basis

Figure 28: Comparison of observed and simulated sediment loss from the watershed for the validation period on a monthly basis

Page 83: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

65

Gully Erosion

The model uses the Universal Soil Loss Equation to estimate the soil loss from each

HRU and also considers channel degradation. This means that the model does not

consider gully erosion in the watershed. However, consideration of gully erosion is

important because there is a big gully in the watershed with an average depth of 8m.

Aerial photographs taken in 1957 and 1982 and a Google Earth image from 2008 were

used to estimate the soil loss contribution from the gully to the watershed outlet, see

Figure 29. The resolution of aerial photos was (after digitizing and geo-referencing)

6.00 m X 6.00 m.

Figure 29: Location and comparison of the largest gully in the watershed for different years

The measured annual average soil loss from 1986 to 1993 was 2785.36 tons/yr, with a

minimum soil loss of 663.39 tons/yr and a maximum soil loss of 6979.77 tons/yr. The

average annual soil loss from the watershed due to gully erosion from 1982 to 2008

Page 84: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

66

was calculated to be 550.15 tons/yr, which is 83% of the minimum annual average soil

loss from the watershed (663.39 tons/yr), 8% of the maximum annual average soil loss

from the watershed (6979.77 tons/yr) and 20% of the average of all the annual average

soil loss (2785.36). This shows that there is a significant sediment contribution from

gully erosion to the total sediment yield from the watershed. Gully erosion occurred

suddenly and the contribution in each year varies highly. This makes the prediction of

soil loss in each year from gullies difficult. Since the separation of soil loss from the

fields and from gullies is impractical, model calibration is also uncertain. Thus, SWAT

fails to simulate soil loss from the watershed in those years where very low sediment

yield and very high sediment yield were observed. The area of the gully in the

watershed is indicated in Table 13.

Table 13: Area of the gully in m2 for the three years

Area_1957 Area_1982 Area_2008 1055.88 1644.24 2761.73

The farmers in the watershed started constructing terraces in 1984. The above analysis

in Table 14 showed that the rate of gully formation after 1982 increases from 301

tons/yr (during the time when there is no construction of terrace in the watershed) to

550 tons/yr (during the time when the farmers are practicing terraces). Construction of

terraces encourages infiltration. As much water infiltrated to the bottom layers, there

will be a high chance for occurrence of soil piping. Physical observations in the

watershed showed the occurrence of soil piping in the watershed. Soil piping is a

major cause of channel head extension, rilling and gullying in landscapes as diverse as

semi-arid climates from Arizona to East Africa (Jones, 1994). Parker and Higgins

(1990) and Dardis and Beckedahl (1988) all regard piping as a major cause of

Page 85: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

67

gullying. In addition to this, gully formation may occur due to the under-cutting of

sides of gullies by a high velocity runoff.

Table 14: Calculation of mass of soil loss due to gully erosion for the three different years

Years Increase in Area (m2)

Volume lost (m3)

Mass lost (tons)

Rate of loss (tons/year)

From 1957 – 1982 588.36 4706.88 7531.01 301.24

From 1982 – 2008 1117.49 8939.92 14303.87 550.15

Spatial Variation of Runoff and Soil Erosion

Areas covered with teff, barely and corn produces the maximum surface runoff in the

watershed. Most of the areas covered with corn produce surface runoff varying from

232 – 393 mm of water in a year. Most of the areas covered with teff produce an

annual surface runoff varying from 532 – 1402 mm of water. Even if most of the plots

covered by barley produced the same amount as the annual average surface runoff

from the watershed (197 mm) some plots produce runoff varying from 531 – 1402

mm. Areas in the watershed covered with forest produce the least surface runoff

varying from 0 – 100 mm of water. Based on location, the middle parts of the

watershed are observed to produce high surface runoff. The top pars of the watershed

with a slope less than 5% are also found to produce high surface runoff. The map in

Figure 30 shows the areal variation of surface runoff.

Page 86: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

68

Figure 30: Map of extent of surface runoff in each HRU

Most parts of the watershed contribute to soil loss ranging from 0 – 1 tons/ha.

Significant correlation was not observed between a high rate of soil loss and soil types

found in the watershed. The highest sediment loss from fields is observed to be from

teff and corn. Most of the plots covered with teff and corn contribute 50 – 334 tons/ha

soil. Considering fields coved with barely, those that produce high surface runoff are

observed to lose high amount of soil varying from 20– 50 tons/ha. The map in Figure

31 shows the areal variation of sediment loss.

Page 87: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

69

Figure 31: Map of extent of sediment loss from each HRU

Scenarios

The different scenarios were compared and analyzed on a monthly basis. The water

yield and the sediment yield from the watershed were analyzed separately.

Flow

Base scenario (Scenario I), Scenario II (no tillage activity) and forestation (scenario

IV) produce essentially the same monthly flow pattern and amount except some slight

change at the end of the rainy season for twenty years’ worth of observed simulated

data, see Figures 32 and 35. Scenario III (Terrace) is observed to reduce the flow from

the watershed and found to give the lowest water yield compared to the other

scenarios. The decrease in runoff due to terrace practice (Scenario III) is considerably

higher at the maximum point of the runoff for a particular year and at also the same at

Page 88: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

70

the end of rainy season. No decrease in runoff is observed at the beginning of the rainy

season. But the pattern of runoff in all the scenarios is the same. During the rainy

season, terracing (Scenario III) results a decrease of 1.7% in runoff. This is likely a

direct consequence of changing the slope in the model and therefore likely not realistic

since interflow depends on the overall slope and not on the slope of the terraces.

Unlike Scenario III the percentage decrease in the flow in the dry season is not very

high in Scenario IV resulting in a 3.6% decrease. Thus water consumption during dry

season is not affected by this management option (forestation). The decrease in water

yield from the watershed in the rainy season due to forestation is 1.3% which is not

substantially less than the runoff decrease due to Scenario III.

In Scenario V, avoiding terraces resulted in an increase in flow. The flow increase in

this scenario has the same pattern as the decrease in flow due to the addition of more

terraces (Scenario III) and the runoff in the rainy season increases by 1.24% (Figure

33).

Annually, Scenario III is found to yield the least flow. Forestation conserve further

10mm of water compared to the water conserved by the existing conservation

structures (Figure 34). Avoiding terraces leads to further 15mm loss of water.

Page 89: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

71

Figure 32: Comparison of average monthly flow in each scenario

Figure 33: Pattern of decrease or increase in flow in different scenarios compared to the base scenario

Page 90: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

72

Figure 34: Annual average water yield form the watershed in each scenario

Sediment

Around 99% of the mean annual suspended sediment load is transported between May

and October. The months November, December, January and February are months

when there is no soil loss from the watershed. Thus these months are not the main

concern in the analysis of sediment yield and when comparing scenarios. Unlike the

differences observed in flow among scenarios, the differences observed in sediment

loss among different scenarios have a similar pattern throughout a year (Figures 36

and 37).

Page 91: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

73

Figure 35: Hydrograph in each scenario on a monthly basis (see table in Appendix VI)

Figure 36: Sediment yield from the watershed in each scenario on a monthly basis (see table in Appendix VI)

Page 92: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

74

Figure 37: Comparison of average monthly sediment loss from the watershed in each scenario

Zero-tillage activities (Scenario II) results in a soil loss decrease from April to

September when compared to the base scenario. There was a 6% decrease in soil loss

at the beginning of the plowing season (from April to June) and a 1.64% decrease

during the rainy season.

In Scenario III (terracing), decreases in soil loss in all the months were observed. In

the months from March to April there is a 61% decrease in soil loss from each HRU

due to the provision of more terraces (0.04t/ha). These are the months when plowing is

started in each year. For the rainy season (June, July, and August) there is a decrease

in soil loss 3.07 t/ha/month (64% of the soil loss in this season in the base Scenario).

For the months at the end of rainy season (September, October and November) the

decrease in soil loss is 0.34% tons/ha/month which is 61% of the soil loss rate in these

months in the base Scenario. Increasing terraces (Scenario III) shows a constant

decrease of soil loss in percentage in all the months under consideration.

Page 93: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

75

Forestation (Scenario IV) results in a decrease of soil loss from April to November,

which is from the start of the plowing season to the end of the rainy season. There is

19% decrease in soil loss from each filed compared to the sediment yield in the base

scenario in these months. The soil loss decreased by 0.011 tons/ha/month which is

13% of the soil loss rate in these months in the base Scenario. During the heaviest rain

in the rainy season (June, July and August) this management option resulted in a

decrease of soil loss by 0.86 t/ha/month, which is 18% of the soil loss rate in the base

Scenario in these months. At the end of rainy season (September, October and

November) a decrease in soil loss due to the forestation of some parts of the watershed

is 0.1 t/ha/month which is 21% of the soil loss rate observed in these months in the

base scenario (Figure 38).

The Scenario that did not consider the practice of terraces in the watershed (Scenario

V) resulted in a soil loss increase with the same pattern as the decrease in soil loss due

to the construction of more terraces (Scenario III). But Scenario V resulted in a higher

soil loss compared to the amount of soil conserved by the construction of more

terraces.

The annual soil loss conserved by further terracing is 10t/ha/yr and loss of soil by

avoiding the terraces is 20t/ha/yr. Forestation can reduce soil loss further by 5t/h/yr

(Figure 39).

Page 94: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

76

Figure 38: Pattern of decrease or increase in sediment loss from the watershed in different scenarios compared to the base scenario

Figure 39: Annual average water yield form the watershed in each scenario

Page 95: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

77

CHAPTER FIVE

CONCLUSIONS AND RECOMMENDATIONS

The result from sensitivity analysis showed that the runoff is most sensitive to the

groundwater parameters. Thus, for further accuracy of the model a detailed study of

the groundwater properties (the groundwater depth, the alpha factor etc) are essential.

SWAT-WB approach avoids the SCS curve number method and instead uses water

balance approach and introduce a new coefficient EDC which affect the distribution of

runoff and percolation. EDC needs careful calibration and investigation.

The SWAT-WB model under-estimates flow in the middle of the rainy season. This

could possibly be corrected and the model performance improved by more accurately

determining the value of soil profile depth, of each soil type, which is available for

saturation (D) and is directly influences the EDC.

The runoff production in Anjeni watershed is mainly of saturation excess flow and

infiltration excess flow being a trivial runoff production process in the watershed. This

trivial runoff production process results in the model under-prediction of runoff during

the dry season. As the result of the main runoff production process in the rainy season,

saturation excess flow, high runoff in a day in the watershed is not produced due to

high rainfall but instead it depends on the antecedent moisture conditions.

Simulation of flow by the model is found to be excellent on a monthly basis and

satisfactory on a daily basis. Thus SWAT-WB performs very well and can be used for

runoff simulation for watersheds with the same runoff production process as that of

Anjeni. As studies like Lui et al., 2008 and Collick et al., 2008 have shown, most of

the watersheds in the Ethiopian highlands behave in a way that is similar to Anjeni

Page 96: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

78

watershed. That is, the runoff production in these areas is largely due to saturation

excess flow from saturated areas in the landscape. Therefore, this model can be used

as a tool to analyze hydrological processes in Ethiopian highlands.

The sensitivity analysis for soil erosion showed that the soil loss is most sensitive to

channel properties, exponent and factor for calculation of maximum sediment

concentration that can be transported by the water. The channel properties define how

loose and erodible the channel walls are. Generally, sensitivity analysis showed that

the sediment loss from the watershed is more sensitive to channel properties than HRU

properties. 70% of the top 10 parameters for which the model is most sensitive are

those which define channel properties.

In most of the years the simulated value of soil loss from the watershed fails to

estimate the peak soil loss in a year. Gully erosion in the watershed was found to be

very significant. The rate of gully erosion increases by 240 tons/yr after the

construction of terraces starting from 1984. Thus since SWAT has no ability to predict

gully erosion, soil loss tends to be under predicted.

With the assumption that SWAT totally fails to simulate gully erosion, the effect of

application of terraces in the watershed starting from the year 1984 is found to save

2064 tons/yr. This value is calculated with the assumption that the increase in gully

erosion is totally due to the construction of terraces. With the same concept further

construction of terraces in the future (Scenario II) saved 932 tons/year. Even if the

plots are divided into two for further terrace construction the amount that we can

further save is only 932 tons/yr which is 45% of the amount saved by the existing

conservation practices.

Page 97: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

79

If gully erosion is controlled with an efficiency of 90% the result would be a savings

of 495 tons/year of soil. Forestation reduces erosion by 333 tons/year. Thus, gully

rehabilitation with an erosion controlling efficiency of 90% and forestation together

would save 828 tons of soil in a year. These two management options are feasible and

can be accepted by the farmers without much effort in addition to saving 88% of the

amount of soil that can be saved by further construction of terraces. Further

construction of terraces in the watershed is not feasible since the plots are only 20m –

30m width and it is also impractical to divide these plots. This management option

also has a negative influence on the availability of water in the dry season. Thus,

further construction of terraces is not a feasible option for conserving soil and

availability of water.

Zero-tillage can save 45 tons of soil loss in a year which is comparatively very small.

This management option also requires considerable effort from farmers and is unlikely

to be acceptable to them. Thus, forestation of degraded lands and bush lands together

with rehabilitation of gullies with 90% efficiency are the best option for controlling

soil erosion and a sustainable development option in the watershed.

This study hasn’t incorporated the change in productivity due to provision of each

management options. Thus, the effect of these management options on productivity

especially further construction of terraces needs to be studied. If there is a

considerable productivity change due to further construction of terraces then this

option can be practiced as it saved 104 tons/yr than the management option

recommended in this study. In addition, the farmers may need to change their practices

to adapt climate change. If implementing the management options is planned,

Page 98: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

80

considering the impact of climate change on these management options, formulated in

this study, is very important.

Page 99: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

81

REFERENCES

Abegaz Gizachew, 1995. Soil erosion assessment: Approaches, magnitude of the

problem and issues on policy and strategy development (Region 3). Paper

presented at the Workshop on Regional Natural Resources Management

Potentials and Constraints, Bahir Dar, Ethiopia, 11–13 January 1995. Bureau

of Natural Resources and Environmental Protection, Bahir Dar, Ethiopia. 9 pp.

Arnold, J.G., Allen P.M., Muttiah R., and Bernhardt G., 1995. Automated Base-Flow

Separation and Recession Analysis Techniques. Ground Water 33:1010-1018.

Arnold, J.G., and Allen P.M.. 1996. Estimating Hydrologic Budgets for Three Illinois

Watersheds. Journal of Hydrology. 176: 57-77.

Arnold, J.G., and P.M. Allen. 1999. Automated methods for estimating baseflow and

ground water recharge from streamflow records. Journal of the American

Water Resources Association 35:411-424.

Arnold, J.G., J.R. Williams, R.H. Griggs, and N. B. Sammons. 1990. SWRRB - A

Basin Scale Simulation Model for Soil and Water Resources Management.

Texas A & M Press.

Arnold, J.G., Williams, J. R., Srinivasan, R. and King K. W., 1996, SWAT-Soil and

Water Assessment Tool, USDA-ARS, Temple, Texas.

Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998. Large area

hydrologic modeling and assessment part I: model development. J. American

Water Resources Association 34(1):73-89.

Arnold, J.G., R. Srinivasan, R.S. Muttiah, and P.M. Allen. 1999. Continental Scale

Simulation of the Hydrologic Balance. Journal of American Water Resource

Association. 35(5): 1037-51.

Page 100: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

82

Arnold, J.G., R.S. Muttiah, R. Srinivasan, and P.M. Allen. 2000. Regional Estimating

of Baseflow and Groundwater Recharge in the Mississippi River Basin.

Journal of Hydrology. 227: 21-40.

Bekele, S., Holden, S.T. 1998. Resource degradation and adoption of land

conservation technologies in the Ethiopian Highlands: A case study in Andit

Tid, North Shewa. Agricultural Economics 18: 233-47

Bosshart, Urs., 1995. Catchment Dischrge and Suspended Sediment Transport as

Indicators of Physical Soil and Water Conservation in the Minchet Catchment,

Gojam Research Unit. A case study of the northern Highlands of Ethiopia.

SCRP Research Report, Berne University, Addis Ababa, 104p.

Bosshart, Urs, 1997. Measurement of river discharge for the SCRP research

catchments : gauging station profiles. Berne University in Association with

Ministry of Agriculture, Ethiopia. Soil and Conservation Research Program;

Report 31, 1997.

Collick, A.S., Z.M. Easton, E. Adgo, S.B. Awulachew, G. Zeleke, and T.S. Steenhuis.

2008. Application of a physically-based water balance model on four

watersheds throughout the upper Nile River Basin. Paper presented at the

Workshop on Hydrology and Ecology of the Nile River Basin under Extreme

Conditions, June 16-19, Addis Ababa, Ethiopia.

Cotter, A., 2002. Critical Evaluation of TMDL Data Requirements for Agricultural

Watersheds. M.S. thesis, University of Arkansas, Fayetteville, Arkansas.

Dardis, G.F. and H.R. Beckedahl, 1988. Drainage evolution in an ephemeral soil pipe-

gully system, Transkei, Southern Africa, in Dardis, G.F.and Moon, B. P. (eds)

Geomorphological studies in Southern Africa, Balkema, Rotterdam, 247-65.

Decoursey, D. G. and E. H. Selly, 1988. Mathematical models for point sources water

pollution control. Journal of Soil and Water Conservation. 44(2): 568-576.

Page 101: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

83

De Ploey J., 1974. Mechanical Properties of Hill slopes and their Relation to Gullying

in Central Semiarid Tunisia, Zeitschrift fur Geomorphologie 21, 177-90.

Easton, Z.M., D.R. Fuka, M.T. Walter, D.M. Cowan, E.M. Schneiderman, and T.S.

Steenhuis. 2008. Re-conceptualizing the Soil and Water Assessment Tool

(SWAT) model to predict run off from variable source areas. Journal of

Hydrology, 348(3-4): 279-91.

Fournier, F., 1960. Climat et Erosion. Presses Universitaires de France, Paris.

Green WH, Ampt GA., 1911. Studies on soil physics; The flow of air and water

through soils. Journal of Agricultural Sciences; 4: 11-24.

Grizzetti, B., F. Bouraoui, K. Granlund, S. Rekolainen, and G. Bidoglio, 2003.

Modelling Diffuse Emission and Retention of Nutrients in the Vantaanjoki

Watershed (Finland) Using the SWAT Model. Ecological Modelling

169(1):25-38.

Habtegebrial K., Singh B.R., Haile M., 2007. Impact of tillage and nitrogen

fertilization on yield, nitrogen use efficiency of tef (Eragrostis tef (Zucc.)

Trotter) and soil properties. Soil and Tillage Research (94) 55–63.

Haregeweyn, N. and Yohannes, F. 2003. Testing and evaluation of the agricultural

non-point source pollution model (AGNPS) on Augucho catchment, western

Hararghe, Ethiopia. Agriculture Ecosystems & Environment 99 (1-3): 201-212.

Hargreaves G. L., Hargreaves G. H., Riley J. P. 1985. Agricultural benefits for

Senegal River basin. Journal of Irrigation and Drainage Engineering 1985;

111(2): 113-124.

Harmel R. D., Smith P. K., 2007. Consideration of Measurement Uncertainty in the

Evaluation of Goodness-of-fit in hydrologic and Water Quality Modeling.

Journal of Hydrology 337, 326-336.

Page 102: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

84

Herweg, K.and Stillhardt, B. 1999. The Variability of Soil Erosion in the Highlands of

Ethiopia and Eritrea. Research Report 42. Soil Conservation Research Program

(SCRP), Centre for Development and Environment, University of Berne.

Hewlett, J.D., Hibbert A.R., 1967. Factors affecting the response of small watersheds

to precipitation in humid areas, p. 275-290, In W. E. Sopper and H. W. Lull,

eds. Forest Hydrology. Pergamon Press, Oxford.

Hurni, H., 1982. Inception Report. Soil Conservation Research Project, Vol. I.

University of Berne, Switzerland.

James, L.D. and S.J. Burges, 1982. Selection, Calibration, and Testing of Hydrologic

Models. In: Hydrologic Modeling of Small Watersheds, C.T. Haan, H.P.

Johnson, and D.L. Brakensiek (Editors). ASAE Monograph, St. Joseph,

Michigan, pp. 437-472.

Jones, J.A.A., 1994. Subsurface flow and subsurface erosion: further evidence on

forms and controls, in Stoddart, D.R. (ed.). Process and Form in

Geomorphology, Routledge.

Kefeni Kejela. 1987. Preliminary Assessment of the Impact of Soil Erosion and Its

Implication for Soil Conservation. A Case study based on the Soil Erosion

Data from Anjeni Station. A thesis presented to the School of Graduate Studies

in the Addis Ababa University, Addis Ababa, Ethiopia.

Kefeni Kejela 1985. The Soils of the Anjeni Area – Gojam Research Unit, Ethiopia.

Research Report 27. Soil Conservation Research Project, Centre for

Development and Environment, University of Berne.

Kirsch, K., A. Kirsch, and J.G. Arnold. 2002. “Predicting Sediment and Phosphorus

Loads in the Rock River Basin using SWAT.” Trans. ASAE 45(6): 1757-69.

Knisel, W.G., 1980. ed., CREAMS: A Field Scale Model for Chemicals, Runoff, and

Erosion from Agricultural Management Systems. Washington, D.C.: U.S.

Page 103: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

85

Department of Agriculture, Agricultural Research Service Conservation

Research Report No. 26.

Leavesley, G. H., R. W. Lichty, B. M. Troutman, and L. G. Saindon, 1983.

Precipitation-runoff modeling system user’s manual, U.S. Geol. Sur. Water

Resour. Invest. Rep. 83-4238, 207 pp.

Legate D. R., and McCabe G. Jr., 1999. Evaluating the Use of Goodness-of-fit

Measures in Hydrologic and Hydro-climatic Model Validation. Water

Resource Research, Volume 35, No. 1, 233-241.

Legesse D., C. Vallet-Coulomb and F. Gasse. 2003. Hydrological response of a

catchment to climate and land use changes in Tropical Africa: case study South

Central Ethiopia. Journal of Hydrology, 275: 67-85

Leonard, R.A., Knisel, W.G., Still, D.A., 1987. GLEAMS: groundwater loading

effects of agricultural management systems. Trans. Am. Soc. Agric. Eng. 30,

1403–1418.

Liu, B.M., A.S. Collick, G. Zeleke, E. Adgo, Z.M. Easton, and T.S. Steenhuis. 2008.

Rainfall-discharge relationships for a monsoonal climate in the Ethiopian

Highlands. Hydrological Processes, 22(7): 1059-67

Marshall, H.G., 1982. Breeding for tolerance to heat and cold. In: M.N. Christiansen

and C.F. Lewis (Editors), Breeding Plants for Less Favorable Environments.

Wiley, New York, pp. 47-70.

Matalas, N. C. 1967. Mathematical Assessment of Synthetic Hydrology. Water

Resources Research 3(4): 937-945

McHugh OV. 2006. Integrated water resources assessment and management in a

drought-prone watershed in the Ethiopian highlands. PhD dissertation,

Department of Biological and Environmental Engineering. Cornell University

Ithaca NY.

Page 104: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

86

McIntyre, DS 1974, in Loveday, J (ed) Methods of Analysis for Irrigated Soils.

Commonwealth Agricultural Bureaux Technical Commmunication No 54,

Farnham Royal, England.

Mohammed, A., Yohannes, F., Zeleke, G, 2004. Validation of agricultural non-point

source (AGNPS) pollution model in Kori watershed, South Wollo, Ethiopia.

International Journal of Applied Earth Observation and Geoinformation 6: 97–

109

Nash, J. E. and J. V. Sutcliffe. 1970, River flow forecasting through conceptual

models part I: A discussion of principles, Journal of Hydrology, 10, 282-290.

Neitsch, S.L., J.G. Arnold, J.R. Kiniry and J.R. Williams. 2001. Soil and Water

Assessment Tool User’s Manual, Version 2000.

Neitsch, S.L., J.G. Arnold, J.R. Kiniry, and J.R. Williams. 2002. Soil and Water

Assessment Tool User’s Manual, Version 2000. Grassland, Soil and Water

Research Laboratory, Temple, Texas GSWRL Report 02-02 Blackland

Research and Extension Center, Temple, Texas BRC Report 02-06. Texas

Water Resources Institute, College Station, Texas TWRI Report TR-192.

Neistch J. R., Arnold J. G., Kiniry J. R., Srinivasan R. and Williams J. R., 2004.

Input/Output File Documentation Version 2005. Grassland, Soil and Water

Research Laboratory, Agricultural Research Service, Blackland Research

Center, Texas Agricultural Experiment Station, Temple, Texas 76502

Neitsch SL, Arnold JG, Kiniry JR, Williams JR., 2005. Soil and Water Assessment

Tool, Theoretical Documentation: Version 2005. Temple, TX. USDA

Agricultural Research Service and Texas A&M Blackland Research Center.

Nicks A D., 1974. Stochastic generation of the occurrence, pattern and location of

maximum amount of daily rainfall. p. 154 – 171 In proc. Symp. Statistical

Page 105: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

87

Hydrology Aug – Sept 1971, Tuscon, AZ. US Department of Agriculture,

Misc, Publ. No. 1275.

Njau E. C., 1996. Generalized Derivation of the Angstrom and Angstrom-Prescott

Equations. Renewable Energy, Vol. 7, No. 1, pp. 105-108.

NRCS, Northern Plains Regional Office. 1996. State of the Land for the Northern

Plains Region. USDA, Natural Resources Conservation Service, Northern

Plains Regional Office, Lincoln, NE. 64 p.

Parker, G.G., Sr. and C. G. Higgins, 1990. Piping and pseudokarst in dry lands, in

Higgins, C.G. and Coates, D.R. (eds), Ground water geomorphology: the role

of subsurface water in earth-surface processes and landforms, Geological

Society of America Special Paper 252, 77-110.

Patterson K. P., 2007. Integrating Population, Health, and Environment in Ethiopia,

Population Reference Bureau: Making the Link, 2007.

Persaud N., Lesolle D., Ouattara M., Coefficients of the Angstrijm-Prescott equation

for estimating global irradiance from hours of bright sunshine in Botswana and

NigerAgricultural and Forest Meteorology (88) 27-35.

Reeve, M.J.,Carter, A.D.,1991.Water release characteristic. In:

Smith,K.A.,Mullins,C.E. (Eds.), Soil Analysis. Physical Methods. Marcel

Dekker, New York, pp. 111–160.

Reungsang P., Kanwar R. S., Jha M., Gassman P. W., Ahmad K., and Saleh A., 2005.

Calibration and validation of SWAT for the Upper Maquoketa River

Watershed. Working Paper 05-WP 396. Center for Agricultural and Rural

Development, Iowa State University, Ames, Iowa 50011-1070.

www.card.iastate.edu

Revfeim K.J.A., 1997. On the relationship between radiation and mean daily sunshine.

Agricultural and Forest Meteorology (86) 183-191.

Page 106: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

88

Richardson, C. W. 1981. Stochastic simulation of daily precipitation, temperature, and

solar radiation. Water Resource Research 17 (1): 182-190.

Richardson, C. W. and D. A. Wright. 1984. WGEN: a model for generating daily

weather variables. U.S. Department of Agriculture, Agricultural Research

Service, ARS-8.

Santhi, C., J.G. Arnold, J.R. Williams, L.M. Hauck, and W.A. Dugas, 2001a.

Application of a Watershed Model to Evaluate Management Effects on Point

and Nonpoint Source Pollution. Transactions of the American Society of

Agricultural Engineers 44(6):1559-1570.

Santhi, C., J.G. Arnold, J.R. Williams, W.A. Dugas, R. Srinivasan, and L.M. Hauck,

2001b. Validation of the SWAT Model on a Large River Basin With Point and

Nonpoint Sources. Journal of American Water Resources Association

(JAWRA) 37(5):1169-1187.

Setegn, S.G., Srinivasan, R., Dargahi, B. (2008): Hydrological Modeling in the Lake

Tana Basin, Ethiopia using SWAT model. The Open Hydrology Journal Vol 2,

49-62.

Sharpley, A.N. and J.R. Williams, eds. 1990. EPIC-Erosion Productivity Impact

Calculator, 1. Model documentation. U.S. Department of Agriculture,

Agricultural Research Service, Tech. Bull. 1768.

Smedema, L.K. and D.W. Rycroft. (1983). Land drainage-planning and design of

agricultural drainage systems, Cornell University Press, Ithica, N.Y.

Soil Conservation Research Program (SCRP), 2000. Area of Anjeni, Gojam, Ethiopia:

Long-term Monitoring of the Agricultural Environment 1984-1994. Soil

Erosion and Conservation Database. Centre for Development and

Environment, University of Berne.

Page 107: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

89

Srinivasan, R., T.S. Ramanarayanan, J.G. Arnold, and S.T. Bednarz, 1998. Large Area

Hydrologic Modeling and Assessment. Part II: Model Application. Journal of

American Water Resources Association (JAWRA) 34(1):91-101.

Steenhuis, T.S., M. Winchell, J. Rossing, J. Zollweg, and M.F. Walter. 1995. SCS run

off equation revisited for variable-source run off areas. J. Irrigation Drainage

Eng., 121(3): 234-8.

Tripathi, M.P., R.K. Panda, and N.S. Raghuwanshi. 2003. Identification and

Prioritisation of Critical Subwatersheds for Soil Conservation Management

using the SWAT Model. Biosystems Engineering (2003) 85(3):365-379,

doi:10.1016/S1537-5110(03)00066-7.

USAID, 2000. Amhara National Regional State food security research assessment

report. Available at: http://crsps.org/amhara/amhara_rpt.PDF

USDA-NRCS. 2004. Part 630: Hydrology. National Engineering Handbook. Available

at: http://policy.nrcs.usda.gov/media/pdf/H_210_630_9.pdf. Accessed 3

January 2008

USDA Soil Conservation Service. 1972. National Engineering Handbook Section 4

Hydrology, Chapters 4-10, 1972.

Walling D.A. (1984) The sediment yields of African Rivers In: Proceedings of the

Challenges in African hydrology and Water Resources. IAHS Publ. No. 144.

Werner C., 1986. Soil Conservation Experiment in the Anjeni Area. Gojam Research

Unit (Ethiopia). University of Berne, Switzerland: Soil Conservation Research

Project 13.

White E.D., Z.M. Easton, D.R. Fuka, E.S. Collick, E. Adgo, M. McCartney, S.B.

Awulachew, Y.G. Selassie, and T.S. Steenhuis. 2008. Adapting the soil and

water assessment tool (SWAT) for the Nile Basin (unofficial data describes

Page 108: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

90

that 70% of the cost of operation and maintenance in the Blue Nile part of

Sudan is spent on sediment related and canal maintenance).

White K. L. and Chaubey I., 2005. Sensitivity Analysis, Calibration and Validations

for a Multisite and Multivariable SWAT Model. Journal of the American

Water Resources Association (JAWRA), 41(5):1077-1089.

Wilcox, B. P., Rawls W. J., Brakensiek D. L., Wright J. R., 1990. Predicting Runoff

from Rangeland Catchments: A comparison of two models. Water Resource

Research, (26), 2401 – 2410.

Williams, J.R. 1975. Sediment-yield prediction with universal equation using runoff

energy fac-tor. p. 244-252. In Present and prospective technology for

predicting sediment yield and sources: Proceedings of the sediment yield

workshop, USDA Sedimentation Lab., Ox-ford, MS, November 28-30, 1972.

ARS-S-40.

Williams, J.R., 1995. Chapter 25: The EPIC model. In: V.P. Singh (eds.), Computer

models of Watershed hydrology, pp. 909-1000.

Williams, J.R., A.D. Nicks, and J.G. Arnold. 1985. Simulator for water resources in

rural basins Journal of Hydrology Engineering 111(6): 970-986.

Winchell M., R. Srinivasan R., Di Luzio M. and Arnold J. G., 2007. ArcSWAT

Interface for SWAT2005. Grassland, Soil & Water Research Laboratory,

USDA Agricultural Research Service, Temple, Texas.

Zeleke, Gete, 2000. Landscape Dynamics and Soil Erosion Process Modeling in the

North-western Ethiopian Highlands. African Studies Series A 16, Geographica

Bernensia, Berne.

Zeleke, Gete, 1998. Soil Map of the Anjeni research area. Soil Conservation Research

Program, Centre for Development and Environment, Berne, Switzerland.

Page 109: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

91

APPENDIX Appendix I: Parameters in SWAT database for each crops in the watershed

CPNM WWHT FRST RNGE RNGB CORN BARL ALFA SOYB FLAX TEFF CROP NAME

Winter Wheat

Forest-Mixed

Range-Grasses

Range-Brush Corn

Spring Barley Alfalfa Soybean Flax Teff

BIO_E 30 15 34 34 39 35 20 25 15 35 HVSTI 0.4 0.76 0.9 0.9 0.5 0.54 0.9 0.31 0.76 0.9 BLAI 4 5 2.5 2 3 4 4 3 5 4 FRGRW1 0.05 0.05 0.05 0.05 0.15 0.15 0.15 0.15 0.05 0.5 LAIMX1 0.05 0.05 0.1 0.1 0.05 0.01 0.01 0.05 0.05 0.02 FRGRW2 0.45 0.4 0.25 0.25 0.5 0.45 0.5 0.5 0.4 0.89 LAIMX2 0.95 0.95 0.7 0.7 0.95 0.95 0.95 0.95 0.95 0.95 DLAI 0.5 0.99 0.35 0.35 0.7 0.6 0.9 0.6 0.99 0.85 CHTMX 0.9 6 1 1 2.5 1.2 0.9 0.8 6 0.6 RDMX 1.3 3.5 2 2 2 1.3 3 1.7 3 1.3 T_OPT 18 30 25 25 25 25 20 25 30 25 T_BASE 0 10 12 12 8 0 4 10 10 6 CNYLD 0.025 0.0015 0.016 0.016 0.014 0.021 0.025 0.065 0 0.05 CPYLD 0.0022 0.0003 0.0022 0.0022 0.0016 0.0017 0.0035 0.0091 0 0.004 BN1 0.0663 0.006 0.02 0.02 0.047 0.059 0.0417 0.0524 0.01 0.03 BN2 0.0255 0.002 0.012 0.012 0.0177 0.0226 0.029 0.0265 0 0.02 BN3 0.0148 0.0015 0.005 0.005 0.0138 0.0131 0.02 0.0258 0 0.012 BP1 0.0053 0.0007 0.0014 0.0014 0.0048 0.0057 0.0035 0.0074 0 0.002 BP2 0.002 0.0004 0.001 0.001 0.0018 0.0022 0.0028 0.0037 0 0.0015 BP3 0.0012 0.0003 0.0007 0.0007 0.0014 0.0013 0.002 0.0035 0 0.0013 WSYF 0.2 0.01 0.9 0.9 0.3 0.2 0.9 0.01 0.01 0.9 USLE_C 0.15 0.05 0.01 0.05 0.15 0.1 0.15 0.2 0.05 0.3 GSI 0.006 0.002 0.005 0.005 0.007 0.008 0.01 0.007 0 0.008 VPDFR 4 4 4 4 4 4 4 4 4 4 FRGMAX 0.75 0.75 0.75 0.75 0.75 0.75 0.75 0.75 0.75 0.75 WAVP 6 8 10 10 7.2 7 10 8 8 8 CO2HI 660 660 660 660 660 660 660 660 660 660 BIOEHI 39 16 39 39 45 45 35 34 16 46 RSDCO_PL 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 OV_N 0.14 0.1 0.15 0.15 0.14 0.14 0.06 0.14 0.1 0.3 CN2A 62 36 49 39 67 62 31 67 36 65 CN2B 73 60 69 61 77 73 59 78 60 79 CN2C 81 73 79 74 83 81 72 85 73 84 CN2D 84 79 84 80 87 84 79 89 79 86 FERTFIELD 1 0 0 0 1 1 0 0 0 0 ALAI_MIN 0 0.75 0 0 0 0 0 0 0.75 0 BIO_LEAF 0 0.3 0 0 0 0 0 0 0.3 0 MAT_YRS 0 50 0 0 0 0 0 0 50 0 BMX_TREES 0 1000 0 0 0 0 0 0 1000 0 EXT_COEF 0.65 0.65 0.33 0.33 0.65 0.65 0.65 0.45 0.65 0

Page 110: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

92

Appendix II: Parameters in SWAT database for each soil layers in the watershed

SNAM AJLILE

AJVELU

AJEURE

AJHUAL

AJHANI

AJDYCA AJHAAL

AJHAAC AJHALX AJHUNI

NLAYERS 1 6 3 5 4 6 5 5 6 6 HYDGRP A B B C C B C B B B SOL_ZMX 500 1400 1300 1400 2000 1380 2150 1700 1850 1380 ANION_EXCL 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 SOL_CRK 0 0.01 0.01 0.03 0.01 0 0.01 0.01 0.01 0.002 TEXTURE SCL SCL SiL C CSiL L C L-CL SCL C-CL

SOL_Z1 200 200 250 200 200 200 200 200 200 200 SOL_BD1 1.1 1.45 1.08 1.1 1.1 1.1 1.1 1.1 1.45 1.1 SOL_AWC1 0.11 0.11 0.12 0.11 0.11 0.11 0.11 0.11 0.11 0.11

SOL_K1 5 5 6.8 1 4.34 7 4.34 4.34 7 7 SOL_CBN1 2 0.5 1.6 2 2 2 2 2 0.5 2

CLAY1 50 25 53.6 50 50 50 50 50 25 50

SILT1 33 31 25.7 33 33 33 33 33 31 33

SAND1 17 44 20.7 17 17 17 17 17 44 17

ROCK1 5 0.01 0 5 5 5 5 5 0.01 5 SOL_ALB1 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 USLE_K1 0.22 0.3 0.23 0.22 0.22 0.22 0.22 0.22 0.3 0.22 SOL_EC1 0 0 0.5 0 0 0 0 0 0 0

SOL_Z2 127 280 750 330 900 320 500 350 280 390 SOL_BD2 2.5 1.46 1.15 1.27 1.27 1.45 1.3 1.37 1.46 2.5 SOL_AWC2 0.5 0.11 0.19 0.11 0.11 0.13 0.13 0.09 0.11 0.1

SOL_K2 400 37.2 6.8 1 4.54 13 4.54 5.52 25 7 SOL_CBN2 0.58 0.52 0.3 0.8 1.5 1.1 1.4 0.22 0.52 0.58

CLAY2 5 28 73.6 10 23 32 63 44 22.6 35

SILT2 25 6 15.7 20 50 24 17 4 23 20

SAND2 70 66 10.7 70 27 44 20 52 54.4 45

ROCK2 98 0 0.01 0 0 0 0.01 2 0 0 SOL_ALB2 0.08 0.13 0.13 0.09 0.13 0 0.13 0.13 0.13 0.08 USLE_K2 0 0.3 0.22 0.2 0.22 0.34 0.22 0.11 0.18 0.14 SOL_EC2 0 0 0.045 0 0 0 0 0 0 0

SOL_Z3 0 360 1300 430 1000 650 900 680 450 420 SOL_BD3 0 1.45 1.17 1.28 1.28 1.45 1.3 1.42 1.45 1.5 SOL_AWC3 0 0.11 0.19 0.1 0.11 0.13 0.11 0.15 0.1 0.11

Page 111: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

93

SNAM AJLILE

AJVELU

AJEURE

AJHUAL

AJHANI

AJDYCA AJHAAL

AJHAAC AJHALX AJHUNI

SOL_K3 0 34.8 6.8 1 5.16 25 5.16 10.56 25 7 SOL_CBN3 0 0.63 0.1 0.4 1.3 1 1.1 0.21 0.63 1

CLAY3 0 52 71.6 15 60 30 61 43 41.6 24

SILT3 0 8 15.7 24 25 20 20 7 12 16

SAND3 0 40 12.7 61 15 50 19 50 46.4 60

ROCK3 0 0 0 0 0 0 0 1 0 0 SOL_ALB3 0 0.13 0.13 0.09 0.13 0.2 0.13 0.13 0.13 0.12 USLE_K3 0 0.3 0.2 0.2 0.22 0.2 0.22 0.28 0.18 0.14 SOL_EC3 0 0 0.06 0 0 0 0 0 0 0

SOL_Z4 0 710 0 770 2000 900 1600 900 720 710 SOL_BD4 0 1.49 0 1.22 1.22 1.39 1.3 1.49 1.49 1.3 SOL_AWC4 0 0.1 0 0.1 0.11 0.13 0.12 0.1 0.1 0.3

SOL_K4 0 33.6 0 1 4.24 25 4.24 33.6 25 7 SOL_CBN4 0 0.4 0 0.3 0.5 1 0.6 0.2 0.4 1

CLAY4 0 51 0 17 71 30 75 44 51.6 23.1

SILT4 0 5 0 26 20 20 20 6 10 16.5

SAND4 0 45 0 57 9 50 5 50 38.4 60.4

ROCK4 0 0 0 0 0 0 0 0 0 0 SOL_ALB4 0 0.13 0 0.09 0.13 0.2 0.13 0.13 0.13 0.2 USLE_K4 0 0.3 0 0.2 0.22 0.2 0.22 0.3 0.18 0.14 SOL_EC4 0 0 0 0 0 0 0 0 0 0

SOL_Z5 0 1120 0 1400 1000 1220 2150 1700 1470 1000 SOL_BD5 0 1.48 0 1.13 1.5 1.45 1.3 1.48 1.48 1.5 SOL_AWC5 0 0.1 0 0.1 0.4 0.1 0.14 0.1 0.1 0.4

SOL_K5 0 36 0 1 400 24 4.34 36 25 7 SOL_CBN5 0 0.2 0 0.2 2 1 0.4 0.2 0.2 2

CLAY5 0 38 0 25 23.1 25 79 43 54.3 23.1

SILT5 0 8 0 24 17.2 20 14 10 23.4 17.2

SAND5 0 54 0 51 59.7 55 7 47 22.3 59.7

ROCK5 0 0 0 0 0 0 0 0 0 0 SOL_ALB5 0 0.13 0 0.09 0.2 0.2 0.13 0.13 0.13 0.2 USLE_K5 0 0.3 0 0.2 0.14 0.2 0.22 0.3 0.18 0.14 SOL_EC5 0 0 0 0 0 0 0 0 0 0

SOL_Z6 0 1400 0 1635.2 1380 1380 1350 1350 1850 1380 SOL_BD6 0 1.49 0 1.1 1.4 1.53 1.49 1.49 1.49 1.4

Page 112: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

94

SNAM AJLILE

AJVELU

AJEURE

AJHUAL

AJHANI

AJDYCA AJHAAL

AJHAAC AJHALX AJHUNI

SOL_AWC6 0 0.1 0 0.11 0.5 0.12 0.2 0.2 0.1 0.5

SOL_K6 0 36 0 4.24 40 50 36 36 25 7 SOL_CBN6 0 0.12 0 1.24 0.9 1 0.12 0.12 0.12 0.9

CLAY6 0 25 0 60 23.6 22 17 17 54.3 24

SILT6 0 5 0 12.7 18.1 16 57 57 24.8 18

SAND6 0 70 0 27.3 58.3 62 26 26 20.9 58

ROCK6 0 0 0 0 0 14 0 0 0 0 SOL_ALB6 0 0.13 0 0.09 0.19 0.11 0.13 0.13 0.13 0.19 USLE_K6 0 0.3 0 0.2 0.14 0.2 0.3 0.3 0.18 0.14 SOL_EC6 0 0 0 0 0 0 0 0 0 0

SOL_Z7 0 1800 0 2422.4 0 0 1800 1800 1800 0 SOL_BD7 0 1.47 0 1.1 0 0 1.47 1.47 1.47 0 SOL_AWC7 0 0.21 0 0.09 0 0 0.21 0.21 0.21 0

SOL_K7 0 36 0 4.04 0 0 36 36 36 0 SOL_CBN7 0 0.1 0 0.34 0 0 0.1 0.1 0.1 0

CLAY7 0 16 0 63.6 0 0 16 16 16 0

SILT7 0 59 0 16.6 0 0 59 59 59 0

SAND7 0 25 0 19.8 0 0 25 25 25 0

ROCK7 0 0 0 0 0 0 0 0 0 0 SOL_ALB7 0 0.13 0 0.09 0 0 0.13 0.13 0.13 0 USLE_K7 0 0.3 0 0.2 0 0 0.3 0.3 0.3 0 SOL_EC7 0 0 0 0 0 0 0 0 0 0

Page 113: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

95

Appendix III: Parameters in SWAT database for Urban land uses in the watershed

URBNAME URLD

URBFLNM Residential-Low

Density FIMP 0.12 FCIMP 0.1 CURBDEN 0.24 URBCOEF 0.18 DIRTMX 225 THALF 0.75 TNCONC 460 TPCONC 196 TNO3CONC 6 OV_N 0.1 CN2A 31 CN2B 59 CN2C 72 CN2D 79 URBCN2 98

Page 114: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

96

Appendix IV: A. Sliding of sides of gullies

Page 115: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

97

Appendix IV: B. Soil piping in channel sides and springs in the watershed

Page 116: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

98

Appendix IV: C. Soil piping in gullies and side sliding

Piping

Page 117: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

99

Appendix V: Parameters used for Weather Generator in SWAT Model

TMPMX TMPMN TMPSTDMX TMPSTDMN PCPMM PCPSTD PCPSKW PR_W1_ PR_W2_ PCPD RAINHHMX SOLARAV DEWPT WNDAV

25.19 6.48 1.22 1.72 19.16 2.26 4.68 0.1 0.19 1.1 31.6 7.11 1.07 2.11 26.43 7.92 1.4 2.06 15.34 2.18 4.12 0.19 0.49 1 10.8 17.01 -2.72 2.45 26.47 9.65 1.66 1.99 49.855 3.92 3.1 0.21 0.52 4.9 15.6 20.46 2.03 2.36 26.06 10.83 2.1 1.77 64.72 4.36 3.1 0.32 0.69 5.9 18.4 16.06 5.49 2.2 25.11 11.28 2.23 1.56 110.455 6.21 2.29 0.73 0.92 11 30.8 7.68 9.31 2.02 21.76 10.82 2.12 1.9 316.35 10.04 1.2 0.66 0.97 24.6 29.8 1.63 9.28 1.73 19.54 10.65 1.5 1.11 429.12 11.82 1.47 0.65 0.96 29.1 35.3 4.48 12.39 1.5 19.46 10.66 1.6 1.76 392.845 10.18 1.19 0.55 0.9 28.85 36.5 10.4 12.33 1.62 21.03 9.85 1.45 1.3 258.665 7.95 1.25 0.35 0.7 24.25 35.9 15.7 10.61 1.86 22.28 8.89 1.28 1.78 153.93 7.37 2.38 0.16 0.46 12.5 38.4 15.19 8.45 2.15 23.56 7.32 1.13 2.8 36.295 2.68 3.38 0.1 0.28 3.8 13.1 9.78 5.88 2.18 24.29 6.01 1.1 1.52 18.46 1.86 3.78 0.08 0.26 1.6 10.5 2.36 12.22 2.11

Page 118: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

100

Appendix VI: Observed and Simulated Flow and Sediment loss in Calibration

Time Qobs Qsim SEDobs SEDsim (mm) (mm) (t/ha) (t/ha)

Jan-86 14.73 4.79 0.00 0 Feb-86 7.24 2.47 0.00 0 Mar-86 8.08 2.28 0.00 0 Apr-86 7.76 1.82 0.00 0 May-86 11.79 1.58 1.77 0 Jun-86 38.30 50.07 3.93 2.59 Jul-86 139.56 150.88 4.92 5.39

Aug-86 174.36 143.89 5.28 2.43 Sep-86 117.20 117.36 2.40 0.71 Oct-86 69.18 71.02 1.38 0.26 Nov-86 18.56 20.67 0.01 0 Dec-86 21.39 10.94 0.00 0 Jan-87 14.51 4.19 0.00 0 Feb-87 7.24 2.73 0.00 0 Mar-87 12.12 3.38 0.00 0 Apr-87 10.76 2.5 0.14 0 May-87 14.44 17.79 0.56 0.69 Jun-87 87.41 135.99 3.07 5.61 Jul-87 173.98 182.35 4.06 6.19

Aug-87 234.71 220.16 5.23 4.5 Sep-87 89.11 118.77 0.63 0.62 Oct-87 44.21 59.12 0.10 0.38 Nov-87 22.00 15.56 0.00 0 Dec-87 14.73 9.41 0.00 0 Jan-88 14.73 4.22 0.00 0 Feb-88 13.78 2.83 0.00 0 Mar-88 9.32 1.72 0.00 0 Apr-88 7.76 1.04 0.00 0 May-88 8.02 2.33 0.00 0 Jun-88 21.45 80.34 0.52 4.28 Jul-88 257.53 169.37 2.09 5.7

Aug-88 192.36 213.1 1.41 4.56 Sep-88 89.67 112.68 0.66 0.47 Oct-88 90.45 118.3 0.74 1.37 Nov-88 28.14 34.34 0.00 0 Dec-88 22.20 16.22 0.00 0

Page 119: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

101

Time Qobs Qsim SEDobs SEDsim (mm) (mm) (t/ha) (t/ha)

Jan-89 14.73 8.17 0.00 0.02 Feb-89 7.24 3.18 0.00 0 Mar-89 11.36 3.03 0.00 0 Apr-89 13.56 3.18 0.00 0 May-89 12.05 7.15 0.00 0.06 Jun-89 9.29 35.29 0.02 0.83 Jul-89 153.76 215.98 2.45 8.68

Aug-89 161.06 163.82 1.54 2.95 Sep-89 131.07 137.73 1.01 0.91 Oct-89 42.44 63.64 0.00 0.08 Nov-89 22.93 20.75 0.00 0 Dec-89 18.78 9.71 0.07 0 Jan-90 22.68 6.18 0.00 0.01 Feb-90 11.57 2.69 0.00 0 Mar-90 8.02 2.01 0.00 0 Apr-90 7.76 1.49 0.00 0 May-90 10.48 9.54 0.39 0.17 Jun-90 12.34 29.22 0.57 1.29 Jul-90 135.13 165.7 7.93 6.24

Aug-90 266.41 218.87 16.08 4.94 Sep-90 162.03 170.42 5.08 2.75 Oct-90 47.89 80.62 0.00 0.11 Nov-90 21.94 25.13 0.00 0 Dec-90 17.55 11.38 0.00 0 Jan-91 11.70 4.63 0.00 0 Feb-91 8.97 2.4 0.00 0 Mar-91 8.02 2.14 0.00 0 Apr-91 7.76 1.42 0.00 0 May-91 8.83 2.11 0.03 0 Jun-91 12.35 53.62 0.00 3.99 Jul-91 176.13 218.94 9.88 8.56

Aug-91 187.58 192.54 6.62 3.68 Sep-91 132.73 146.64 3.75 1.48 Oct-91 37.85 51.26 0.06 0.04 Nov-91 21.51 15.85 0.01 0 Dec-91 14.73 7.65 0.00 0 Jan-92 14.73 3.66 0.00 0 Feb-92 12.48 2.02 0.00 0

Page 120: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

102

Time Qobs Qsim SEDobs SEDsim (mm) (mm) (t/ha) (t/ha)

Mar-92 8.45 1.73 0.00 0 Apr-92 13.14 5.41 0.94 0.06 May-92 9.89 13.99 0.22 0.3 Jun-92 36.82 64.8 3.67 2.44 Jul-92 117.66 131.1 4.14 2.93

Aug-92 220.39 160.86 2.30 2.96 Sep-92 105.18 98.14 1.33 0.4 Oct-92 109.64 103.35 1.66 0.61 Nov-92 27.24 35.31 0.00 0 Dec-92 18.57 19.66 0.00 0.01 Jan-93 14.73 5.76 0.00 0 Feb-93 11.79 2.96 0.00 0 Mar-93 8.02 2.26 0.00 0 Apr-93 8.02 2.37 0.00 0 May-93 8.57 1.81 0.00 0 Jun-93 57.79 90.3 4.25 5.02 Jul-93 170.70 169.38 7.57 5.21

Aug-93 247.16 241.83 6.82 5.75 Sep-93 218.40 203.13 10.70 4.37 Oct-93 66.26 104.89 0.50 0.31 Nov-93 25.43 38.21 0.00 0 Dec-93 14.73 16.52 0.00 0

Page 121: SWAT TO IDENTIFY WATERSHED MANAGEMENT OPTIONSsoilandwater.bee.cornell.edu/Research/international/docs/Biniam_Thesis... · SWAT-WB model simulates saturation excess flow in order to

103

Appendix VII: Observed and Simulated Flow and Sediment loss in Validation

Time Qobs Qsim Sed obs Sed sim (mm) (mm) (ton/ha) (ton/ha)

Jan-95 14.72 3.98 0.00 0 Feb-95 10.48 2.24 0.00 0 Mar-95 8.01 1.66 0.00 0 Apr-95 8.38 1.9 0.11 0 May-95 12.55 5.71 0.45 0.01 Jun-95 51.03 80.84 8.69 5.23 Jul-95 175.54 187.13 6.69 6.1

Aug-95 170.69 177.73 5.28 2.69 Sep-95 138.45 166.73 5.08 2.56 Oct-95 46.38 71.27 0.00 0.09 Nov-95 18.87 24.32 0.00 0 Dec-95 18.25 10.84 0.00 0 Jan-96 11.99 5.2 0.00 0 Feb-96 10.19 2.77 0.00 0 Mar-96 24.58 7.4 3.65 0.07 Apr-96 24.26 11.57 0.51 0.39 May-96 49.86 36.47 2.77 1.41 Jun-96 100.26 120.44 4.17 4.09 Jul-96 222.31 237.27 10.58 9.52

Aug-96 151.75 170.28 4.29 2.29 Sep-96 89.94 108.83 1.46 0.44 Oct-96 44.82 47.51 0.18 0.04 Nov-96 22.20 18.47 0.00 0.01 Dec-96 15.01 7.89 0.00 0 Jan-00 19.07 3.81 0.00 0 Feb-00 13.77 2.09 0.00 0 Mar-00 14.72 1.47 0.00 0 Apr-00 10.78 2.67 0.00 0 May-00 10.39 3.59 0.00 0 Jun-00 80.05 104.16 7.81 5.7 Jul-00 184.33 218.5 8.03 8.9

Aug-00 156.04 172.36 4.80 2.73 Sep-00 87.74 100 1.75 0.43 Oct-00 76.73 118.33 0.00 1.46 Nov-00 57.11 43.11 0.00 0 Dec-00 33.59 20.74 0.00 0