CHAPTER 6: Decadal Erosion Rate Controls - Introduction 118m.rezaeian/thesis/eroison...

CHAPTER 6: Decadal Erosion Rate Controls - Introduction

118

6-1 Motivation

Erosion interacts with tectonic and climatic processes to shape the topography of active mountain

belts. Compressional orogens with the highest rates of rock uplift have the highest rates of denudation. Their

uplift rates are controlled by erosion through an isostatic response mechanism. This interplay of tectonic

uplift and erosion is affected by local climate, vegetation, steepness of the topography and erodibility of the

rock mass (e.g., Koons, 1987; Pinter & Brandon, 1997; Willett, 1999). Of these, climate is often thought to be

the dominant variable, while vegetation and substrate strength are commonly assumed to be of secondary

importance, and held constant in geodynamic models. However, due to the influence of vegetation as a

function of precipitation, the relationship between rainfall and erosion is nonlinear and complex.

Decadal average erosion rates are known for many catchments in the Alborz Mountains (Chapter 5).

Erosion rates vary across the mountain belt, but not in the simple way expected from consideration of the

pattern of precipitation. The north flank of the mountain belt has annual precipitation totals far outstripping

those on the south flank, but erosion rates peak in the south. It is evident that other factors control the pattern

of erosion in the Alborz. In this chapter, these controls will be investigated. Specifically, I have quantified

precipitation, runoff, stream power, vegetation cover, slope, and relief, and substrate properties, and

normalized these factors where possible. The complex interactions between these potential controls make it

difficult to quantify the erosional effects of individual factors with precision; therefore, non-linear

correlations with thresholds are characteristic of decadal erosion and its controls. Using dimensionless

normalized values, makes it possible to compare different entities statistically and investigate the interaction

between them. Special attention has been paid to the seasonality of precipitation and runoff which reflects not

only the erosive impact of rainstorms in wet seasons, but also the attenuating effects of seasonally changing

vegetation (e.g., Douglas, 1967).

Seismicity has not been considered in detail. Although recent, large earthquakes have caused

significant mass wasting in the Alborz Mountains, insufficient data is at hand to evaluate the relation between

erosion and seismicity quantitatively.

Rezaeian M., 2008, Coupled tectonics, erosion and climate in the Alborz Mountains, Iran. PhD thesis, University of Cambridge; 219 p.


CHAPTER 6: Decadal Erosion Rate Controls-Statistical Approaches

119

6-2 Statistical Approaches

6-2-1 Introduction

Data on erosion and potential controls comes in different formats and resolutions. Erosion rates

have been calculated from at-a-station hydrometric measurements, resulting in catchment-wide average

values. Similarly, meteorological data are for specific stations, but these stations are different in number

and location from the nodes in the hydrometric network. In contrast, topographic data, and proxies for

vegetation density are remote sensed, with spatially continuous and uniform coverage. These variables

can be assessed within individual cells of a grid covering the entire mountain belt. Geomechanical

properties, on the other hand, are inferred from geological maps, without a robust calibration. These

properties are grouped in broad classes, and can not be quantified on a graded scale.

This diversity of data format and density makes a direct comparison between variables difficult.

The first challenge, therefore, is to homogenise the data. This can be done by compounding all available

data into representative values for geographic units. The unit of choice is the catchment, specifically the

drainage basins with a hydrometric station. Alternatively homogenisation can be achieved by

extrapolation of at-a-station data to a full grid of estimated values. Both approaches have limitations and

advantages. In this study, I have combined them to obtain complementary results.

Any analysis of multiple, inter-related variables in a large domain is likely to be hostage to spatial

complexity and the interference of individual effects. This limits the degree of certainty with which

individual controls and responses can be identified and quantified, and has caused a large part of observed

variance in erosion rates in other studies to remain unexplained (cf., Dadson et al., 2003). By segregating

data into geographic domains with a reduced heterogeneity of selected variables, for example lithology or

vegetation, it is possible to isolate the effects of other variables on erosion. I have studied erosion and its

forcing on the scale of the integral Alborz Mountains, but I have also split available data into a northern

and a southern domain, defined as the north and south flank of the mountain belt, respectively, in the hope

that this would result in a better resolution of the mechanisms of erosion.

It is clear from results presented in Chapter 5 that erosion of the Alborz Mountains is highly

seasonal. It does, therefore, make sense to look not only at annual average values of variables, but also at

seasonal variability. Deeper levels of temporal complexity, rich as they may be, have not been tapped in

this study.

6-2-2 Homogenisation

The process of translating or extrapolating information from fine to coarse scales is usually referred to as

up-scaling, and the reverse process is down-scaling. Among four general methods for the rescaling of data

as identified by King (1991), I have applied Lumping to compound girded data into characteristic values

for geographic units, i.e., up-scaling, and Direct Extrapolation to obtain girded values from at-a-station

data, i.e. down-scaling.

Lumping is the simplest method for up-scaling by which coarse-scale mean values are derived from

averaging fine-scale, often pixelated variables, such as vegetation density, slope, relief, and precipitation.

Lumping assumes that the mathematical formulation of processes in fine-scale models remains valid at


CHAPTER 6: Decadal Erosion Rate Controls-Statistical Approaches

120

coarser scales, or that larger scale systems behave in the same, or a similar way as the average fine-scale

system. This assumption holds only if the equations that describe the system are linear. Lumping is

known to lead to considerable bias, because it doesn’t account for temporal or spatial variability in

processes and ignores non-linear changes with scale (Rastetter et al., 1992).

In this study, the geographic unit for lumping is the catchment. This choice is dictated by the nature of the

data on erosion and sediment transport, which has been collected at hydrometric stations. Therefore,

catchment-wide average values of all other variables have been calculated for all gauged catchments.

Lumped data can be analysed with or without regard for the size of the catchment. It can be argued that

the weight of a data point should be determined by the size of the area to which it applies. The simplest

weighting scheme therefore uses a direct proportionality of data weight to catchment size. It has been

applied in this study. In recognition of the fact that this is not necessarily the most robust weighting

strategy, and in the absence of an established protocol, catchment data have also been considered in un-

weighted format.

Lumping methods are notorious for their suppression of spatial heterogeneity and its effect on response

variables (Turner, 1989). These methods do not effectively exploit the potential of the data sets with the

finest spatial scales in a larger ensemble. To do this, coarser data sets must be artificially refined. Direct

extrapolation is the simplest way of achieving this. In this method, the catchment-wide average value of a

variable, as assessed from at-a-station measurements, is assigned to each cell of a geographic grid that is

located entirely within that catchment. Cells straddling one or more catchment boundaries are assigned a

value equivalent to the average of catchment values weighted for the relative extent of the different

catchments within the cell. In this way, a full grid of values is obtained. More sophisticated extrapolation

schemes would apply smoothing strategies. In the absence of an established protocol, I have applied a

direct extrapolation strategy in this study.

6-2-3 r-Squared Value

I have investigated the forcing and mechanisms of erosion in the Alborz Mountains by constraining the

relationships between erosion and individual forcing factors. The strength of these relations has been

evaluated by means of the r-squared measure.

r-squared is a statistical measure of how well a regression line approximates real data points. It is

a descriptive measure with values between zero and one, indicating how good one term is at predicting

another. An r-squared of 1.0 (100%) indicates a perfect fit. The formula for Pearson’s product-moment

correlation coefficient, r is:

r(X,Y) = [ Cov (X,Y) ] / [ StdDev (X) . StdDev (Y) ] (6.2.1)

where X and Y are two independent variables, Cov is the covariance of these variables, and StdDev is the

standard deviation of a given variable. r-squared values have been calculated for linear and non-linear

best fits to the data, and are given for each pairing.


CHAPTER 6: Decadal Erosion Rate Controls - Precipitation

121

6-3 Precipitation

6-3-1 Introduction

Precipitation is involved in many different erosion mechanisms. Rain splash may detach loose

particles from Earth’s surface. Soil detachment and transport by rain splash is usually the first step in soil loss

and sediment transport. Directionality of drop impacts can give rise to a net displacement of mass, and

rainfall on hill slopes drives a down slope flux of sediment because ballistic path length is greatest in that

direction. Rainfall is partitioned into infiltration and runoff. Infiltration feeds seepage flow, which may cause

formation of erosional pipes in substrates. It also sets up gradients in pore water pressure, increases the

weight of permeable substrate and decreases its cohesion. Together, these effects can cause slope failure, and

down slope displacement of large volumes of sediments.

Water that doesn’t infiltrate accumulates at the surface, forming runoff on hill slopes. Once depth and

velocity of flow are sufficient to overcome to the strength of the substrate, erosion occurs. Runoff

concentration in channels enables rivers to carry sediment, and to erode valley deposits and bedrock.

Precipitation may fall in the form of snow. Although snow fall in itself is not erosive, it can cause eroding

avalanches, and spring melt of substantial snow packs may create erosive river discharges. A portion of

rainfall evaporates, and vegetation may act to intercept raindrops, transferring water to the surface by gentle

stem flow. These mechanisms can complicate simple relationship between rainfall, runoff and erosion.

The spatial and temporal distribution of precipitation in mountain belts is influenced by: i) The

synoptic weather systems, and ii) Orography. A brief summary of orography is given, followed by a summary

of the synoptic weather system of north Iran.

Mountain ranges have distinct orographic precipitation patterns. These are thought to control patterns

of erosion and rock exhumation on geological time scales, and can have profound effects to shape the

mountain ranges and their tectonic regime through asymmetric erosion (e.g., Beaumont et al. 1992, Willett,

1999, Reiners et al. 2003; Roe, 2005).Moist air rising over a topographic barrier expands, causing its capacity

to retain moisture to decrease. Simultaneously, the air cools and water vapour condensates, forming droplets.

Together, these two mechanisms drive precipitation on mountain flanks facing into predominant air streams.

The distribution of orographic precipitation is set by the steepness of the topographic front and the velocity of

atmospheric flow across it. Often, rising air has dried before it reaches the orogen ridge pole, and orographic

precipitation maxima may be located well away from the topographic divide. Across the divide, the air mass

can fall with reverse orographic effects, and a rain shadow results (e.g., Browning & Hill, 1981; Roe, 2005).

Air can also flow around a topographic obstacle. Blocking of atmospheric flow leads to range-

parallel flow towards areas with lower topography, typically at the tips of mountain ranges (e.g., Medina &

Houze, 2003). In such cases, precipitation does not penetrate deep into mountain belts, but precipitation rates

may be high at the leading edge of the mountain topography. Flow of moist air along a mountain front causes

advection of water vapour into certain corridors, where precipitation rates may be untypically high as a result.



122

6-3-2 Climate Synopsis

The major mountain ranges of the Alborz and Zagros play an influential role in temporal and spatial

distribution of precipitation across the Iranian plateau. Only limited amounts of moist air penetrate Iran from

different sources and water vapour flux is restricted by the geographic distribution of nearby water masses.

Precipitation mainly originates from the Mediterranean Sea, Black Sea, and Caspian Sea. The Red Sea and

Persian Gulf-Oman Sea provide moisture to the air masses entering the country from the south west (e.g.,

Khalili, 1984; Alijani & Harman, 1985; Nazemosadat & Cordery, 2000).

The average annual precipitation on the Iranian plateau is around 250 mm, giving rise to an arid to

semi-arid climate in Iran, except in mountain areas exposed to moist air masses. The Caspian Sea is a major

source of moisture for north Iran and the Caspian lowlands are the wettest part of Iran with annual

precipitation in excess of 1000mm. In this region, the precipitation is a direct function of evaporation in the

Caspian basin, which is most intense in summer. As a result, summers are wet on the Caspian coast, and the

adjacent highlands.

The mountain slopes of the northwest Zagros and the southwest Alborz receive a considerable

amount of orographic precipitation (Alijani & Harman, 1985; Nazemosadat & Cordery, 2000). It is associated

with an eastward propagating, mid-latitude cyclonic system from the Mediterranean region, which produces

heavy winter precipitation across much of Iran (Martyn, 1992). This system accounts for up to half of the

total annual precipitation in Iran (Alijani et al., 2007). Most of the incoming water vapour is intercepted by

the high mountain ranges of the region, while the interior high plains receive only a small amount of

precipitation. In winter, precipitation falls as snow at higher elevations (Syed et al., 2006).

In summer, Mediterranean cyclonic systems affect only the north-western, north-eastern, and Caspian

regions (Alijani & Harman, 1985). Due to the southward shift of the global pressure belts in winter in the

northern hemisphere (Barth & Steinkohl, 2004); they affect the whole country in winter. Spring is a transition

from wet winter to dry summer (Alijani & Harman, 1985).

Annual rainfall variability is high in the arid and semi-arid regions, with values of the coefficient of

variation of precipitation ranging from 18% in the north-west of the Alborz to 50% in south-east of the

mountains (Dinpashoh et al., 2004; Modarres & Silva, 2007).

6-3-3 Climate of the Alborz Mountains

6-3-3-1 Direction of Air Flow

The rainfall in the Caspian lowland has long been thought to be due to orographic lifting of unstable

air arriving from the Caspian Sea (Ganji, 1968). However, Khalili (1973) explained the south Caspian

precipitation pattern based on a cyclonic mechanism involving air moving from a high pressure center over

Siberia. A branch of the air flow crosses the Caspian Sea from NE to SW and precipitation occurs as a result

of thermodynamic destabilization of the Siberian cold air as it crosses the warmer sea surface.



123

The dominant direction of airflow can be seen from wind direction data catalogued by the NCEP

(National Centers for Environmental Prediction), and published by NOAA’s climate program office. This

catalogue contains monthly average wind directions for 2.5 x 2.5 degree grid cells, calculated based on data

for 1970-2001.

In the southern Alborz, autumn and winter winds are to the NE, while spring and summer winds are

to the SW. In the northern Alborz and the Caspian coast, wind directions are more variable. In summer, and

to a degree in autumn, winds are from the NE; winter winds are from the E, and in spring from the N. Air

flow from the Caspian basin is blocked by high topography of the central Alborz, and water vapour is

advected westward along the mountain front during much of the year. As the topographic barrier curves

around the SW tip of the Caspian basin, it occludes the lateral air flow and forces high precipitation rates.

Given the orientation and curvature of Alborz Mountains around the South Caspian Basin, the

dominant northeast - southwest air flow produces a main flow parallel to the east Alborz and perpendicular to

the mountain belt in the west.

6-3-3-2 Spatial Distribution

The Alborz Mountains are the major highland of northern Iran. They are a boundary between the

coastal plains of the Caspian region, with humid or sub-humid climate, and the Central Iranian Plateau, with

arid or semi-arid climate. This produces a sharp gradient in precipitation across the mountain range. I have

used daily precipitation data from 411 meteorological stations operated by the Ministry of Energy of Iran

measured over a period of 2-33 years (TAMAB, unpublished data), and monthly precipitation totals from 466

stations operated by the Iranian Meteorological Organization (IRIMET, http://www.irimet.net/) over a period

of 2-40 years (Fig. 6.3.1), to determine the spatial pattern of annual precipitation across the Alborz

Mountains.

Fig. 6.3.1: Total annual precipitation in the Alborz, derived from 877 meteorological stations.

Locations of weather stations operated by TAMAB and IRIMET in northern Iran are shown as white circles.



124

Figure 6.3.1 shows the spatial pattern of the total annual precipitation. It ranges from 85 mm in the

southeastern Alborz up to 1800 mm in the western Caspian plain, and precipitation rates decrease from west

to east and from north to south.

The west-southwest and east-southeast sectors of the mountain belt have least annual precipitation

(<300 mm) while the south-central Alborz near northern Tehran and Karaj and the highlands of the main

divide have higher precipitation totals (600 mm/yr) than adjacent sectors in the south flank of the mountain

belt. Low precipitation totals are not limited to the southern Alborz: some areas such as the Haraz valley to

the north of Damavand (Fig. 3.1 for location) receive <400mm/yr. Thus, the precipitation pattern of the

Alborz reflects the mode of air flow in the region.

6-3-3-3 Precipitation Cross Sections

Precipitation gradients across the mountain belt have been investigated along nine N-S cross sections:

Talesh, Sefid Rud, Shah Rud, Taleqan, Chalus, Tehran, Haraz, Firuz Kuh (Talar), Neka (Fig. 6.3.2 & 6.3.3).

AFT data along most of these sections have been reviewed in Chapter 2 (Section 2-2-15-5-3).

Fig. 6.3.2: Location map of cross sections overlaid on the total annual precipitation map of the Alborz.

The distribution of annual precipitation totals across the mountain belt has some common patterns, as

well as along-strike differences. In the south flank of the mountain belt, precipitation rates are lowest (<200

mm/yr) at the mountain front and in the adjacent interior. Throughout the south, precipitation rates increase

steadily with elevation of the topography in a strong orographic pattern. Precipitation rates peak on high

ridges, and drop dramatically beyond these barriers in intramontane basins and longitudinal valleys. The

highest precipitation rates in the south flank are found in the central Chalus, Tehran and Haraz sections,

where the main divide is highest. Away from the highest topography, orography of the southern Alborz is



125

Chalus

200

400

600

800

1000

1200

Ann

ua

l pre

cip

itatio

n (m

m)

0 20 40 60 80 100 120

0

1000

2000

3000

4000

Ele

va

tion

(m

)

Kilometer south

Talesh

0 10 20 30 40 50 60 70

Kilometer west

200

400

600

800

1000

1200

An

nu

al p

recip

itatio

n (m

m)

0

500

1000

1500

2000

2500

Ele

va

tion

(m

)

Sefid Rud

200

400

600

800

1000

1200

1400

1600

1800

2000

Ann

ua

l pre

cip

itatio

n (m

m)

0 20 40 60 80 100 120 140

0

500

1000

1500

2000

2500

3000

Ele

vation

(m

)

Kilometer south

Tehran

Kilometer south

200

400

600

800

1000

1200

Ann

ua

l pre

cip

itatio

n (m

m)

0 20 40 60 80 100 120

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Ele

va

tio

n (

m)

weaker. Although individual ridges are associated with elevated precipitation rates, a general, increasing

trend at the scale of the mountain belt is less pronounced in the peripheral areas.

In northern flank of the mountain belt, precipitation rates increase with elevation only in the Sefid

Rud, Shah Rud and Taleqan sections of the western Alborz. Further east, precipitation rates are highest in the

coastal plain, decreasing steadily towards the mountain belt interior.

However, where sufficient data are available for the high mountains, precipitation rates appear to increase

again on the northern rise to the main divide.

In north flank of the Alborz, total annual precipitation is anti-correlated with altitude; it is the highest

in the coastal plain and decreases with altitude with a gradient of 22-68 mm of rain per 100 meters (e.g.,

Khalili, 1973). This pattern is observed along the Chalus, Tehran and Haraz sections.

This general pattern is violated where the coastal plain is wide. Two main epicenters of high

precipitation are observed around the Sefid Rud and Talar-Tajan alluvial fans far from the coast in the west

and central-eastern part of coastal plain, respectively. Both are located upwind from major barriers (Medina

& Houze, 2003), and an ascending trend in the precipitation-altitude relation is observed in these sections.

The precipitation gradient is 24-28 mm per 100 m, with a zone of maximum precipitation at around 2000 to

2400 m (e.g., Khalili, 1973). This pattern is considered as a uniform precipitation enhancement by altitude.

Taleqan

200

400

600

800

1000

1200

1400

Ann

ua

l pre

cip

itatio

n (m

m)

0 20 40 60 80 100 120 140

0

1000

2000

3000

4000

Ele

va

tion

(m

)

Kilometer south

Shah Rud

0

200

400

600

800

1000

1200

1400

1600

An

nu

al p

recip

itatio

n (m

m)

0 20 40 60 80 100 120 140

0

500

1000

1500

2000

2500

3000

3500

4000

Ele

va

tion

(m

)

Kilometer south



126

Haraz

200

400

600

800

1000

Ann

ual p

recip

itatio

n (m

m)

0 20 40 60 80 100 120 140 160

0

1000

2000

3000

4000

5000

6000

Ele

vatio

n (

m)

Kilometer south

Fig. 6.3.3: Topography and annual precipitation along nine cross sections (for location see Fig.

6.3.2). Maximum, minimum, and mean elevation within a 20-km-wide swath oriented perpendicular to the

strike of the Alborz Mountains are shown in grey. Average annual precipitation shown in blue. Error bars

show 2σ interval .

6-3-3-4 Transverse and Longitudinal Valleys

Transverse valleys are conspicuous as they cut across tectonically controlled geological structures

forming pronounced gorges or canyons through prominent topographic barriers (e.g., Alvarez, 1999). In

Figure 6.3.1, high rainfall totals along the main transverse valleys of Chalus, Haraz and Talar illustrate the

role of deep trunk valleys, admitting water vapour from the north deep into the mountain interior. Transverse

valleys act as conduits for wet front transport in mountain belts, and as a result, the have greater precipitation

totals than the surrounding areas (e.g., Thiede et al., 2004).

The Haraz valley forms a deep gorge with high relief (4000 m at Mount Damavand) in northern slope

of the Alborz. Its head waters are located close to southern mountain front, and the catchment occupies the

widest segment of the northern slope in the mountain belt. In its middle reaches, the Haraz River turns around

a major topographic barrier, Mount Damavand. This is where much of the water vapour pouring into the

mountain belt through this corridor, rains out. The precipitation profile along the Haraz valley (Fig. 6.3.3)

shows a high amount of precipitation (1000mm) in the coastal plain, decreasing into the mountain belt with a

minimum of 270 mm in the Baladeh longitudinal valley, and then rising again to 656 mm in the northern

slope of Mount Damavand. Damavand is too high to have high precipitation rates at its peak, as can be seen

from the drop in precipitation around the main divide in this section.

Longitudinal valleys define first-order topographic lows inside mountain belts. They tend to sit behind frontal

ridges that parallel the gross structural grain (Koons, 1995; Jamieson et al., 2004). These ridges constitute

Neka

0 20 40 60 80 100 120

0

500

1000

1500

2000

2500

3000

3500

Ele

va

tion

(m

)

Kilometer south-east

0

100

200

300

400

500

600

An

nu

al p

recip

itatio

n (m

m)

0

200

400

600

800

1000 An

nua

l pre

cip

itatio

n (m

m)

0 20 40 60 80 100 120 140 160

0

1000

2000

3000

4000

Ele

va

tion

(m

)

Kilometer south

Firuz Kuh



127

200 400 600 800 1000 1200 1400 1600

0

100

200

300

400

500

600

700

800

900

Spring

Summer

Autumn

Winter

To

tal

se

as

on

al

pre

cip

ita

tio

n (

mm

)

Total annual precipitation (mm)

barriers to uprising air flow and generate rain shadows inside the mountain belt. Precipitation is suppressed

along interior valleys due to adiabatic heating and compression of descending air, but upslope enhancement

of the precipitation occurs over the windward valley slopes toward divide. The longitudinal valleys of Shah

Rud in the south-west and Baladeh in the northern slope of the mountain belt clearly show this precipitation

pattern. For the Shah Rud valley the effect is visible in three sections: Sefid Rud, Shah Rud and Taleqan

(marked on Fig. 6.3.3).

6-3-4 Seasonality of Precipitation

Daily precipitation data from the Ministry of Energy of Iran have been aggregated in average total

seasonal precipitation rates, across the Alborz. Winters have average precipitation of 171±65 mm, but spring

and summer, with average precipitation totals of 165±100 and 158±250 mm, respectively, are only

marginally drier. Autumn is considerably drier throughout the mountain belt with 74±70 mm rainfall.

However, these mountain belt averages hide significant shifts in the location of precipitation, borne out most

clearly in the large range of the summer averages.

To explore the seasonality in more detail, total seasonal precipitation was plotted against total annual

precipitation for 411 weather stations in the Alborz Mountains (Fig. 6.3.4).

Fig. 6.3.4: Seasonal precipitation and annual precipitation, for 411 gauging stations, operated by

TAMAB. Bars show 2σ interval within binned annual precipitation.

Throughout the Alborz, autumn is the driest season, but the wettest season changes with the total

annual precipitation. In areas with relatively low annual precipitation totals, winter is the wettest season, with

significant additional rainfall in spring, while summers are very dry.

Across the mountain belt winter precipitation varies least, and in wetter areas, total annual

precipitation is increasingly dominated by summer rainfall. The cross-over from a winter/spring dominated

precipitation regime to a summer-dominated regime is at about 800 mm of precipitation per year.



128

0 200 400 600 800 1000 1200 1400 1600 1800 2000

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Spring

0 200 400 600 800 1000 1200 1400 1600 1800 2000

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Summer

0 200 400 600 800 1000 1200 1400 1600 1800 2000

0.0

0.1

0.2

0.3

0.4

0.5

0.6 Autumn

To

tal se

aso

na

l p

recip

ita

tio

n n

orm

alize

d b

y t

ota

l a

nn

ua

l p

recip

ita

tio

n


0 200 400 600 800 1000 1200 1400 1600 1800 2000

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Winter

The trends are best illustrated by normalizing seasonal precipitation rates by the annual total

precipitation at a station (Fig. 6.3.5). Where annual total precipitation >800 mm, summer precipitation is 30-

50% of the total. Winter precipitation is ~40-65% of the annual total in drier areas where annual total

precipitation < 800mm. Spring proportions decrease slightly with increasing annual precipitation, while

autumn contribution increases marginally.

Fig. 6.3.5: Seasonal proportion of total annual precipitation in 411 meteorological gauging stations in

the Alborz.

Thus there is a sharp shift in seasonality between the dry (< 800 mm) and wet part of the mountain

belt (> 800 mm). The spatial pattern of wet seasons is illustrated in Figure 6.3.6. In the north flank of the

Alborz, summer is the wettest season with 30-50 % of the annual rain total. This seasonal bias is strongest in

the west of the mountain belt, and decreases to the east (Fig. 6.3.6a). The southern Alborz has a semi-

Mediterranean precipitation regime with dry summers and autumns, and precipitation peaking in winter as

snow and spring as rain (e.g., Khalili, 1973); winter precipitation makes up approximately 40-65 % of the

annual total in the areas where the precipitation is less than 800 mm per year (Fig. 6.3.6b).

The relative variability (average of the absolute deviations of the values above from their

mean) of seasonal precipitation proportions evolves with the annual total precipitation (Figure 6.3.7). When

normalized by the annual total precipitation, the variability of seasonal precipitation is greatest where the

annual total precipitation is smallest. The variability decreases steadily to a minimum of 0.07 at an annual

total precipitation of ~800 mm, and is approximately constant at that value in wetter areas.



129

0 200 400 600 800 1000 1200 1400

0.05

0.10

0.15

0.20

Vari

ab

ilit

y o

f n

orm

alized

seaso

nal p

recip

itati

on


Fig. 6.3.6: Proportion of annual precipitation during summer (a) and winter (b) in the Alborz,

estimated from daily precipitation data from 411 stations operated by TAMAB.

Fig. 6.3.7: Variability of normalized seasonal precipitation plotted against total annual precipitation,

for 411 meteorological gauging stations in the Alborz. Bars show 2σ interval within precipitations bins. (A

polynomial best fit to the data: y = 1E-07x2 - 0.0002x + 0.16 has R

2 = 0.19, and Pearson’s correlation

coefficient = -0.40).

(a)

(b)



130

The spatial distribution of the seasonal normalized precipitation is shown in Figure 6.3.8, in which a

strong contrast is evident between north and south flank. The northern Alborz has low variability, decreasing

trend from west to east; the southern Alborz has high variability, especially in central-west regions. This

pattern is a function of the synoptic climate discussed in section 6.3.2. In the next section I explore the link

between precipitation and erosion.

Fig. 6.3.8: Variability of seasonal precipitation normalized by annual precipitation in the Alborz,

estimated from daily precipitation data from 411 stations operated by TAMAB.

6-3-5 Precipitation and Erosion

6-3-5-1 Annual Total Precipitation and Erosion

Three statistical approaches, lumping without weighting, lumping with weighting and direct

extrapolation have been applied to constrain the correlation between total annual and seasonal precipitation,

and erosion in the Alborz Mountains.

Lumped un-weighted total annual precipitation for catchments (Fig. 6.3.9) are weakly positively

correlated with catchment wide annual erosion rate with R2= 0.01, Pearson’s correlation coefficient = 0.1.

However, when the data is weighted according to catchment area, a weak anti-correlation found with

R2= 0.21; Pearson’s correlation coefficient = -0.25 (Fig. 6.3.10). This result is dominated by larger

catchments.

The direct extrapolation method has been applied to account for spatial heterogeneity within

catchments. For this purpose, the Alborz domain was split into discrete cells of unit catchment size and

annual precipitation for each cell was plotted against the estimated erosion rate (Fig. 6.3.11).

This method, too finds a weak anti-correlation between erosion rate and total precipitation with R2= 0.02,

Pearson’s correlation coefficient = -0.1; which implies that heterogeneity of precipitation totals within

catchments decreases the statistical correlation.



131

200 400 600 800 1000 1200

0.0

0.1

0.2

0.3

0.4

0.5

0.6

An

nu

al

ero

sio

n (

mm

)

Annual precipitation (mm)

200 400 600 800 1000 1200

0.0

0.2

0.4

0.6

0.8

1.0

1.2

An

nu

al

ero

sio

n (

mm

)


These results indicate that annual total precipitation is not a strong control on catchment erosion rates

in the Alborz Mountains, and that this finding is not an effect of sub-catchment scale heterogeneity of

precipitation and its forcing.

As expressed in R square values, the correlation between annual erosion rate and total precipitation is

weakest for direct extrapolation and highest for lumping with weighting. This implies a major heterogeneity

of annual precipitation within catchments.

Fig. 6.3.9: Annual erosion and annual precipitation data, binned, for 86 drainage basins in the Alborz.

Error bars show 2σ interval. (A polynomial best fit to the data: y = 9E-08x2 - 5E-05x + 0.18 has R

2= 0.01).

Fig. 6.3.10: Annual erosion plotted against annual precipitation, for 86 drainage basins in the Alborz.

Each data point is weighted according to catchment area. Bars show 2σ interval within precipitation bins. (A

polynomial best fit to the data: y = -9E-08x2 + 0.0003x + 0.01 has R2= 0.21).



132

200 400 600 800 1000

0.0

0.1

0.2

0.3

0.4

0.5

An

nu

al

ero

sio

n (

mm

)


Fig. 6.3.11: Annual erosion plotted against annual precipitation estimated for 2515 data points in the

Alborz. Bars show 2σ interval within precipitation bins. (A polynomial best fit to the data: y = 4E-07x2 -

0.0006x + 0.39 has R2= 0.02)

6-3-5-2 Seasonal Precipitation and Erosion

Annual erosion and the variability of seasonal precipitation are weakly positively correlated with R2=

0.09; Pearson’s correlation coefficient=0.19 (Fig. 6.3.12).

The spatial pattern of the variability of seasonal precipitation (Fig. 6.3.8) matches some first-order

features of the decadal erosion pattern. Watersheds with high variability are located in southern part of the

mountain belt where erosion rates are high. In contrast, catchments in the north Alborz where erosion rates

are lower mainly have low variability of seasonal precipitation.

The seasonality of erosion is coupled with precipitation (Figure 6.3.13). The main seasons for

sediment supply are spring and autumn in the dry and wet part of the Alborz, respectively. Surprisingly, the

wettest seasons of winter and summer contribute less to the total erosional flux. This issue is addressed in

runoff section. Two different trends are recognised in Figure 6.3.13. Annual precipitation is negatively

correlated with the proportion of annual erosion in spring, but positively in autumn. Watersheds with annual

precipitation < 800mm contribute significantly to annual erosion (40-65%) in spring; therefore, an anti-

correlation is dominant. However, watersheds with annual precipitation >800mm have a significant part of

annual erosion (~40%) in autumn.

The combination of these two different trends overprints any correlation between annual precipitation

and erosion.

The role of seasonal precipitation becomes even clearer when a focus is developed on individual

seasons. This has been done by segregating catchments with winter precipitation total > 25% of the annual

total. This cut off coincides broadly with the 800 mm annual total precipitation mark. In 43 catchments where

winter precipitation is > 25% of the annual total, annual average erosion rates are positively correlated with

the proportion of the total annual precipitation in winter, with R2 = 0.26; Pearson’s correlation coefficient =

0.46 (Fig. 6.3.14). In these catchments, winter precipitation is seen to drive erosion.



133

0.24 0.26 0.28 0.30 0.32 0.34 0.36 0.38 0.40 0.42 0.44

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

An

nu

al e

ro

sio

n (

mm

)

Proportion of total annual precipitation in winter

300 400 500 600 700 800 900 1000 1100 1200

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0 Spring

Summer

Autumn

Winter

Fra

cti

on

of

an

nu

al

ero

sio

n


0.04 0.06 0.08 0.10 0.12 0.14 0.16

0.0

0.2

0.4

0.6

0.8

An

nu

al e

ro

sio

n (

mm

)

Variability of total seasonal precipitation normalized

by total annual precipitation

Fig. 6.3.12: Annual erosion plotted against variability of seasonal precipitation estimated in 88

drainage basins. Bars show 2σ confidence interval within variability of precipitation bins. (A polynomial best

fit to the data: y = 52.0x2 - 7.95x + 0.46 has R

2= 0.09).

Fig. 6.3.13: Proportion of annual erosion plotted against annual precipitation estimated for 79

watersheds. Bars show 2σ confidence interval within precipitation bins.

Fig. 6.3.14: Annual erosion plotted against proportion of annual precipitation in winter estimated in

43 drainage basins in which receives > 25 % of annual precipitation. Bars show 2σ confidence interval within

proportion of precipitation bins. (A polynomial best fit to the data: y = 12.72x2 - 7.22x + 1.13 has R

2= 0.26,

and Pearson’s correlation coefficient = 0.46)



134

0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.32 0.34 0.36

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

An

nu

al

ero

sio

n (

mm

)

Proportion of annual precipitation in summer

In contrast, no correlation has been found between erosion and the proportion of precipitation in

summer in all catchments where this proportion > 0.15% (R2 = 0.004; Pearson’s correlation coefficient = -

0.06) (Fig. 6.3.15).

Fig. 6.3.15: Annual erosion rate plotted against proportion of annual precipitation in summer

estimated in 43 drainage basins in which summer precipitation > 15 % of annual precipitation. Bars show 2σ

confidence interval within proportion of precipitation bins. (A polynomial best fit to the data: y = 1.36x2 -

0.91x + 0.33 has R2 = 0.004)

In summary, erosion of the Alborz Mountains appears to depend on seasonal rather than annual total

precipitation. In particular, winter precipitation correlates with catchment erosion rates, and rates are highest

where the proportion of winter precipitation is greatest. The proportion of summer precipitation has no

influence on catchment erosion rate.


CHAPTER 6: Decadal Erosion Rate Controls - Vegetation

135

6-4 Vegetation

6-4-1 Introduction

Vegetation impedes erosion by increasing soil cohesion and intercepting precipitation, but it also

facilitates chemical weathering (e.g., Leopold et al. 1995; Collin et al., 2004; Istanbulluoglu et al., 2005).

The capacity of runoff to detach material is proportional to shear stress and a portion of the applied fluid

shear stress is partitioned between plants and soils in the presence of vegetation. Vegetation increases the

threshold shear stress, below which the rate of material detachment from a soil and regolith is negligible.

Additional cohesion in the soil profile provided by vegetation also acts to stabilize the slopes against

landsliding (Prosser et al., 1995; Istanbulluoglu et al., 2004; Tucker et al., 2006).

The effect of vegetation on erosional processes depends on vegetation type, climate, geology,

topography, and anthropogenic and natural disturbances.

In forests, dense upper- and understory vegetation combine with well developed litter layers to protect the

soil surface from rain splash. A dense vegetation cover with a relatively shallow root system protects

slopes against erosion by runoff, but has little impact on the propensity to deep-seated failure. Forest

vegetation with deep-reaching root clumps is capable of holding together steep slopes in weathered

bedrock, lowering the rate of landsliding. Changes of vegetation cover can have a profound effect on

erosion rate. They can be brought about in many ways. For example, earthquake can cause dieback

through the shearing of tree root mass, reducing water uptake and health of trees, and thereby promoting

elevated rates of slope failure in epicentre areas. Tree fall has been recognized as a major type of

disturbance (Garwood et al., 1979; Wells et al., 2001).

Lightning fires can clean tracts of dryland vegetation, greatly reducing the threshold for erosion

(e.g., Hevely, 1980; Wondzell & King, 2003). Ashfall from volcanic eruptions changes the chemistry of

soils and the runoff infiltration balance, resulting in sweeping and long-lived vegetation changes. Finally,

forest clearance and tillage have resulted in increased soil loss and erosion in many places while irrigation

and terracing have reduced these processes elsewhere.

Overall, there is thought to be a negative correlation of erosion with vegetation for a given

topography: as vegetation density increases, the erosion rate decreases. In tectonically active areas, this

may result in steeper slopes and lower drainage density, and therefore greater relief in densely vegetated

areas. In the long run, this topographic change could negate the clamping of erosion by vegetation.

A strategy to deconvolve the effect of vegetation from other controls on erosion has not yet been

designed. In this section, I will document vegetation type and density in the Alborz, and explore their

influence on erosion.

6-4-2 Flora in the Alborz

Based on plant species, climate and physiography, Iranian habitats fall into four ecological zones.

From north to south they are: Hyrcanian, Irano-Touranian, Zagros, Gulf-Omanian (Heshmati, 2007).

The location of the Alborz Mountains between the South Caspian Sea and the Central Iranian

Plateau imposes a major climatic gradient; consequently, a great biological diversity with variegated flora



136

and fauna is observed. The mountain range encompasses a Hyrcanian province in the North and an Irano-

Touranian province in the south.

The Hyrcanian forests of the north Alborz are a relict of the Tertiary Hyrcanian flora. They

occupy the flanks and summits of the coastal mountains at the southern margin of the Caspian Sea,

linking across the Aras River with humid forests on the south-eastern shores of the Black Sea through

Transcaucasia. This province is often treated as a part of the Euro-Siberian region (Blondel & Aronson,

1999; Bobek, 1991). Pterocarya fraxinifolia is an example of a Tertiary relic element growing in the

lower-altitude forest of the Hyrcanian province since the Miocene-Pliocene (Zohary, 1973; Akhani &

Salimian, 2003; Ramezani et al., 2008).

Dense Hyrcanian forest constitutes a green belt over the northern slopes of the Alborz Mountains.

The width of the forested zone varies from 12-70 km and the total area is 19,250 km 2. It extends from sea

level to an altitude of 2,800 m asl, and consists of mainly deciduous forest of beech Fagus orientalis,

hornbeam Carpinus, maple Acer, lime Tilia, alder Alnus and oak Quercus castanaefolia. Ground flora is

abundant and includes grasses, herbs, ferns, shrubs and saplings. Within this zone, dense vegetation is

supported by high annual precipitation from 600 to 2,000mm, a considerable part of which falls in

summer. Above the timber line at elevations of 2,800m to 3,300m perennial spiny cushion-like plants

predominate. This flora shows affinities with the vegetation of the high mountains of the Hindu Kush

(UNDP, 2003; Rouhi-Moghaddam et al., 2007; Ramezani et al. 2008).

An Irano-Touranian arid to semi-arid flora dominates the southern slope of the Alborz

Mountains. Inside the mountain belt, it consists of an open forest steppe zone with a width of 30-90 km,

connected to the flat lands of Central Iran with desert vegetation further to south. Trees and shrub species

of this flora are significantly resistant to summer drought and heat, and can tolerate winter cold.

Dominant species include sub-steppic Juniperus sabina, J. communis, Amigdallus scoparia, Onobrychis

cornuta, Acantholimon spp., Astragalus spp., Artemisia aucheri, Alleum spp., Bromus tumentellus,

Pistacia vera, Berberos integessima, Acer spp.

In the Irano-Touranian zone the maximum precipitation is 500 mm per year in the mountains and

200 mm per year on the flats (UNDP, 2003; Heshmati, 2007).

Anthropogenic disturbance is important in the northern and southern periphery of the mountain belt

where the population density is high and substantial farming activity is common. Although the present

day vegetation pattern broadly fits the ecological zonation, deforestation and agriculture have locally

affected the vegetation density, particularly in low relief areas and along major valleys. In the north

Alborz, this has reduced vegetation biomass, but along the southern margin of the mountain belt irrigation

has created grass zones along waterways.

6-4-3 Vegetation Index

The Vegetation Index (VI) is an empirical measure of vegetation mass at the land surface.

Vegetation indices can be constrained from multiple-wavelength satellite images. To enhance the

vegetation signal, two (or more) frequency bands can be combined. Commonly used bands are the red

(0.6-0.7 mm) and NIR (near infra-red) wavelengths (0.7-1.1 mm).



137

Two vegetation index algorithms, NDVI and EVI, are used globally. The standard Normalized Difference

Vegetation Index (NDVI), also referred to as greenness index, is most often used, but susceptible to error

and uncertainty over variable atmospheric and canopy background conditions. It represents the absorption

of visible red radiation and reflected near infra red radiation within the vegetation canopy, expressed on a

scale from -1 to +1. Values between -0.2 and 0.05 are characteristics for snow, inland water bodies,

deserts and exposed soils, and values from about 0.05 to 0.7 are associated with progressively increasing

amounts of green vegetation (Myneni et al., 1997).

In 1995, Liu & Huete proposed the Enhanced Vegetation Index (EVI), a modified NDVI

combining blue, red, and near infra red bands from Moderate Resolution Imaging Spectrometer (MODIS)

data. This index has improved sensitivity in high biomass regions and improved vegetation monitoring

through a decoupling of the canopy background signal and a reduction in atmosphere influences (Tucker,

1979; Liu & Huete, 1995; Huete et al., 1997; Huete et al., 1999; Huete et al., 2002; Matsutisha et al.

2007).

EVI data has been proposed by the MODIS (Moderate Resolution Imaging Spectroradiometer)

Land Discipline Group as a global-based vegetation index for monitoring the Earth’s terrestrial

photosynthetic vegetation activity.

6-4-3-1 Enhanced Vegetation Index (EVI) in the Alborz

I have used EVI to characterise the vegetation state of the Alborz Mountains. Enhanced

Vegetation Index data for the Alborz at a spatial resolution of 250m were obtained from

http://iridl.ldeo.columbia.edu/, managed by NOAA's climate program office and Columbia University.

The data were produced by the MODIS VI algorithm on a per-pixel basis, relying on multiple

observations over 16-day periods to generate a composite product. The goal of compositing

methodologies is to select the best observation for a pixel, from all the retained, filtered data, to represent

each pixel over the 16-day compositing period. The time interval covered by the data used in this study is

January to December 2006.

Thus, the EVI data allow an analysis of seasonal vegetation changes. It is assumed that the broad

pattern observed in 2006 is representative of the vegetation state of the Alborz Mountains over the longer

time interval covered by the erosion data described in chapter 5.

Raw EVI data was downloaded as ASCII in ARC map, converted to grid and projected; I built up

seasonal and annual EVI maps from the original data by averaging the pixel values over multiple 16-day

intervals. The mean EVI was calculated per pixel and catchment-wide, and used to examine the

correlation of erosion with vegetation mass by means of three statistical approaches of lumped un-

weighted, weighted and direct extrapolation.

6-4-4 Spatial Pattern of EVI

The Enhanced Vegetation Index map of north Iran (Fig. 6.4.1) shows a sharp contrast between

the northern and southern flank of the Alborz Mountains. The wet, north flank has a high EVI, commonly

> 0.15 and the dry, southern flank has EVI <0.15.



138

Fig. 6.4.1: Mean annual EVI in northern Iran derived from MODIS (Moderate Resolution

Imaging Spectroradiometer) data for the time interval January - December 2006. The grid is shown at

250m resolution. Individual values represent the average of all 16-day composites during the observation

period.

Catchment-wide values of mean annual EVI (Fig. 6.4.2) range from 0.05 to 0.41, with two

subsets of EVI = 0.05 - 0.20 and EVI = 0.25 - 0.45 corresponding to the Hyrcanian and Irano-Touranian

provinces, respectively (Fig. 6.4.3).

Fig. 6.4.2: Smoothed mean annual EVI in the Alborz. Mean EVI calculated from MODIS

(Moderate Resolution Imaging Spectroradiometer) observations, smoothed at the catchment scale using a

circular moving mean with 15km diameter, and plotted with1 km spatial resolution.

Catchments in the SW and central Alborz, and some catchments around the main divide have a

mean annual EVI of 0.05-0.1. Low EVI values are also found in a section of the Haraz valley, a north-

draining longitudinal valley situated in the rainshadow of the northern frontal range. Catchments with

high EVI (0.3-0.35) are located in northernmost part of the mountain range, mainly in the NW, where the

annual precipitation rates are highest.



139

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45

0

5

10

15

20

25

30

Nu

mb

er

of

ca

tch

me

nts

Mean annual EVI

Hyrcanian province Irano-Touranian province

Fig. 6.4.3: The frequency of mean annual EVI in the Alborz. Catchment-average EVI values for

108 drainage basins are bi-modally distributed, showing the strong contrast between the Irano-Touranian.

province and the Hyrcanian province.

6.4.5 Seasonality of EVI

Throughout the Alborz, vegetation biomass is strongly seasonal. I have calculated pixel- specific

seasonal means for spring (21th March-21

th June), summer (22

th June-22

th september), autumn (23

th

September-21th December) and winter (22

th December-20

th March). Figure 6.4.4 shows seasonal means in

winter (6.4.4a) and summer (6.4.4b) as two end members. The mean EVI value in winter is low, ranging

from -0.09 to 0.30. The lowest values are characteristic of the area covered by snow in Irano-Touranian

province; the higher values are characteristic of Hyrcanian province. The contrast between north and

south flank persists in summer, but the EVI values range in a different interval between 0.1 and 0.55.

Winter mean EVI values are lowest in an area between the peaks of Damavand in the central

Alborz and Alam Kuh in the central-west Alborz, including the highlands above Tehran-Karaj. This

region has wet winters and very dry summers. In areas with high annual mean EVI, summers are wetter

than winters. Thus, where the biomass is low, precipitation falls when vegetation cover is weakest, and

where the biomass is high, precipitation is concentrated in times with the densest cover.

Seasonality in the EVI is also evident in catchment-wide data (Fig. 6.4.5). Winters have the

lowest EVI; summers and springs have similar, high EVIs.

In all seasons, the mean seasonal EVI increases with the annual mean. Moreover, the absolute

range of seasonal mean values increases with the annual mean EVI.



140

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Spring

Summer

Autumn

Winter

Me

an

se

aso

na

l E

VI

Mean annual EVI

Fig. 6.4.4: Pixel-specific mean seasonal EVI derived from MODIS for (a) winter and (b) summer 2006 at

the 250 m resolution.

Fig. 6.4.5: Mean seasonal and annual EVI values for 108 drainage basins in the Alborz.

When normalized with respect to the annual mean EVI (Fig. 6.4.6) the relative variability of seasonal

means changes systematically with mean annual EVI, and the Irano-Touranian and Hyrcanian provinces



141

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

Spring

Summer

Autumn

Winter

Me

an

se

as

on

al

EV

I n

orm

ali

ze

d b

y m

ea

n a

nn

ua

l E

VI

Mean annual EVI

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.2

0.4

0.6

0.8

Va

riab

ilit

y o

f n

orm

ali

zed

sea

so

na

l E

VI

Mean annual EVI

can be distinguished. For mean annual EVI< 0.15, seasonality of EVI decreases with increasing annual

mean values. For mean annual EVI> 0.15, seasonality of EVI is only weakly dependant on annual mean

values. This difference in relative strength of seasonality is strongly expressed in the winter and summer

means.

Fig. 6.4.6: Mean seasonal EVI normalized by mean annual EVI plotted against mean annual EVI

for 108 drainage basins.

Variability of seasonal EVI against annual EVI indicates the areas with higher EVI values have

lower seasonal variability of EVI (Fig. 6.4.7). The spatial distribution of this variability (Fig. 6.4.8) has a

dominant high in the high mountains of the central and west Alborz, where winter precipitation is as

snow. This area includes the high peaks of Damavand and Alam Kuh and the watersheds they feed. In

these areas summer is the main growing season, and, the summer EVI reaches up to 2.25 times mean

annual EVI, while in winter it ranges between 0.5 and -0.5.

Fig. 6.4.7: Variability of mean seasonal EVI normalized by mean annual EVI, plotted against

mean annual EVI, binned, for 108 drainage basins. Error bars show 2σ interval. (A polynomial best fit the

data: y = 3.32x2 - 2.42x + 0.71 has R

2 = 0.70, Pearson’s correlation coefficient = -0.81)



142

Fig. 6.4.8: Spatial pattern of variability of mean seasonal EVI normalized by mean annual EVI

estimated for 88 drainage basins. The grid is shown at 1 km resolution with a 15 km diameter smoothing

mean applied.

6-4-6 EVI and Precipitation

The spatial distribution of mean annual EVI (Fig. 6.4.1 & 6.4.2) matches the spatial pattern of

mean annual precipitation. Both show a general decrease of values from north to south and also from west

to east. Statistically, mean annual EVI is strongly correlated with mean annual precipitation (R2

= 0.50;

Pearson’s correlation coefficient = 0.74) (Fig. 6.4.9). Watersheds with low EVI <0.15 have a mean annual

precipitation of 300 to 600 mm; watersheds with high EVI > 0.15 have a mean annual precipitation

greater than 500 mm.

Crucially, these two domains have a marked and different seasonality of precipitation. This is

best seen in the seasonality of proportions of annual precipitation totals (Fig. 6.4.10).

Watersheds with EVI <0.15 have a different seasonal dynamic than more densely vegetated

drainage basins, except for autumn when the whole range receives the least proportion of annual

precipitation.

In areas with low mean annual EVI, up to 80% of precipitation falls in winter and spring (Fig.

6.4.10). The proportion of cold season precipitation gradually decreases as total annual rainfall, and with

it annual EVI, increases. In areas with the highest EVI, cold season contributions to total annual

precipitation can be as low as 35%.

In areas with low EVI, summer precipitation can be as little as ~ 10% of the annual total, and

where EVI is high, its contribution to the annual total can be up to 35%. Throughout, autumn

precipitation is a relatively minor component of the annual total everywhere in the mountain.

Combining Figures 6.4.9 and 6.4.10, it is clear that watersheds with substantial precipitation

throughout year constantly are highly vegetated with EVI > 0.15. In contrast, watersheds with EVI < 0.15

are dry in autumn and summer but have precipitation in winter and spring. The seasonal drought and

limited precipitation during the wet seasons together result in a low vegetation density (EVI < 0.15).



143

200 400 600 800 1000 1200

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

Me

an

an

nu

al

EV

I

Mean annual precipitation (mm)

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.1

0.2

0.3

0.4

0.5 Spring

Summer

Autumn

Winter

No

rm

ali

ze

d s

ea

so

na

l p

re

cip

ita

tio

n

Mean annual EVI

Interestingly, higher summer precipitation appears to enhance biomass. In contrast, higher winter

and spring precipitation cause a decrease of biomass. One possible mechanism for this is erosion (See

section 6-4-7).

Fig. 6.4.9: Mean annual EVI plotted against mean annual precipitation, for 108 drainage basins in

the Alborz. Bars show 2σ interval within precipitation. (A polynomial model fits the data with: y = -3E-

07x2 + 0.0007x - 0.12; R

2 = 0.52, Pearson’s correlation coefficient = -0.73).

Fig. 6.4.10: Seasonal proportion of mean annual precipitation and mean annual EVI for 97

drainage basins in the Alborz. Error bars show 2σ interval.

6-4-7 EVI and Erosion

To explore vegetation controls on erosion rate, three statistical approaches of lumping without

weighting, lumping with weighting and direct extrapolation were employed.

Range-wide statistics are relatively poor as a result of the combination of two fundamentally different

ecological zones across the mountain belt.

Unweighted, catchment-wide mean annual EVI is poorly anti-correlated with erosion (Fig.

6.4.11) (R2= 0.13; Pearson’s correlation coefficient = -0.03). However, when catchment-wide values are

weighted for catchment size, then a more meaningful anti-correlation emerges (R2= 0.20; Pearson’s

correlation coefficient = -0.33) (Fig. 6.4.12). Using direct extrapolation, a weaker anti-correlation is

found between erosion rate and EVI with R2= 0.06 and Pearson’s correlation coefficient = -0.22 (Fig.

6.4.13).



144

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.0

0.2

0.4

0.6

An

nu

al

ero

sio

n (

mm

)

Mean annual EVI

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.0

0.2

0.4

An

nu

al

ero

sio

n (

mm

)

Annual EVI

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.0

0.2

0.4

0.6

An

nu

al

ero

sio

n (

mm

)

Annual EVI

Fig. 6.4.11: Annual erosion rate and EVI for 89 drainage basins in the Alborz. Error bars show 2σ

interval. (A polynomial model fits the data with: R2 = 0.13y = 9.94x

2 - 4.47x + 0.59)

Fig. 6.4.12: Annual erosion rate and mean annual EVI estimated for 89 drainage basins in the

Alborz. Each data point is weighted according to catchment area. Error bars show 2σ interval. (A

polynomial model fits the data with: y = 8.93x2 - 4.35x + 0.65; R

2 = 0.20)

Fig. 6.4.13: Annual erosion rate and mean annual EVI for 2515 grid cells in the Alborz. EVI was

sampled discretely across the range and plotted against the interpolated local erosion rate. Error bars show

2σ interval. (A polynomial model fits the data with: y = 2.84x2 - 1.65x + 0.37; R

2 = 0.06).



145

These results indicate that at the mountain belt scale vegetation may have an effect on erosion

rate, such that increased vegetation density suppress erosion rates, as expected.

Sub-catchment scale variability of biomass decreases the correlation of vegetation and erosion. A

linear fit to the EVI-erosion data may obscure the true nature of their relationship. In all three statistical

approaches, erosion rates are lowest for intermediate EVI values of EVI= 0.20-0.25. Erosion rates

increase progressively on either side of this intermediate EVI range. A possible interpretation is that at

low EVI, lack of vegetation permits easy detachment of weathering products, and at high EVI, associated

high rainfall rates drive erosion regardless of vegetation cover. This would give rise to a strong

seasonality of erosion as a function of EVI.

The effect of EVI on the seasonality of erosion is illustrated in Figure 6.4.14. Spring is by far the

dominant erosion season in catchments with mean annual EVIs of up to 0.20. The importance of spring

erosion subsides as EVI increases beyond this value, and autumn gradually gains erosional importance.

This transition has been examined in more detail in Figure 6.4.15, in which watersheds with

annual EVI >0.15 are classified based on the seasonality of erosion. In north flank of the Alborz, where

most of these catchments are located, areas with the highest proportion of erosion in spring have the

lowest mean annual erosion rates, indicating an anti-correlation between EVI and erosion similar to the

dry southern Alborz. Only watersheds with very high mean annual EVI have high mean annual erosion

rates. In these catchments, erosion occurs in autumn, when mean seasonal EVI is significantly lower than

in spring and summer. Low seasonal vegetation cover may be a primary cause of high seasonal erosion

erosion rates.

Denser vegetation in growth seasons adds shielding, and soil strength. It may also modulate

surface runoff and impede transport of entrained sediment by increasing surface roughness. In general,

grass or vegetation filter strips trap sediments eroded upslope by modifying the hydraulic characteristics

of flow; they reduce flow turbulence and velocity, thus increasing infiltration and sediment deposition.

Vegetation can reduce runoff by 10-90 % and trap 77-97 % of sediment, according to flow velocity and

vegetation roughness (e.g., Abu-Zreig, 2001; Rey, 2004; Fiener & Auerswald, 2006; Legu´edois et al.,

2008). Trapped sediments are released in times with lower vegetation, which results in lower infiltration

and higher runoff. In the Alborz this occurs in autumn. This discussion anticipates a detailed examination

of runoff and its effects on erosion in the next section.



146

0.20 0.25 0.30 0.35 0.40

0.0

0.2

0.4

0.6

0.8

1.0 Spring

Summer

Autumn

An

nu

al

ero

sio

n (

mm

)

Annual EVI

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Spring

Summer

Autumn

Winter

Fracti

on

of

an

nu

al ero

sio

n

Annual EVI

Fig. 6.4.14: Seasonal fraction of annual erosion and mean annual EVI for 79 drainage basins in

the Alborz. Error bars show 2σ interval.

Fig. 6.4.15: Annual erosion and EVI for watersheds with EVI > 0.15. Each data bin is classified

according to the season of peak erosion (typically, this is when >40% of total annual erosion occurs).

Error bars show 2σ interval.


CHAPTER 6: Decadal Erosion Rate Controls - Runoff

147

0 500 1000 1500 2000 2500

0

5

10

15

20

25

30

Nu

mb

er

of

ca

tch

me

nts

Annual runoff (mm)

6-5 Runoff

6-5-1 Introduction

Runoff is the portion of rainfall which is not lost to infiltration or evapo-transpiration; it may

include recharge from groundwater and melt water from snow packs. Base flow is the portion of runoff

that results from seepage of ground water into a channel. In many locations, infiltration excess and

saturation excess are the main mechanisms for generating runoff. Infiltration excess overland flow, or

Hortonian flow (Horton, 1933) occurs when the rainfall intensity exceeds the infiltration capacity of the

substrate. Saturation excess overland flow occurs when the soil water content reaches the storage capacity

and surplus rain cannot be accommodated in pore spaces (Wondzella & King, 2003; Smith & Goodrich,

2005). Semiarid and sub-humid zones are prone to infiltration excess runoff because of predominance of

relatively short, high intensity rainfalls on bare rock surfaces. In contrast, saturation excess runoff is more

common in humid areas with greater rainfall volumes but lower intensities. In these areas there is no

excess runoff early in the wet season (Mccartney et al., 1998; Smith & Goodrich, 2005).

Runoff is a principal driver of erosion by detaching and entraining sediment and by adding to the

transport capacity of channelised flow (e.g., Cerda, 1997). Vegetation modifies overland flow as

discussed in section 6-4. In bare zones, infiltration excess runoff is in direct interaction with substrate and

slope, driving erosion effectively; in contrast, large infiltration in vegetated zones precludes any

significant interaction.

6-5-2 Spatial Distribution of Runoff

The spatial distribution of runoff in the Alborz Mountains is known from daily discharge data

from 108 hydrometric stations operated by the Ministry of Energy of Iran, covering 5-55 years (TAMAB,

unpublished data). Specific annual runoff has been computed as the annual average river discharge

divided by drainage area upstream of the gauging station. The values range from 30-2000 mm between

stations. The modal runoff in the Alborz is 200-250 mm/yr, and the frequency count of specific annual

runoff per catchment has an asymmetric distribution with positive skew and coefficient of variation of

86% for the population of gauged catchments (Fig. 6.5.1).

Fig. 6.5.1: The frequency of specific annual runoff from gauged catchments in the Alborz. This

plot includes runoff data from 108 drainage basins.



148

0 200 400 600 800 1000 1200 1400 1600

0

2

4

6

8

0

2

4

6

8

Annual runoff (mm)

Spring

Summer

Se

as

on

al

run

off

(m

m/d

ay

)

Autumn

Winter

Figure 6.5.2 shows the spatial pattern of annual runoff smoothed at 15 km length scale. It

decreases from west to east and from north to south. The south flank of the mountain belt has low runoff

values of < 250 mm/yr, except for the central Tehran- Karaj sector and the highlands on the main divide.

In the northern Alborz, runoff is lowest in the east and some segments of the Haraz valley (See Fig. 3.1

for the location).

Fig. 6.5.2: Total annual runoff in the Alborz, derived from daily discharge measurements at 108

hydrometric stations operated by TAMAB. Runoff data was smoothed at catchment scale using a circular

moving mean with 15km diameter, and displayed at 1 km spatial resolution.

6-5-3 Seasonality of Runoff

Runoff is seasonal in the Alborz (Fig. 6.5.3). Throughout the mountains, spring has the highest

runoff rate, and summer has the lowest runoff rate. Autumn has low runoff rates in catchments with

annual runoff <600-700 mm, but high runoff rates in catchments with annual runoff > 600-700mm.

Fig. 6.5.3: Seasonal runoff and annual runoff estimated for 108 drainage basins in the Alborz.

Bars show 2σ interval within annual runoff bins.



149

0 200 400 600 800 1000 1200 1400

0.1

0.2

0.3

0.4

0.5

0.6

Spring

Summer

Autumn

Winter

Annual runoff (mm)

No

rm

ali

ze

d s

ea

so

na

l ru

no

ff

0 200 400 600 800 1000 1200 1400

0.05

0.10

0.15

Va

ria

bil

ity

of

no

rm

ali

ze

d s

ea

so

na

l ru

no

ff

Annual runoff (mm)

In order to better characterize the intra-annual variability of catchment runoff, average seasonal

runoff was normalized by total annual runoff. Results are shown in Figure 6.5.4. This figure reveals two

main domains with different dynamics. Catchments with total annual runoff < 600 mm have more than

40% and up to 65 % of total annual runoff in spring and less than 15 % in summer. In contrast,

catchments with the total annual runoff > 600 mm have a more even distribution with 20-35 % of the total

annual runoff in each of four seasons.

Fig. 6.5.4: Normalized seasonal runoff plotted against total annual runoff for 108 drainage basins

in the Alborz. Bars show 2σ intervals within annual runoff bins.

Total annual runoff and variability of normalized seasonal runoff are weakly anti-correlated (Fig.

6.5.5), with R2= 0.11, Pearson’s correlation coefficient of -0.31. Again, there appear to be two domains.

Where total annual runoff is less than about 600 mm, the average variability is 0.11-0.13, but the range of

catchment-specific values is large. Where total annual runoff is more than about 600 mm, the average

variability is 0.05-0.07, and the range of catchment specific values is small, specifically in catchments

with very high runoff totals.

Fig. 6.5.5: Variability of normalized seasonal runoff as a function of annual runoff for 108

drainage basins. Bars show 2σ intervals within annual runoff bins. (A polynomial model fits the data

with: y = 9E-09x2 - 6E-05x + 0.13; R

2 = 0.11)



150

300 400 500 600 700 800 900 1000 1100 1200

0

200

400

600

800

1000

1200

1400

1600

To

tal

an

nu

al

ru

no

ff (

mm

)


The spatial pattern of the seasonal runoff variability (Fig. 6.5.6) displays a strong visual

correspondence with first order features of the annual erosion pattern. Watersheds with high variability

located in south-western and central part of the mountain belt have the highest erosion rates, and in the

extreme northeast erosion and runoff seasonality appear to peak together.

Fig. 6.5.6: Variability of the proportion of annual runoff in four seasons in the Alborz, estimated

from daily discharge measurements at 108 gauging stations operated by TAMAB. The map pattern was

smoothed at catchment scale using a circular moving mean with 15km diameter, and displayed at 1 km

spatial resolution.

6-5-4 Runoff and Precipitation

Annual runoff is strongly correlated with annual precipitation. The best fit model is a polynomial

function of degree two, with R2= 0.36 and Pearson’s correlation coefficient = 0.57 (Fig. 6.5.7).

Catchments with the highest amounts of precipitation, mainly located in north-northwest of the mountain

belt, have the greatest runoff; and catchments in the west-southwest and east-southeast of the Alborz

combine low runoff and low precipitation.

Fig. 6.5.7: Total annual runoff and total annual precipitation in 96 drainage basins in the Alborz.

Bars show 2σ interval within precipitation bins. Straight line is 1:1 relation. Runoff is mainly less than

precipitation. (A polynomial model fits the data with: y = 0.001x2 - 0.47x + 280.19; R

2 = 0.38)



151

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Spring

Summer

Autumn

Winter

Pro

po

rti

on

of

an

nu

al

ru

no

ff

Proportion of annual precipittaion

200 400 600 800 1000 1200

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 Spring

Summer

Autumn

Winter

No

rm

ali

ze

d s

ea

so

na

l ru

no

ff


The non-linear relation of runoff and precipitation implies that for total annual precipitation of

less than about 600 mm, an increase in precipitation does not systematically result in an increase in

runoff. Above this value, runoff does increase with precipitation.

In most catchments runoff < precipitation, as expected. The mountain belt wide average ratio of runoff :

precipitation is 0.63. However, it is 1.05 for catchments with annual runoff >600 mm and 0.43 for

catchments with annual runoff <600mm. At stations where total runoff exceeds total precipitation (12 out

of 106 catchments), anomalously high discharge may be due to seepage and spring flow into the

catchment, insufficient overlap between the intervals over which precipitation and runoff data were

collected, or measurement error.

Seasonal runoff is tied with annual precipitation, according to Figure 6.5.8, but the link is

complex (Fig. 6.5.9). Spring has the greatest runoff across the mountain belt except where annual

precipitation exceeds about 1m in the north-northwest Alborz. There, winter and autumn have the highest

fraction of runoff even though summer is the wet season. In drier parts of the mountain belt with annual

precipitation < 800 mm, the spring runoff peak is paired with a winter precipitation peak.

Fig. 6.5.8: Normalized seasonal runoff plotted against annual precipitation for 96 drainage basins

in the Alborz. Bars show 2σ interval within precipitation bins

Fig. 6.5.9: Proportion of annual runoff plotted against proportion of annual precipitation, in each

of four seasons, for 96 drainage basins in the Alborz. Bars show 2σ interval within bins of proportional

annual precipitation.



152

300 400 500 600 700 800 900 1000 1100 1200

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

Va

ria

bil

ity

of

no

rm

ali

ze

d s

ea

so

na

l ru

no

ff


The variability of total seasonal runoff normalized by total annual runoff is anti-correlated with

total annual precipitation, where R2 = 0.22; Pearson’s correlation coefficient = -0.5 (Fig. 6.5.10).

For precipitation < 800 mm/yr, the seasonal runoff variability can range widely from 0.02 to 0.2, but

where precipitation > 800 mm/yr, it never exceeds 0.10.

Fig. 6.5.10: Variability of normalized seasonal runoff plotted against annual precipitation for 108

drainage basins in the Alborz. Bars show 2σ interval within precipitation bins. (A polynomial model fits

the data with: y = 2E-08x2 - 0.0001x + 0.18; R

2 = 0.22)

6-5-5 Dynamics of Seasonal Runoff

Seasonal runoff is modulated by snow accumulation and melt, evapo-transpiration, and

infiltration and base flow. This gives rise to a complex precipitation-runoff dynamic in the Alborz

Mountains.

In the northern Alborz, summers are wet and autumns dry but runoff peaks in autumn. The lag in

runoff may be due in part to significant evapo-transpiration and infiltration in summer, while exfiltration

and seepage of summer rain could swell autumn base flow, but this remains ill-constrained. High

fractions of autumn runoff are limited to the frontal range of the northern Alborz, where summer

precipitation is greatest (Fig. 6.5.11a).

In the south flank of the mountain range a lag time exists between the winter precipitation high

and the runoff peak in spring. The highest fraction of annual runoff occurs during spring throughout the

southern Alborz (Fig. 6.5.11b).

At high altitude, most winter precipitation is as snow. It accumulates, suppressing winter runoff (Fig.

6.5.12). Snow melt in spring combines with spring precipitation in the southern Alborz, while seepage

may also contribute to the delay of the runoff peak.

In a study of 27 years of runoff from a catchment above the Amir Kabir reservoir in the southern

Alborz, Mashayekhi and Mahjoub (1991) have found that the lag time between the winter precipitation

peak and the runoff peak is due to snow melt from mid-March at lower elevation until end May at higher

elevation. Moussavi et al. (1989) have also argued that the runoff in spring is mainly generated by

snowmelt in the south flank of the Alborz.



153

0.15 0.20 0.25 0.30 0.35 0.40

0.10

0.15

0.20

0.25

0.30

0.35

Pro

po

rtio

n o

f an

nu

al ru

no

ff i

n w

inte

r

Proportion of annual precipitation in winter

Fig. 6.5.11: Proportion of annual runoff during (a) autumn and (b) spring in the Alborz, derived

from daily discharge measurements at 108 hydrometric stations operated by TAMAB. Runoff was

smoothed at the catchment scale using a circular moving mean with 15km diameter, and displayed at1 km

spatial resolution.

Fig. 6.5.12: Proportion of annual runoff in winter plotted against proportion of annual

precipitation in winter for 96 drainage basins in the Alborz. Bars show 2σ interval within precipitation

bins. (A linear model fits the data with: y = -0.40x + 0.35; R2 = 0.21; Pearson’s correlation coefficient = -

0.46).

(a)

(b)



154

0 100 200 300 400 500 600 700 800 900

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

An

nu

al

EV

I

Annual runoff (mm)

0 100 200 300 400 500 600 700 800 900 1000 1100

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

An

nu

al

EV

I

Annual runoff (mm)

6-5-6 Runoff and EVI

Although both runoff and vegetation density are linked with annual precipitation, annual EVI is

poorly correlated with annual runoff; R2 = 0.09 and Pearson’s correlation coefficient=0.31 (Fig. 6.5.13).

Mean annual EVI in catchments in south flank of the Alborz is anti-correlated with total annual runoff (R2

= 0.24; Pearson’s correlation coefficient = -0.43), but northern cachments display a positive correlation of

these variables (R2 = 0.21; Pearson’s correlation coefficient = 0.39).

Fig. 6.5.13: Annual EVI and runoff for 106 drainage basins Bars show 2σ interval within runoff

bins. (A polynomial model fits the data with: y = -2E-08x2 + 0.0001x + 0.19; R

2 = 0.09).

Fig. 6.5.14: Same as Fig. 6.5.13, but with data segregated for north and south flank of the

mountain belt, shown as open and filled squares, respectively. (A polynomial model fits the data with: y =

-6E-08x2 + 0.0002x + 0.19; R

2 = 0.21, and y = 9E-08x

2 - 0.0001x + 0.11; R

2 = 0.24, for north and south,

respectively.).

Moreover, annual EVI is strongly anti-correlated with the variability of seasonal runoff as

normalized by annual runoff, R2= 0.38 and Pearson’s correlation coefficient= -0.62 (Fig. 6.5.15). This

reflects the steady wet climate of the densely vegetated north flank of the mountain belt where high

infiltration rates suppress variability of runoff, and the highly seasonal runoff of the barren south flank,

where topographic slope and lithology may add to a high hydrological variability ( see sections 6.6 and

6.7).

Given the importance of the seasonality of runoff in dry part of the Alborz (Fig. 6.5.8 & 6.5.9),

normalized seasonal runoff has been explored in conjunction with annual EVI (Fig. 5.5.16).



155

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

An

nu

al

EV

I

Variability of normalized seasonal runoff

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

Spring

Summer

Autumn

Winter

An

nu

al

EV

I

Normalized seasonal runoff

Fig. 6.5.15: Annual EVI plotted against variability of normalized seasonal runoff for 108

drainage basins in the Alborz. Bars show 2σ interval within runoff bins. (A polynomial model fits the data

with: y = 1.84x2 - 1.64x + 0.38; R

2 = 0.38).

Fig. 6.5.16: Annual EVI and seasonal runoff normalized by annual runoff for 108 drainage basins

in the Alborz. Bars show 2σ interval in runoff bins.

A significant anti-correlation exists between annual EVI and the proportion of annual runoff in

spring, when a major amount of erosion occurs (Chapter 5), confirming the reduced effect of vegetation

impedance during peak erosion.

Vegetation also appears to affect the ratio of runoff and precipitation (Fig.6.5.17). With

increasing EVI, the runoff/precipitation ration decreases steadily to a minimum at annual EVI = 0.15-

0.20. In areas with denser vegetation, the ratio increases gradually with EVI.

Pearson’s correlation coefficient and R2 are improved, when they are estimated for north and

south flank catchments separately. Pearson’s correlation coefficient and R2 are -0.69 and 0.52 in the south

flank and 013 and 0.11 in the north flank, respectively. The meaningful anti-correlation of EVI and the

runoff/precipitation ratio in the south flank of the Alborz may imply that vegetation can affect erosion by

modulating the runoff/precipitation ratio.



156

0 500 1000 1500 2000

0.0

0.2

0.4

0.6

0.8

An

nu

al

ero

sio

n (

mm

)

Annual runoff (mm)

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

An

nu

al

run

off

/ a

nn

ua

l p

rec

ipit

ati

on

Annual EVI

Fig. 6.5.17: The ratio of annual runoff / precipitation plotted against annual EVI for 95

watersheds in the Alborz. Bars show 2σ interval within EVI bins. (A polynomial best fit the data : y =

9.78x2 - 4.46x + 0.89 has R

2= 0.11; Pearson’s correlation coefficient = -0.02)

6-5-7 Runoff and Erosion

The strength of the runoff control on erosion rate in the Alborz Mountains has been explored with

two statistical approaches, lumping without weighting and lumping with weighting. The direct

extrapolation method is not applicable in this case because runoff is an independent variable (See section

6.2).

Both methods show a non-linear relation of annual runoff and erosion, with an erosion maximum

at intermediate runoff values. However, the maximum is different between these methods. Lumped, un-

weighted catchment data show a near linear increase of erosion rate with runoff to a maximum at about

1300 mm total annual runoff. When weighted for catchment size, the data show an erosion maximum at

about 500 mm total annual runoff (Fig. 6.5.18 & 6.5.19, respectively). The large difference between these

results underlines the influence of weighting.

Seasonal variability of runoff and erosion are weakly positively correlated (Fig. 6.5.20), with R2 =

0.04; Pearson’s correlation coefficient = 0.20.

Fig. 6.5.18: Annual erosion and runoff for 89 drainage basins in the Alborz. Bars show 2σ

interval within runoff bins. (A polynomial model fits the data with: y = -4E-08x2 + 0.0002x + 0.12; R

2 =

0.06; Pearson’s correlation coefficient = 0.20).



157

0 200 400 600 800

0.0

0.2

0.4

0.6

An

nu

al

ero

sio

n (

mm

)

Annual runoff (mm)

0.00 0.05 0.10 0.15 0.20

0.0

0.2

0.4

0.6

An

nu

al ero

sio

n (

mm

)

Variability of total seasonal runoff normalized by total annual runoff

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-0.15

0.00

0.15

0.30

0.45

0.60

0.75

0.90

An

nu

al e

rosio

n (

mm

)

Annual runoff / annual precipitation

Fig. 6.5.19: Annual erosion and runoff for 89 drainage basins in the Alborz. Each data point is

weighted according to catchment area. Bars show 2σ interval within runoff bins. (A polynomial model

fits the data with: y = -0.35x2 + 0.58x + 0.12; R

2 = 0.09; Pearson’s correlation coefficient = 0.16)

Fig. 6.5.20: Annual erosion plotted against variability of total seasonal runoff normalized by total

annual runoff for 88 drainage basins in the Alborz. Bars show 2σ interval within runoff bins. (A

polynomial model fits the data with: y = -0.68x2 + 1.05x + 0.12; R

2 = 0.04).

Fig. 6.5.21: Annual erosion plotted against annual runoff / precipitation for 95 watersheds in the

Alborz. Data are segregated for north and south flank catchments, shown with open and filled squares,

respectively. (Polynomial models fit the data for south and north flanks with: y = 0.62x2 - 0.04x + 0.15;

R2 = 0.35, and y = -0.28x

2 + 0.56x - 0.01; R

2 = 0.08, respectively). Bars show 2σ interval within

runoff/precipitation bins.



158

0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Spring

Summer

Autumn

Winter

Fra

cti

on

of

an

nu

al

ero

sio

n

Fraction of annual runoff

Erosion is relatively highly correlated with the runoff/precipitation ratio in both flanks of the

mountain belt. R2 and Pearson’s correlation coefficient are 0.08 and 0.27, and 0.35 and 0.57, in the north

flank and south flank, respectively.

Importantly, the seasonality of erosion is coupled with the seasonality of runoff (Fig. 6.5.22).

When the seasonal fraction of annual runoff is less than about 0.25, it scales linearly and directly with the

seasonal fraction of annual erosion, except during winter when erosion is always disproportionately low.

However, in catchments with a highly seasonal runoff, erosion rates are disproportionately high during

seasons with most discharge. This is true for the north flank of the mountain belt, where autumns have

most discharge, as well as for the south flank of the mountain belt, where discharge peaks in spring. Thus,

seasons with high runoff, set the erosion pattern, and on the scale of the mountain belt, runoff seasonality

caused by snow melt matches erosion “hot spots”. This link is explored next.

Fig. 6.5.22: Proportion of annual erosion plotted against proportion of annual runoff for four

seasons, for 78 watersheds in the Alborz. Bars show 2σ interval within runoff bins. Solid line is 1:1

relation.

6-5-8 Dynamics of Snowmelt-Generated Runoff

Episodic floods provide a primary mechanism for transport of large amounts of sediment; in

some cases more than 90% of the total annual sediment load is transported in several days of high runoff

as reported in different areas with different climate regime (e.g., Parker & Troutman, 1989; Dadson et al.,

2005; Beylich et al., 2006).

One efficient mechanism of flood generation is snowmelt runoff on a frozen substrate (e.g., Wade

& Kirkbride, 1996; Ferrick & Gatto, 2005; Beylich et al., 2006; Yavuz et al., 2006). This mechanism is

most effective in terms of erosion when thermal shock and freeze-thaw process drive high physical

weathering rates, producing regolith especially during periods of snow accumulation and melt. As a

result, material is available for erosion from hillslopes when transport capacity is greatest.

Daily observations of suspended sediment concentration and water discharge in the Haraz and

Sefid Rud catchments (Pazwash, 1982; Sadeghi et al., 2005), have shown episodic floods in late winter -

early spring, in which daily sediment discharge can exceed the mean annual value.



159

0.30 0.35 0.40 0.45 0.50 0.55 0.60

0.0

0.1

0.2

0.3

0.4

0.5

0.6

An

nu

al

ero

sio

n (

mm

)

Proportion of total annual runoff in spring

0 100 200 300 400 500

0

100

200

300

400

500

600

To

tal

se

as

on

al

ru

no

ff (

mm

)

Total seasonal precipitation (mm)

The erosional response to snowmelt-generated runoff is severe and proves to be a stronger control

on erosion than the total annual precipitation. It dominates the sediment flux from the southern Alborz.

Within this area, a higher correlation exists between the seasonal runoff and erosion. Specifically, for

catchments in which spring runoff exceeds 26% of the annual total runoff, annual erosion increases

systematically with increasing proportional importance of spring runoff.

A polynomial function of degree two (y = 1.37x2 - 0.59x + 0.15) defines the correlation between

the seasonal runoff and erosion with R2 = 0.18 and Pearson’s correlation coefficient = 0.40 (Fig. 6.5.23).

Fig. 6.5.23: Annual erosion as a function of the proportion of annual runoff in spring for 64

drainage basins with > 26 % of annual runoff in spring. Bars show 2σ interval within runoff bin. (A


2 = 0.18)

6-5-9 Dynamics of Groundwater-Generated Runoff

Groundwater can make up in excess of 50-90% of stream flow (Arnold, 2000). In the north

Alborz, it must constitute the bulk of the autumn runoff, when river discharge exceeds precipitation (Fig.

6.5.24). This contrasts with summers when precipitation exceeds runoff by 2.4 times. Given the fact that

summer erosion is low in the north Alborz despite heavy precipitation, and high in autumn when runoff

peaks, the ground water dynamic deserves some further attention.

Fig. 6.5.24: Total seasonal runoff and precipitation in summer and autumn for 64 catchments in

the north flank of the Alborz. Filled circles are summer data; open circles are autumn data. Solid line is

1:1 relation.



160

0.26 0.28 0.30 0.32 0.34 0.36 0.38

0.0

0.2

0.4

0.6

0.8

1.0

An

nu

al ero

sio

n (

mm

)

Proportion of total annual runoff in autumn

It has been noted that riparian evapo-transpiration is greater in wet periods than in dry periods. In

the north Alborz, where EVI peaks in summer, this may lead to a large seasonal loss of surface and

groundwater. Only deeper groundwater is relatively unaffected, and this may dominate recharge of

surface flow in autumn. In that season, lower EVI reflects reduced biomass, and water uptake and evapo-

transpiration loss may be less as a result (e.g., Federer, 1973; Daniel, 1976; Tallaksen, 1995). This could

be a primary cause of high flow rates in autumn, in the wettest part of the north Alborz. If this

interpretation is correct, then there is a time lag of several months between precipitation and outflow of

groundwater in these catchments The lengths of the flow paths involved are not known.

Where this effect is large and autumn runoff contributes disproportionately to the annual total

runoff, erosion rates appear to be affected. This can be seen in the runoff and erosion statistics for

catchments where the proportion of total runoff in autumn is greater than 25% (annual precipitation >700

mm). In these catchments annual erosion rates increase with the proportion of autumn runoff. The data

are best fit by a non-linear model with R2 = 0.44; Pearson’s correlation coefficient = 0.44) (Fig. 6.5.25).

However, only when the proportion of autumn runoff exceeds about 35% of the annual total its effects on

catchment erosion is important.

The mechanism by which, strong autumn recharge drives catchment erosion, has not been studied

in detail. Recharge is focused in topographic lows and unlikely to affect hillslopes. Therefore, autumn

sediment mobilization must occur in channels rather than on hillslopes. Sediment delivery to channels

may be a more gradual process, or concentrated in summer months when precipitation rates are high, but

onward transport happens when stream flow is greatest, and this is recorded at gauging stations.

Fig. 6.5.25: Annual erosion rate plotted against the proportion of annual runoff in autumn for 32

drainage basins with > 25 % of annual runoff in autumn. Bars show 2σ interval within runoff bins. (A


2 = 0.44)

In summary, overland flow caused by direct precipitation, enhanced by snow-melting in spring,

or groundwater recharge in autumn, and modified by evapo-transpiration are the main mechanisms of

runoff generation in the Alborz.

Across the mountain belt, erosion rates increase with annual runoff, peak at intermediate runoff

values, and then decrease. This pattern is overprinted by a strong seasonality of the runoff control on

erosion. In the semi-arid southern and high Alborz, winter precipitation accumulates as snow and



161

meltwater merges with spring rain to maximize runoff at a time when the products of thermal weathering

of friable rocks are first exposed. As a result, erosion rates peak in spring in this part of the mountain belt,

and this mechanism may well be responsible for the highest erosion rates of the entire region. In the north

flank of the Alborz Mountains, heavy summer rains do not translate in significant runoff, probably due to

significant evapo-transpiration loss and infiltration through dense vegetation. However, deep ground

water recharges autumn stream flow, and where this is most effective, autumn runoff drives catchment

erosion, probably through the purging of sediment concentrated in stream beds and valley floors.


CHAPTER 6: Decadal Erosion Rate controls-Lithology

162

6-6 Lithology

6-6-1 Introduction

In this section, an effort is made to establish a conceptual approach to classifying geological

substrate on the basis of its mechanical properties. In the absence of any systematic measurement of

geomechanical properties of rocks in the Alborz, porosity is identified as a convenient mechanical proxy

measure of rock strength. Porosity is also a first control on the weathering susceptibility of rocks.

Infiltration and seepage of meteoric water depends on the interaction of fluids with contrasting

lithologies. Since the mineral-fluid interface, including that in pores, is the primary location of weathering

reactions (Fitzner, 1990), porous rocks with greater ability to hold fluids from precipitation or melt water

tend to have higher chemical weathering rates than lithologies with low porosity.

Porosity is also crucial in terms of freeze-thaw processes in which hydraulic pressure in the voids

during freezing increases the pore-size distribution and average pore radius (Fitzner, 1990) leading to a

loss of cohesion, and a capacity to accept and retain more moisture. This results in a significant reduction

of the unconfined compressive strength in porous rocks; in contrast, rocks with low porosity permit

neither the infiltration nor the pressurization of fluid required for significant freeze-thaw damage (e.g.,

Hall, 1999; Hale & Shakoor, 2003; Chen et al., 2004; Ferrick & Gatto, 2005; Yavuz et al., 2006;

Lindqvist et al., 2007).

Strength, porosity, saturation coefficient, water absorption, coefficient of volumetric expansion

and thermal conductivity are the general mechanical properties which are considered to be most important

for weathering. Of these, porosity and strength are fundamental in both chemical and physical weathering

(Gerrard, 1988; Schaetzl & Anderson, 2005).

Rock strength can affect mountain belts in two ways. In fine scales, it controls the landscape via

weathering, but in larger scales, bedrock strength can limit local relief in a mountain range, and once

hillslopes approach a mechanically limited steepness, landsliding will lower ridgelines at the pace set by

the rate of river incision (e.g., Montgomery, 2003). The latter is addressed in the section on topographic

controls on erosion. In this section rock strength and weathering controls are investigated.

Compressive strength of a given rock type varies within a wide range, depending on porosity,

cementation, degree of weathering, formation heterogeneity, grain size angularity, and degree of

interlocking of mineral grains, and finally, orientation of load application with respect to microstructure

(e.g., foliation in metamorphic rocks and bedding planes in sedimentary rocks).

In this respect, the intrinsic variability of intact rock properties causes the association between

strength and rock type to be limited. Uni-axial compression of rock samples in a laboratory is the most

frequently used means of determining rock strength; strength measurements are broadly related to dry

density, and therefore to porosity (e.g., Colback & Wild, 1965; Burshetin, 1969; Goodman, 1989;

Hawkins & McConnell, 1992; Waltham, 2002; Vásárhelyi & Ván, 2006; Pariseau, 2007). Moreover,

saturation state of the pore space has been found to be a major control on rock strength. In many cases,

strength decreases remarkably after only 1% water saturation, but the rate of reduction in strength

depends upon lithology (Burshetin, 1969; Hawkins & McConnell, 1992).



163

6-6-2 Substrate in the Alborz

The Alborz Mountains have a broad range of stratigraphic units from Precambrian to Quaternary,

covering a wide range of lithologies.

The carbonates of the Alborz mainly consist of Paleozoic-Mesozoic limestone and occasionally

Paleocene-early Neogene limestones. They vary from thick, highly fractured, karstic masses to thin

unfractured beds. Carbonate rocks dominate in the eastern Alborz and, together with meta-sediments,

cover a significant area in the north flank of the north-central and west Alborz.

Impermeable sediments and metamorphic rocks with a low porosity are mainly Mesozoic in age

but older, Paleozoic and some Precambrian units crop out in the internal part of the north flank of the

Alborz. This rock class comprises of siltstones, sandstones, conglomerates and low grade metamorphic

rocks, in places alternated with impermeable layers of shale, schist and marl. Together with carbonates,

these rocks dominate the north flank of the Alborz Mountains, and form about 62% of all outcrop in the

integral Alborz. This group includes the Upper Proterozoic Kahar Formation consisting of argillaceous

and siliceous slate with intercalation of quartzite, dolomite (Berberian, 1974; Geological Survey of Iran,

1991b).

Igneous rocks of the Alborz include granites, lava flows, volcanic ash, and tuff and tuffeous

conglomerates and sandstones, and have a wide range of geological ages from Precambrian to

Quaternary. They share geomechanical characteristics with young clastic sedimentary rocks. Rocks of

this class crop out in a wide band across the central-southern part of the range. They are abundant in the

west but feature less in the eastern Alborz. Outcrops of volcanics in the western Alborz join to the

northern Alborz at the westernmost end of the mountain belt, but in the north-east limited outcrops of this

class mainly comprise conglomerate-sandstone.

Evaporites of the Alborz include marls, salts and gypsum of Paleogene and Neogene age. This

class, which coexists with young volcanic rocks, has outcrops in the south central Alborz. In addition,

small outcrops are found in intramontane basins in the west Alborz; in combination with igneous rocks,

they cover 37% of the whole study area. The metapelitic-metavolcanic complex of Late Ordovician

Gorgan Schists (Geological Survey of Iran, 1991a; Ghavidel-syooki, 2007), in the NE Alborz, is

considered volcanic, and therefore classified in this second group.

In Figure 6.6.1, the porosity and strength of these or similar rock types are indicated using

measurements on samples collected elsewhere (after: Pariseau, 2007; Afrouz 1992; Dincer et al., 2004).

Basalt, fine grained limestone, shale and marble have a low porosity (0.3-0.85 %) but high

strength (70-116 MPa). Volcanic tuff, evaporitic rocks, sandstone and andesite have higher porosities,

between 4% and 21.5%, but their strength decreases steadily with increasing porosity from 43 to 27.5

MPa.



164

V. Tuff Evapo. Sand st. Andesite Basalt Lime st. Shale Marble

0

20

40

60

80

100

120

140

Un

co

nfi

ne

d c

om

pre

ss

ive

str

en

gth

(M

Pa

)

0.01

0.1

1

10

100

Po

ro

sity

%

Fig. 6.6.1: Unconfined compressive strength (UCS), and porosity of selected rock types estimated

from a total of 570 measurements (data from: Pariseau, 2007; Afrouz 1992; Dincer et al., 2004). Porosity

is displayed by open circles and UCS by filled circles. Bars show 2σ interval.

6-6-3 Spatial Distribution

The broad outcrop pattern of the principal lithological classes of the Alborz Mountains is shown

in Figure 6.6.2. There are two classes: (I) strong, low porosity rocks, including carbonates, meta-

sedimenatry rocks, and (П) weak, high porosity rocks, including igneous, evaporates and conglomerate-

sandstone. Their outcrop pattern has been obtained from the hydro-geological map of Iran. Each class has

a broad variability in geomechanical properties due to a large degree of lithological heterogeneity,

geological structure, and weathering. Moreover, a higher variability is expected in class (П) due to a large

range of porosity from 4 % to 21.5 %.

Rocks in class (I) dominate the north Alborz, with minor outcrops of rocks in class (II), along the

northern range front and in isolated inliers. The south flank of the mountain belt is dominated by rocks in

class (П) with carbonates of class (I) exposed in some deep incised valleys and in the hanging wall of

main thrust faults. Variability of properties within individual formations is illustrated with an example in

the next section.



165

Fig. 6.6.2: Rock class map of the Alborz showing two main lithological units (compiled from:

Ministry of Energy of Iran, 1983).

6-6-4 Geomechanical Properties of the Karaj Formation

The tuffeous Eocene Karaj Formation is the only formation for which detailed geomechanical

data are available. These data were collected during the construction of a freeway from Tehran to the

Caspian Sea along two S-N tunnels with lengths of 300 m and 6,300 m (Yassaghi & Salari Rad, 2005;

Yassaghi et al., 2005). The Eocene Karaj Formation is a major component of the southern Alborz. The

magmatic sequence of the formation mainly consists of submarine volcano-sedimentary rocks of mid

Eocene age (Dedual, 1967), including green tuffs with volcanic intercalations (Emami, 2000). Along the

tunnelled sections, rocks mainly consist of lithic (up to 45%), vitiric (up to 35%) and crystalline (up to

20%) tuffs (Yassaghi et al., 2005).

Physical and mechanical characteristics of these three types of tuff and other subvolcanic rocks

are listed in Table 6.6.1. The porosity in samples varies from 7.2 to 15.2 %; it is higher in vitric and lithic

tuffs than in crystalline tuffs and andesitic-basalts.

Unconfined compressive strength is higher in vitric and lithic tuff than crystalline tuff and

andesitic-basaltic, and is negatively correlated with porosity (Table 6.6.1).

In order to investigate the effect of water on the rock strength Yassaghi et al. (2005) have also

measured unconfined compressive strength in saturated samples. Their results indicate that tuffs lose

strength by an average of 30% when saturated. The strength reduction is larger for samples with higher

porosity; it reduces from about 25% in the crystalline tuffs to 43% in the vitric ones.

This implies that volcanic tuffs and specially vitric tuffs are likely to weather fast, and prone to

failure during heavy or sustained winter and spring precipitation.



166

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

An

nu

al

ero

sio

n (

mm

)

Proportion of catchment area covered by igneous-evaporitic rock

Rock type Porosity % UCS for natural sample

(MPa)

UCS for saturated sample

(MPa)

Lithic tuff 7.2-11.7 87-110 58-71

Vitric tuff 10.9-15.2 80-105 49-55

Crystalline tuff 9.4-10.5 100-142 78-106

Andesitic-basaltic 9.4-10.5 100-120 -

Tables 6.6.1: Mechanical properties of Karaj Formation (Yassaghi & Salari-Rad, 2005; Yassaghi et al.,

2005).

6-6-5 Substrate and Erosion

Here, the correlation between annual erosion rate and substrate is explored using catchment wide

erosion data and the proportion of the catchment area covered by each lithological class. Data have been

analysed in un-weighted and weighted format. The third statistical method of direct extrapolation is not

applicable for lithological control analyses because it is scale independent (See section 6.2).

A significant correlation exists between mean annual erosion and the fraction of a catchment

covered by igneous, evaporitic and sedimentary rocks (class П) (Fig. 6.6.3 & 6.6.4). Catchments with a

greater proportion of class II rocks have higher annual erosion rates. The correlation is enhanced when

the data is weighted for catchment area with R2 = 0.13 and 0.36 for un-weighted and weighted

approaches, respectively. Similarly, Pearson’s correlation coefficient, an index of similarity, increases

from 0.38 to 0.60 when weighting is applied.

As a complement, Figures 6.6.5 and 6.6.6 show an anti-correlation between mean annual erosion

and the fraction of a catchment covered by carbonates and clastic-metamorphic rocks (class I).

Catchments with a greater proportion of class (I) rocks have lower erosion rates. Again, the statistical

correlation improves when the data are weighted according to catchment size, R2 = 0.20 and 0.36 for un-

weighted and weighted approaches respectively. Pearson’s correlation coefficient is -0.51 and -0.57 for

un-weighted and weighted data, respectively.

Fig. 6.6.3: Annual erosion plotted against proportion of catchment area covered by class П rocks.

Bars show 2σ interval within coverage bins. (A polynomial model fits the data with: y = 0.04x2 + 0.24x +

0.12 has R2 = 0.13)



167

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

An

nu

al

ero

sio

n (

mm

)

Proportion of catchment area covered by igneous-evaporitic rock

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

An

nu

al

ero

sio

n (

mm

)

Proportion of catchment area covered by carbonate-clastic-metamorphic rock

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.0

0.1

0.2

0.3

0.4

0.5

0.6

An

nu

al

ero

sio

n (

mm

)

Proportion of catchment area covered by carbonate-clastic-metamorphic rock

Fig. 6.6.4: Annual erosion plotted against proportion of catchment area covered by rocks of class

П. Data were weighted according to catchment area. Bars show 2σ interval within coverage bins. (A

polynomial model fits the data with: y = 0.03x2 + 0.53x + 0.07 has R

2 = 0.36)


I. Bars show 2σ interval within coverage bins. (A polynomial model fits the data with: y = -0.65x2 +

0.38x + 0.28 has R2 = 0.20)


I., Data were weighted according to catchment area. Bars show 2σ interval within coverage bins. (A

polynomial model fits the data with: y = -0.37x2 - 0.06x + 0.43 has R

2 = 0.36)



168

In summary, of the main lithologies of the Alborz Mountains, porous tuffs and young

sedimentary rocks may be most susceptible to freeze-thaw weathering and chemical weathering. The uni-

axial compressive strength of tuffs of Karaj Formation drops by up to 43% upon saturation according to

measurements (Yassaghi et al., 2005). These volcanic tuffs dominate the southern Alborz, where up to 60

% of annual precipitation falls as snow at high elevations (Moussavi et al., 1989; Mashayekhi &

Mahjoub, 1991), making this area especially susceptible to weathering and disaggregation due to winter

freeze-thaw and spring snow melt and runoff. The north Alborz is dominated, instead by stronger, less

porous rock types. On the scale of the mountain belt, a relatively strong correlation exists between

catchment erosion rates and rock class. Catchments with a larger proportion of weak and porous rocks

systematically have higher erosion rates.


CHAPTER 6: Decadal Erosion Rate Controls-Topography

169

6-7 Topography

6-7-1 Introduction

Attributes of topography may be important controls on erosion. The elevation of the Earth’s

surface represents a potential energy field, but for this energy to result in erosion, a topographic gradient

is required (cf., Strahler, 1950). Many erosion processes have a rate dependence on the local topographic

slope (e.g., Roering et al., 2007). On a global scale, this is reflected in the erosion rates of large river

catchments. Catchments with the greatest elevation difference between headwaters and base level, and/or

the steepest topographic gradient from headwaters to base level have the highest erosion rates (e.g.,

Ahnert, 1970; Summerfield & Hulton, 1994; Harrison, 2000). However, many river catchments are much

larger than the length scale of most erosion processes, and local relief can be used instead to investigate a

link between topography and erosion. Local relief is defined as the elevation difference between the

highest and lowest points in a landscape over a given distance. Thus, local relief is a measure of the

steepness of a landscape. It can be measured from digital elevation models.

Montgomery and Brandon (2002) have estimated catchment-wide average local relief worldwide.

They have found a strong, positive correlation between relief and catchment erosion rates. However, in

areas with very high erosion rates of >1 mm/yr no relation exists between local relief and erosion rate. In

such areas the modal topographic slope is typically the same as the angle of internal friction of the rock

mass (Schmidt & Montgomery, 1995; Burbank et al., 1996; Binnie, 2005; Hovius & Stark, 2006; Lin et

al., 2008), and tectonically oversteepened slopes collapse in bedrock landslides. This is a landscape with

maximum relief, set by the spacing of the major streams, determined in part by the ambient climate, and

the strength of the rock mass (cf., Roering et al., 2005). It is commonly found in active mountain belts.

This section addresses the relation between erosion, and topographic slope and relief in the

Alborz Mountains. The basis of this study is NASA’s Shuttle Radar Topographic Mission (SRTM) digital

elevation model. It has a posted horizontal resolution of 90 m and a vertical accuracy of 6.5 m in Iran

(Kiamehr & Sjoberg, 2005).

6-7-2 Elevation

The elevation of the Alborz Mountains is summarised in Figure 6.7.1. A first order distinction

can be made between the elevation of the north and south flanks of the mountain belt. The boundary

between the two flanks has been defined as the main divide between catchments draining north into the

Caspian Sea, and catchments draining south towards central Iran. The Shah Rud, located to the south of

the ridge pole of the Alborz Mountains, but draining north via the Sefid Rud, has been included in the

south flank statistics. The south flank of the mountain belt merges into the high plateau of central Iran. It

has very little topography below 1 km asl (Fig. 6.7.1).



170

1,000 2,000 3,000 4,000 5,000

0

1,000

2,000

3,000

4,000

Area (

km

2)

Elevation (m)

0

2

4

6

8

10

12

Pro

po

rtio

n o

f tota

l area

(%)

Fig. 6.7.1: Elevation statistics of the south (filled bars) and north (open bars) flank of the Alborz

Mountains. The absolute surface area, and the fraction of total surface area of the mountain belt, located

within a given elevation bin are shown.

The mean elevation of the south flank is 2011 m asl. The north flank of the Alborz Mountains

has a bimodal distribution of elevations with modal peaks at 400-500 and 1600-1700 m asl., and a mean

elevation of 1617 m asl. However, a relatively large proportion of the north flank is located at elevations

significantly above the second mode, and, more importantly, sub-modal elevations have a much higher

frequency than in the south flank. This reflects the structure of the north flank, with a lower frontal range

associated with the north Alborz fault, relatively low mountains, hydrologically connected with the

Caspian Sea in the east Alborz, and a broad piedmont plain. As a result, the elevation of the north flank

has a high coefficient of variation of 0.59.

6-7-3 Slope

From the SRTM DEM, the local topographic slope was calculated at a length scale of 1000 m.

This length scale is similar to the channelisation length scale in many mountain landscapes (Meunier et

al., 2008). The slope at a point was determined as the gradient of the line of steepest descent, determined

with the flow direction setting in ArcMap, in a circle with a radius of 500, centred on that point. The

resulting slope map is shown in Figure 6.7.2. In the Alborz Mountains, the maximum value of the tangent

of the local topography is about 0.9. Very steep slopes are commonly found in the interior of the

mountain belt, often associated with outcrops of competent lithologies such as massive carbonates and

impermeable clastic-metamorphic rocks. It is important to observe that throughout the mountain belt the

local slope varies significantly at sub-catchment length scales.



171

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0

500

1,000

1,500

2,000

2,500

3,000

Are

a (

km

2)

Slope (tangent)

Fig. 6.7.2: Map of local slopes in the Alborz, computed at 1 km lengthscale from a 90 m Digital

Elevation Model (DEM).

The frequency distribution of local topographic slopes is different between the north and south

flanks of the mountain belt. Figure 6.7.3 shows the slope statistics of both domains, binned at intervals

equivalent to ¼ times the standard deviation of the slope frequency distribution in question. The north

flank of the mountain belt has a classic slope frequency distribution with a modal slope of 18.24° (tangent

= 0.33), close to the mean slope, and a quasi normal distribution around this peak. The broad distribution

has a slight positive skew, probably reflecting the range of rock mass strengths in the North Alborz. In

contrast, the slope frequency distribution in the south flank of the mountain belt is extremely positively

skewed, with a modal slope of only 8° (tangent = 0.15), and a much greater mean slope of 15.02° (tangent

= 0.27).

Fig. 6.7.3: Frequency distribution of local topographic slopes in south (filled bars) and north

(open bars) flank of the Alborz Mountains.

Notwithstanding the apparent match of steep slopes and strong rocks in the high mountains, the

modal topographic slopes are significantly lower than the likely angle of internal friction of the rock mass

(see section 6.6) in most of the Alborz. This may be due in part to the length scale of the slope



172

measurements, which is greater than the length scale of many mechanically controlled hillslope segments.

However, it may also imply that the mountain belt has not systematically attained maximum relief. A

weak, although positive, correlation between the mean local slope of catchments and the proportion of the

catchment consisting of carbonates and clastic-metamorphic rocks (R2 = 0.09; Pearson’s correlation

coefficient = 0.31) confirms this notion, but it is not clear in general that the greater steepness of the

northern Alborz has its origin in rock mechanics. Throughout the mountain belt, deep weathering profiles

are found above intact rock mass, together with important accumulations of colluvium. It is likely that

these materials, and their erosion, rather than the properties of the intact parent rock, set the local

topographic slope (cf., Montgomery, 2001).

The control of slope over erosion in the Alborz has been investigated using three statistical

methods, applied to the integral data set as well as separate data for each of the range flanks. Lumped, un-

weighted catchment-wide averages of topographic slope and erosion rate are poorly, if negatively

correlated (R2 = 0.06; Pearson’s correlation coefficient = -0.17; n = 82). Weighting of the data for

catchment size does not improve the correlation (R2 = 0.00004; Pearson’s correlation coefficient = 0.003;

n = 82), nor does direct extrapolation of the data to take into account spatial heterogeneity at the sub-

catchment scale (R2 = 0.001; Pearson’s correlation coefficient = 0.03; n = 2515).

When split, the data show significant differences between the north and south flanks of the

mountain belt. In the north flank, no relation between slope and erosion rate is detected by any of the

three statistical methods (R2 = 0.09, 0.004, and 0.01; Pearson’s correlation coefficient = -0.30, 0.08, and

0.01 for lumped un-weighted, lumped and weighted, and extrapolated data, respectively). However, there

is some indication that local topographic slope and catchment-wide erosion rates are positively correlated

in the south flank of the Alborz Mountains (R2 = 0.07, 0.03, and 0.18; Pearson’s correlation coefficient =

0.27, 0.17, and 0.36 for lumped un-weighted, lumped and weighted, and extrapolated data, respectively)

(Fig. 6.7.4 displays this correlation for the extrapolated data). There is a hint of threshold behaviour in the

south flank data. Erosion rate increases systematically with mean local slope up to slope values of about

15°, but appears to be constant on steeper slopes. This could signal a transition to a limit landscape

dominated by bedrock landslides in weak volcaniclastic rocks (see section 6.6). For comparison, the angle

of internal friction of comparable rocks in the Coastal Range of east Taiwan is 28° (Ramsey et al., 2006).

In the light of the finding in previous sections that the ratio of annual runoff/precipitation, and

vegetation density are controls on erosion, their correlation with slope is explored here (Fig. 6.7.5 &

6.7.6). A significant positive correlation between annual runoff/precipitation and local topographic slope

in south flank of the Alborz does imply a topographic control on erosion in this area. In the north flank,

this control is little pronounced. Pearson’s correlation coefficients for this relation are 0.13 in the north

and 0.85 in the south (Fig. 6.7.5).



173

0.0 0.1 0.2 0.3 0.4 0.5

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

An

nu

al

ero

sio

n (

mm

)

Mean slope (tangent)

0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

An

nu

al ru

no

ff /

an

nu

al

prec

ipit

ati

on


Fig. 6.7.4: Annual erosion rate and mean slope estimated for 2515 data points in the Alborz,

segregated for north (open squares) and south (filled squares) flank. Bars show 2σ interval within slope

bins (A polynomial model fits the data for south flank with: y = -3.82x2 + 2.40x + 0.08; R

2= 0.18).

In the light of the finding in previous sections that the ratio of annual runoff/precipitation, and

vegetation density are controls on erosion, their correlation with slope is explored here (Fig. 6.7.5 &

6.7.6). A significant positive correlation between annual runoff/precipitation and local topographic slope

in south flank of the Alborz does imply a topographic control on erosion in this area. In the north flank,

this control is little pronounced. Pearson’s correlation coefficients for this relation are 0.13 in the north

and 0.85 in the south (Fig. 6.7.5).

Fig. 6.7.5: The ratio annual runoff/precipitation plotted against mean local slope for 95 drainage

basins in the Alborz. Data for north flank (open squares) and south flank (filled squares) have been

separated. Bars show 2σ interval within slope bins. (Polynomial models fit the data for north and south

flanks with: y = 16.07x2 - 5.93x + 0.97; R

2 = 0.09, and y = 1.33x

2 + 2.89x - 0.04; R

2 = 0.73, respectively).

6-7-4 Relief

Local relief was calculated from the SRTM DEM at an arbitrary length scale of 1km to be able to

capture intra-catchment variability. This length scale has no specific physical significance, but it has been

kept constant in order to enable direct comparison of relief across the mountain belt. Figure 6.7.6 is a

relief map of the Alborz Mountains. Maximum local relief in the mountain belt is about 600 m, in strong

carbonates and recent volcanic rocks of the Damavand area. High relief is also found in the crystalline



174

massif of Alam Kuh, and along substantial outcrops of carbonates and clastic-metamorphic rocks in the

interior of the mountain belt, but at the catchment scale, relief and the proportion of these rocks are

essentially uncorrelated (R2 = 0.05; Pearson’s correlation coefficient = 0.09).

Fig. 6.7.6: Distribution of local relief in the Alborz. The local relief was computed within a

500m-radius circle of every grid cell using a 90 m digital elevation model (DEM), and smoothed at 1km

resolution.

In the north flank of the mountain belt, relief frequencies are normally distributed (Fig. 6.7.7),

with modal and mean relief of 260m and 270 m, respectively. In the south flank the equivalent values are

200 m and 214 m, and the frequency distribution is positively skewed. The coefficient of variation of

relief is nearly identical for both flanks, 0.50 to 0.56.

Not surprisingly, local topographic slope and local relief are strongly correlated in the Alborz

Mountains. Figures 6.7.8 shows the mean local slope and mean local relief of areas with a unit size of 15

km2, equivalent to the size of the smallest gauged catchment in the mountain belt. The data can be

described by a relation R = -1411.7S2 + 1343.3S + 76.14, where R is the mean local relief, S is the mean

local slope (R2 = 0.55; Pearson’s correlation coefficient = 0.69). There are very few places in the

mountain belt with local relief of >500 m at 1 km length scale. The equivalent local slope at that length

scale is 26.5°. This is close to mechanical limites to topographic steepness observed elsewhere. However,

most of the topography does not attain this steepness, and may therefore not be at a landslide threshold.

Instead, weathering-limited mass wasting may be the dominant mode of hillslope erosion.



175

0.0 0.1 0.2 0.3 0.4 0.5

0

100

200

300

400

500

Re

lie

f (m

)

Slope (tangent)

0 100 200 300 400 500 600

0

500

1,000

1,500

2,000

2,500

3,000

3,500

Area

(k

m2)

Relief (m)

Fig. 6.7.7: Frequency distribution of local relief over 1 km in the south (filled bars) and north

(open bars) flanks of the Alborz.

Fig. 6.7.8: Mean relief and slope estimated for 2515 cells in the Alborz. Those were sampled

discretely across the range; unit area is computed according to the smallest watershed. Local relief has

been estimated within a 500m-diameter circle of every grid cell. (R2= 0.55; y = -1411.7x

2 + 1343.3x +

76.14). Bars show 2σ confidence interval.

The control of relief over erosion in the Alborz has been investigated using three statistical

methods, applied to the integral data set as well as separate data for each of the range flanks. On the scale

of the mountain belt, lumped, un-weighted catchment-wide averages of topographic relief and erosion

rate are poorly, if negatively correlated (R2 = 0.003; Pearson’s correlation coefficient = -0.04; n = 87).

Weighting of the data for catchment size does not improve the correlation (R2 = 0.004; Pearson’s

correlation coefficient = 0.03; n = 87), nor does direct extrapolation of the data to take into account spatial

heterogeneity at the sub-catchment scale (R2 = 0.02; Pearson’s correlation coefficient = 0.05; n = 2515).

When split the data show significant differences between the north and south flanks of the

mountain belt. In the north flank, no relation between local relief and erosion rate is detected by any of

the three statistical methods (R2 = 0.03, 0.007, and 0.005; Pearson’s correlation coefficient = -0.16, 0.10,

and 0.09 for lumped un-weighted, lumped and weighted, and extrapolated data, respectively). However,

local topographic relief and catchment-wide erosion rates are positively correlated in the south flank of

the Alborz Mountains (R2 = 0.17, 0.07, and 0.29; Pearson’s correlation coefficient = 0.40, 0.23, and 0.53



176

50 100 150 200 250 300 350 400 450 500

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

An

nu

al

ero

sio

n (

mm

)

Relief (m)

for lumped un-weighted, lumped and weighted, and extrapolated data, respectively). Figure 6.7.9 shows

this correlation for extrapolated data.

Fig. 6.7.9: Annual erosion rate and mean local relief for 2515 data points in the Alborz. Data for

north (open squares) and south (filled squares) have been segregated. Bars show 2σ interval within relief

bins (Polynomial models fit the data for south and north flanks with: y = -2E-06x2 + 0.002x - 0.01; R

2 =

0.29, and y = 2E-06x2 - 0.001x + 0.29; R

2 = 0.06, respectively).

Given the role of the ratio of annual runoff/precipitation as a control on erosion, it’s correlation

with relief is explored here (Fig. 6.7.10). As for local topographic slope, a significant positive correlation

is found between annual runoff/precipitation and local relief in south flank of the mountain belt, but not in

the north (R2 and Pearson’s correlation coefficient are 0.87 and 0.93, and 0.16 and 0.13 for the south and

north flank, respectively).

In conclusion, mean annual erosion rates appear not to be controlled by local slope and

topographic relief at the scale of the mountain belt. However, within the southern Alborz, erosion is

significantly, and positively correlated with both these topographic characteristics. This may be due, in

part, to the effect of topographic slope on the runoff/precipitation ratio, which results in a more efficient

translation from rainfall to surface flow in steeper terrain. Another cause may be the concentration of

vegetation and biomass in parts of the landscape with the lowest topographic slopes and relief. This

tendency is visible both in the south and the north flank of the mountain belt (Fig. 6.7.11). If it is a control

on erosion, then it has been overridden by other factors in the northern Alborz, where topographic

characteristics do not correlate meaningfully with catchment erosion.



177

100 150 200 250 300 350 400

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

An

nu

al ru

no

ff / a

nn

ual p

recip

ita

tio

n

Mean relief (m)

0.10 0.15 0.20 0.25 0.30

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

An

nu

al E

VI


Fig. 6.7.10: The ratio of annual runoff/precipitation plotted against mean relief for 95 watersheds

in the Alborz. Data for north (open squares) and south (filled squares) have been segregated. Bars show

2σ interval within relief bins (Polynomial models fit the data for south and north flanks with: R2 = 0.87; y

= 3E-06 x2 + 0.001x - 0.01 & R

2 = 0.16; y = 1E-05 x

2 - 0.005x + 1.05, respectively).

Fig. 6.7.11: Annual EVI plotted against mean slope for 95 watersheds in the Alborz. Data for

north (open squares) and south (filled squares) have been segregated. Bars show 2σ interval within relief

bins (Polynomial models fit the data for south and north flanks with: R2 = 0.47; y = 1.20 x

2 - 0.71x + 0.18

& R2 = 0.28; y = -1.80 x

2 - 0.09x + 0.37, respectively).


CHAPTER 6: Decadal Erosion Rate Controls-Stream Power

178

6-8 Stream power

6-8-1 Introduction

Stream power provides an expression of the rate of energy expenditure at a given point in a river

system and is inherently linked to sediment transport competence, and the ability of the stream to perform

geomorphic work on its bed. The concept was introduced by Gilbert (1914), and developed by Rubey (1933),

Knapp (1938), Velikanov (1954), and Bagnold (1956, 1966). The expression for total stream power is:

Ω=γQS, (6.8.1)

where Ω is total stream power per unit length of channel (W m−1

), γ is the specific weight of water (9807

N/m2), Q is discharge (m

3/s) and S is the energy slope of the stream. Total stream power is calculated over the

entire width of the channel. It does not specify how much energy is available per unit bed area. Specific

stream power provides a measure of the rate of energy expenditure per unit area of channel bed and is

expressed as:

ω=Ω/W, (6.8.2)

where ω is specific stream power (W/m2) and W is the width (m) of the active channel. Downstream

hydraulic geometry relations established for a range of rivers indicate that at bankfull flow, channel width is

approximately proportional to the square root of discharge (Leopold & Maddock, 1953; Knighton, 1996;

Church, 2002; Finnigan et al., 2005).

I have calculated unit stream power in the Alborz Mountains, using the SRTM DEM with a posted

resolution of 90 m to measure local slope along the line of steepest descent in a 3 x 3 grid cell area, and total

annual discharge at a point calculated by interpolation of runoff measurements discussed in section 6.5. From

high-resolution (<15 m) satellite images, channel width was estimated at 14 stations in the central west

Alborz (Fig. 6.8.1). At these stations, the relationship between channel width and mean daily discharge can

be described by a power law (R2 = 0.49, Pearson’s correlation coefficient = 0.70):

W = 23Q0.56

, (6.8.3)

in good agreement with the general model described above.



179

1 10

10

100

41-115

17-041

15-017

15-011

15-023

41-121

14-02141-109

41-139

41-143

41-117

15-001

15-005

15-007

Ch

an

nel w

idth

(m

)

Mean daily discharge (m3/s)

Fig. 6.8.1: River channel width and mean annual discharge in the Alborz (R2= 0.49; Pearson’s

correlation coefficient = 0.70). The solid line is a nonlinear least-square fit to the data. Widths were measured

as the distance between banks discernible on high resolution satellite images (<15m)

6-8-2 Spatial-Temporal Distribution

Figure 6.8.2 shows the distribution of unit stream power across the Alborz Mountains, with local

values ranging from 0.1 W/m2 to 4.5 W/m

2. High values are found in areas with high runoff and steep slopes

in the north Alborz, especially in catchments receiving snow melt discharge from the high mountains of Alam

Kuh and Damavand. Notably, the small catchments in the northern range front of the mountain belt receive

most precipitation, but have relatively low topographic slopes, subduing local unit stream power values.

Lower values of <2.5-3 W/m2 are characteristic of the drier southern Alborz.

Fig. 6.8.2: Unit stream power in the Alborz, derived from slope map and interpolation of daily

discharge measurements at 81 hydrometric stations operated by TAMAB. The map was smoothed at

catchment scale using a circular moving mean with 15km diameter, and is displayed at 1 km spatial

resolution.



180

0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

An

nu

al

str

ea

m p

ow

er

(W/m

2)

Slope (tangent)

0 200 400 600 800 1000

0.5

1.0

1.5

2.0

2.5

3.0

3.5

An

nu

al

str

ea

m p

ow

er

(W/m

2)

Annual runoff (mm)

Both total annual runoff and local topographic slope are strong controls on the pattern of unit stream

power in the Alborz Mountains. Unit stream power increases systematically with annual runoff up to the

values of about 600 mm (Fig. 6.8.3), typically found in the south flank of the mountain belt, but decreases at

higher discharges. Topographic slope is strongly and linearly correlated with unit stream power (Fig. 6.8.4),

with R2 = 0.56 and Pearson’s correlation coefficient = 0.76. In the north flank of the Alborz Mountains, slope

and runoff are essentially uncorrelated, but in the south flank steeper slopes have higher runoff rates (section

6-7).

Fig. 6.8.3: Average annual unit stream power and runoff for 91 watersheds in the Alborz. Bars show

2σ interval within runoff bins. (A polynomial model fits the data with: y = -2E-06x2 + 0.004x + 0.68; R

2 =

0.22)

Fig. 6.8.4: Average annual unit stream power and mean local slope for 87 watersheds in the Alborz.

Bars show 2σ interval within slope bins. (A linear model fits the data with: y = 11.94x - 0.79; R2 = 0.56)

Across the Alborz Mountains, stream power is negatively correlated with vegetation. In both flanks

of the mountain belt, unit stream power is high, on average 2.25-2.75 W/m2 where EVI<0.10, but the rate of

decrease of unit stream power with increasing EVI is greater in the south flank than in the north flank (Fig.

6.8.5). Given that EVI is positively correlated with precipitation, and therefore runoff (section 6.5), the trends



181

0.16

0.20

0.24

0.28

0.32

0.36

0.40

0.44

0.16

0.20

0.24

0.28

0.32

0.36

0.40

0.44

0.5 1.0 1.5 2.0 2.5 3.0 3.5

No

rma

lize

d s

ea

so

na

l s

tre

am

po

we

r

Spring

Summer

Autumn

Winter

Annual stream power (W/m2)

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

An

nu

al str

eam

po

wer (

W/m

2)

Annual EVI

observed in Figure 6.8.5 imply a strong negative correlation of EVI and topographic slope, where vegetation

density is greatest on gentle slopes (see section 6.7).

Fig. 6.8.5: Average annual unit stream power and mean annual EVI for 76 watersheds in the Alborz.

Data for the south (filled squares) and north (open squares) flank have been segregated. Bars show 2σ interval

within EVI bins. (Polynomial models fit the data with: y = 28.02x2 - 18.09x + 4.17; R

2 = 0.29, and y = -

232.85x2 + 26.28x + 1.23; R

2 = 0.26, for north and south, respectively).

As a consequence of the seasonality of precipitation and runoff (sections 6-3 and 6-5, respectively),

stream power varies through the year in the Alborz Mountains. Due to the constant contribution of local

topographic slope, the seasonality of stream power is subdued, but 32% of the cumulative annual stream

power is achieved in spring, 29% in winter, 25 % in autumn, and 14% in summer.

Across the mountain belt spring is consistently the season with the greatest cumulative stream power,

but where the total annual stream power is smallest, the winter component outweighs the summer and autumn

components, whereas winter has the smallest relative contribution where total annual stream power is greatest

(Fig. 6.8.6).

Fig. 6.8.6: Normalized average seasonal and annual unit stream power for 91 watersheds in the

Alborz. Bars show 2σ interval within stream power bins.



182

0 1 2 3 4

0.0

0.2

0.4

An

nu

al

ero

sio

n (

mm

)

Unit stream power (W/m2)

6-8-3 Stream power and Erosion

Three statistical approaches have been taken to explore the stream power control on erosion in the

Alborz Mountains. Lumped, un-weighted analysis reveals no relation between catchment average values of

total annual unit stream power and catchment erosion rates (R2 = 0.0002; Pearson’s correlation coefficient =

0.04). Direct extrapolation of data gives a similarly poor correlation of the two parameters (R2 = 0.007;

Pearson’s correlation coefficient = 0.04), but weighting of the data according to catchment size yields a very

weak positive correlation of stream power and erosion rate (R2 = 0.04; Pearson’s correlation coefficient =

0.2), dominated by larger catchments (Fig. 6.8.7).

Fig. 6.8.9: Annual erosion rate and average annual unit stream power for 81 drainage basins in the

Alborz. Each data point is weighted based on the catchment area. Error bars show 2σ interval within stream

power bins.

Therefore, it must be concluded that average annual unit stream power is not a first order

control on the ongoing erosion of the Alborz Mountains. This surprising and important finding

implies that topographic and runoff effects on erosion cancel each other on the scale of the mountain

belt, and/or that other controls override the effect of stream power.

Within the physiognomically and lithologically more homogeneous domains of the northern

and southern flanks of the mountain belt, some effects of stream power can be discerned. In the

southern Alborz, annual erosion rates are positively correlated with average annual stream power (R2

= 0.17; Pearson’s correlation coefficient = 0.40) (Fig. 6.8.10), but in the northern Alborz no significant

correlation has been found (R2 = 0.03; Pearson’s correlation coefficient = -0.06).

The relation between catchment erosion and available stream power can also be considered

on a season-by-season basis. The result, shown in Fig. 6.8.11, is similar to that in Fig. 6.5.22 for

erosion and runoff. This is not surprising, given that runoff is a principal term of stream power.

There is a broad trend of increasing seasonal erosion with increasing seasonal stream power, from

which individual seasons deviate. In winter, the proportion of erosion is independent of the



183

0.10 0.15 0.20 0.25 0.30 0.35 0.40

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 Spring

Summer

Autumn

Winter

Frac

tio

n o

f an

nu

al ero

sio

n

Fraction of annual stream power

0.5 1.0 1.5 2.0 2.5 3.0

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

An

nu

al e

ro

sio

n (

mm

)

Annual stream power (W/m2)

proportion of total annual stream power, and in spring and autumn areas with high seasonal stream

power have disproportionally high seasonal erosion rates. The important implication is that other

factors than stream power or its terms suppress erosion in winter, and elevate erosion in spring and

autumn.

Fig. 6.8.10: Average annual erosion and unit stream power for 81 drainage basins in the Alborz. Data

for the south (filled squares) and north (open squares) flank have been segregated. Bars show 2σ interval

within stream power bins. (Polynomial models fit the data with: y = 0.03 x2 - 0.15x + 0.33, y = 0.06 x

2 -

0.04x + 0.20 for north and south, respectively).

Fig. 6.8.11: Proportion of annual erosion and annual unit stream power for 73 watersheds in the

Alborz. Bars show 2σ interval within fraction of stream power bins.

The relation between catchment erosion and available stream power can also be considered

on a season-by-season basis. The result, shown in Fig. 6.8.11, is similar to that in Fig. 6.5.22 for

erosion and runoff. This is not surprising, given that runoff is a principal term of stream power.

There is a broad trend of increasing seasonal erosion with increasing seasonal stream power, from

which individual seasons deviate. In winter, the proportion of erosion is independent of the

proportion of total annual stream power, and in spring and autumn areas with high seasonal stream

power have disproportionally high seasonal erosion rates. The important implication is that other



184

factors than stream power or its terms suppress erosion in winter, and elevate erosion in spring and

autumn.


CHAPTER 6: Decadal Erosion Rate Controls-Discussion

185

6-9 Discussion

Table 6.9.1 and figure 6.9.1 summarise the strength of precipitation, runoff, vegetation density, substrate

properties, topographic attributes and stream power as controls over catchment erosion in the Alborz

Mountains. In this table, correlation strengths are expressed as r-squared values, given for data covering the

integral Alborz, and data covering the north or south flank of the mountain belt only.

Table 6.9.1: R-squared values for polynomial and linear fit (between annual erosion rate and its controls).

Erosion driving control r-squared value (polynomial fit) r-squared value (linear fit)

Precipitation (total) 0.01 0.01

Precipitation (N) 0.08 0.08

Precipitation (S) 0.07 0.03

Runoff (total) 0.06 0.04

Runoff (N) 0.12 0.12

Runoff (S) 0.10 0.004

Runoff/ Prec.(total) 0.08 0.12

Runoff/ Prec.(N) 0.08 0.07

Runoff/ Prec.(S) 0.35 0.32

Vegetation (total) 0.13 0.0009

Vegetation (N) 0.17 0.10

Vegetation (S) 0.22 0.16

Lithology (class I) 0.20 0.26

Lithology (class I) (N) 0.24 0.29

Lithology (class I) (S) 0.37 0.00005

Lithology (class II) 0.13 0.14

Lithology (class II) (N) 0.10 0.15

Lithology (class II) (S) 0.15 0.0002

Slope (total) 0.06 0.04

Slope (N) 0.13 0.09

Slope (S) 0.07 0.14

Relief (total) 0.003 0.002

Relief (N) 0.03 0.02

Relief (S) 0.17 0.16

Stream power (total) 0.0002 0.002

Stream power (N) 0.03 0.004

Stream power (S) 0.17 0.16

The first observation from the table and figure (6.9.1) is that no single variable explains a major part

of the observed pattern of average annual erosion, neither on the scale of the mountain belt, nor within

individual range flanks. However, two variables explain aspects of the erosion pattern at the scale of the

mountain belt. They are lithology (specifically the presence of relatively strong sedimentary and

(meta)sedimentary rocks within the catchment), and vegetation density, and of these two lithology appears to

be the stronger control (R2 = 0.26, and R

2 = 0.13, respectively). Interestingly, both these variables set the

erodibility of the catchment surface, rather than the erosivity of the processes that mobilise material. The

importance of erodibility is confirmed by the fact that the presence of hard rocks in the catchment, rather than



186

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Pre

cip

itatio

n (to

tal)

Pre

cip

itatio

n (N

)

Pre

cip

itatio

n (S

)

Ru

no

ff (tota

l)

Ru

no

ff (N)

Ru

no

ff (S)

Ru

no

ff/pre

cip

itatio

n (to

tal)

Ru

no

ff/pre

cip

itatio

n (N

)

Ru

no

ff/pre

cip

itatio

n (S

)

Slo

pe

(tota

l)

Slo

pe

(N)

Slo

pe

(S)

Re

lief (to

tal)

Re

lief (N

)

Re

lief (S

)

Stre

am

po

we

r (tota

l)

Stre

am

po

we

r (N)

Stre

am

po

we

r (S)

Ve

ge

tatio

n (to

tal)

Ve

ge

tatio

n (N

)

Ve

ge

tatio

n (S

)

Stro

ng

su

bstra

te (c

lass I)

Stro

ng

su

bstra

te N

(cla

ss I)

Stro

ng

su

bstra

te S

(cla

ss I)

We

ak s

ub

stra

te (c

lass II)

We

ak s

ub

stra

te N

(cla

ss II)

We

ak s

ub

stra

te S

(cla

ss II)

R-s

qu

are

d v

alu

e

polynomial fit linear fit

the presence of soft rocks correlates with erosion rates. Weak, porous rocks are more erodible, making it

more likely for factors setting erosivity to dominate the erosion pattern.

Fig. 6.9.1: R-squared value for the polynomial and linear fit between annual erosion and factors

which control the annual erosion (See Table 6.9.1).

Substrate and surface strength are the only demonstrable control on the pattern of erosion on the scale

of the mountain belt. Factors commonly thought to set erosion patterns, including precipitation, runoff,

topographic slope and stream power, do not act notably as controls on this scale. Their effect is masked by

the effects of rock strength and weatherability, and vegetation cover.

Lithologically, two domains can be distinguished within the Alborz Mountains. One is dominated by

the relatively strong rocks (meta)sedimentary rocks mentioned above. The other is dominated by weaker,

more weatherable volcanics, volcaniclastics, conglomerates and evaporites. The boundary between domains

runs approximately parallel to the main divide, such that the first rock type builds the north flank of the

Alborz Mountains, and the second rock type builds the south flank. These domains coincide to first order

with the Hyrcanian forest province in the northern Alborz and the Irano-Touranian arid to semi-arid flora of

the southern Alborz. Thus, there is a major zonation of erodibility within the mountain belt, between a highly

erodible southern zone, and a less erodible northern zone. The boundary between these zones is located

around the main divide of the mountain belt. Within each domain, and especially in the south flank of the

Alborz, other controls on erosion emerge from the statistics.



187

Average annual runoff (R2 = 0.10), the ratio of annual runoff/precipitation (R

2 = 0.35), local relief (R

2

= 0.17), and stream power defined as the product of runoff and local slope (R2 = 0.17), all account

meaningfully for an aspect of the pattern of erosion observed within the south flank of the mountain belt.

They add to the base effects of lithology, notably the presence of strong rocks in the catchment (R2 = 0.37)

and vegetation cover (R2 = 0.22), which continue to dominate on this sub-regional scale. At this scale, the

ratio of runoff/precipitation emerges as a strong control on erosion, and a high runoff efficiency typically

translates into a high erosion rate.

Both runoff and precipitation are highly seasonal throughout the Alborz, and especially in the south

flank of the mountain belt, as is the density of vegetation. In fact, runoff peaks in spring when snow melt

adds to seasonal rains, vegetation cover is low, and the long season of freeze-thaw weathering comes to a

close. This is when erosion rates are highest in the southern Alborz, and maximum erosion rates are found

where the topographic relief and associated stream power are greatest.

In the north flank of the mountain belt, where strong sedimentary and meta-sedimentary rocks

dominate, the effects of the drivers of erosion are more obscured, but the broad conclusions from analysis of

data from the south flank do apply here too. However, the seasonality of erosion is different in the north,

with an emphasis on fluvial sediment transport in autumn in the wettest parts of the mountain belt. I have

attributed this to a combination of high infiltration and evapotranspiration rates with optimal vegetation cover

in summer, which prevent sediment mobilised during rain storms from travelling far down river channels, and

to a remarkable base flow into rivers in the northern Alborz in autumn, which enhances their capacity to

transport clastic loads.

In the final chapter of this thesis, I shall use these findings to explain the long-lived pattern of exhumation of

the Alborz Mountains, and explore consequences for the structural and topographic evolution of this and

other mountain belts.


CHAPTER 6: Decadal Erosion Rate Controls - Introduction 118m.rezaeian/thesis/eroison...

Documents

Transcript of CHAPTER 6: Decadal Erosion Rate Controls - Introduction 118m.rezaeian/thesis/eroison...