International Centre for Integrated Mountain Development

Kathmandu, Nepal

Agro-Ecosystem Monitoring in the HKH region

Faisal M. Qamer

United Nations/Pakistan International Workshopon Integrated Use of Space Technologies for

Food and Water Security, Islamabad, Pakistan

Use of Space Technology and Geospatial Applications in SERVIR programme

Cryosphere / Water

Agriculture andFood Security

Air / Atmosphere

Disaster / Natural Hazards

Ecosystem / Biodiversity

Sustainable Mountain

Development

Interdependent and interlinkages

Common toolsand approaches

Capacity building and networking

Operational information services

Why do we need agricultural monitoring?

Source: USDA ERS

Monthly Average Crop Price Index: 2002-2010 Net

Surplus

Net

Deficit

Annual wheat production minus consumption

Hotspots of Climate Change and Food Insecurity

Source: CGIAR, 2011

HKH region inhabits ~ 210 million people and around 80% of the population are engaged in various land based activities

Food Monitoring and Early Warning Systems

GIEWS - FAO Global Information and Early Warning SystemFEWS Net - USAID Famine Early Warning SystemGMFS – Global Monitoring for Food securityVAM – World Food Programme Vulnerability Analysis and MappingMARS FOOD - Monitoring Agriculture with Remote Sensing (EC/JRC)EARS - Environmental Analysis and Remote SensingDMC - Drought Monitoring Centers (SADC/IGAD) in East Central AfricaSource: GMFS

• Establish past and present status of agriculture conditions in the HKH areas

• Develop spatially referenced socioeconomic data to characterize food security and agriculture production

• Operational Agriculture Monitoring System

Improving knowledge of agriculture production using remote sensing and GIS technologies to support food security analysis in the Himalayan region

Agriculture Monitoring for Food Security in the HKH

Remote Sensing in Agro Ecosystem Monitoring

Vegetation Condition Monitoring

Pettorelli etal 2005

Greenness Index

NDVI: A measure of how much photosynthetically active vegetation is present.

Intra-annual Phenology in Tarai, Nepal

NDVI

Low

High

Compilation of bi-monthly data for last 11 years

MODIS Smoothed and Gap-filled Product at 250m resolution

MODIS Land Product: Mod13Q1Automated downloading script in R

Time-series Data Data Quality

Fourier Curve FitENVI/IDL Script

Weights

Adjust Weight from First Attempt (reduce noises)

Time Series Smoothed Results of 11 Years data

Save Original High Quality Data for processing validation

Landsat image based land cover maps (arable land mask) / Google Earth Images

MODIS Smoothed and Gap-filled Product at 250m resolution

Neighbor Pixels with Same land cover/crop Type

Best Available Inter annual Curve (first neighbor

then typical curve)

Typical Smoothed Curve for each Crop type to identify SoS , Amplitude and

EoS

Data Processing Validation

Mask Crop Area on SoS Image, based on threshold value derived from

seasonal curve

local (per pixel) NDVI mean 10 years (2001-2010)

Start of season assessment of Crop Acreage

Crop growth conditions monitoring and End of Season assessment

Long term trend analysis / (Agro) Ecosystem vulnerability tracking

Data processing framework for (Agro) Ecosystem monitoring using moderate resolution hyper-temporal satellite data

Monotonic trend analysis(persistence mapping)

Deviation of current NDVI (2012 ) condition from mean (anomaly map on every 16th day)Climate anomalies VS vegetation

anomalies

Season Integrated NDVI anomaly maps at the EoS

Climate DataSocio-economic data

Climate anomalies VS vegetation anomalies

Household vulnerability VACA survey data

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Jan Feb March April May June July Aug Sep Oct Nov Dec

Processed

Original

Original Processed

Fourier transformation of poor pixel (given in MOD QA) applied in ENVI/idl.

MODIS Smoothed and Gap-filled Product at 250m resolution

10-25 June 2012

10-25 June 2011

0

50

100

150

200

250

Jan Feb Mar Apr May Jun

2011

2012

0

100

200

300

400

Jan Feb Mar Apr May Jun

2011

2012

Rainfall at Chatra- Koshi

Rainfall at Garuida

pre

cip

itat

ion

Grey area is non- agriculture land

NDVI

Low

High

Delay in summer rain and its impact on Maize crop

25 June 2012

Normal (stable)

Better than normal

Worst than normal

Anomaly ( Z Score)

Delay in summer rain and its impact on Maize crop

Comparison with mean of last 10 years

25 June 2012

25 June 2011

Z = (Current - mean) / stdev.

Koshi basin

Average yield across eco-zones (10 years average)

0

1000

2000

3000

4000

5000

6000

7000

8000

Mountain Hills Tarai

Rice

Maize

Wheat

Rice16% Maize

8%

Wheat6%

other70%

15000

20000

25000

30000

35000

Mountains Hills Tarai (Plains)

Rice

Maize

Wheat

Agriculture areas across Koshi basin

Agriculture areas across eco-zones

km2

Kg/

ha

Variance Analysis across districts and Years for the yield of three major cereal crops

36 Districts and 11 Year crop production data

How this understanding could be disaggregated both spatially and temporally to understand vulnerability?

Will this understanding helps to study water requirements, crop management strategies , socioeconomic resilience?

(Districts)

(Years)

(Districts)

(Years)

0.5

0.55

0.6

0.65

0.7

0.75

2001 2003 2005 2007 2009 2011

Crop season (Dec. – Apr) Max. NDVI

40,000

45,000

50,000

55,000

60,000

65,000

2001 2003 2005 2007 2009 2011

Production

y = 91214x - 3830.8R² = 0.6822

Relationship of Wheat Production and NDVI in Sarlahi District of Nepal Inter-annual changes in wheat crop production and its spatial pattern

ND

VI

Greenness Index (NDVI) as proxy to Agriculture productivity

40,000

45,000

50,000

55,000

60,000

0.5 0.55 0.6 0.65 0.7

Inter-annual changes in wheat crop production and its

spatial pattern2005

2006

2007

2008

Normal (stable)

Better than normal

Worst than normal

ANOMALY (Z Score)

Grey :forested areas , Blue : water bodies

Z = (Current - mean) / stdev.

40,000

50,000

60,000

2001 2003 2005 2007 2009 2011

Production

8,000

10,000

12,000

14,000

2001 2003 2005 2007 2009 2011

5,000

6,000

7,000

8,000

2001 2003 2005 2007 2009 2011

Mo

no

ton

ic t

ren

d a

cro

ss K

osh

i bas

in

30,000

40,000

50,000

60,000

2001 2003 2005 2007 2009 2011

20,000

30,000

40,000

50,000

60,000

2001 2003 2005 2007 2009 2011

Stable

-ve

+ve

Maize Maize

Wheat Wheat

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

WheatMaize Rice

Phenometrics

17 Dec

25 Dec

02 Jan

10 Jan

18 Jan

26Jan

03 Feb

15 Mar

23 Mar

31 Mar

8 Apr

16 Apr

24 Apr

2 May

10 May

18 May

Map of Start of Season (SoS)

for wheat crop in 2011

0-Jan

5-Jan

10-Jan

15-Jan

20-Jan

25-Jan

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Average SoS for Wheat crop during last 10 Years in Sarlahi District

65

70

75

80

85

90

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

48

72

80

88

104

112

128

144

352

Average LoS for Wheat crop during last 10 Years in Sarlahi District Map of Length of Season (LoS)

for wheat crop in 2011

Pattern and processes

• NDVI based productivity

• Start of growing season

• Length of growing season

Response Variables

• Rainfall (W, SP,SM,F)

• Temperature (W, SP,SM,F)

• Radiation

• Evapotranspiration

• Water Stress

• Growing Degree Days

• Soil nutrition

• Topography

• Social structure

Explanatory Variables

W: Winter, SP: Spring, SM: Summer, F: Fall

Timesat parameter setup for

Savitzky-Golay filter

11 years of 16-day MODIS NDVI

11 years monthly precipitation/ temperature

Pastoral dynamicsEcosystem dynamics

Smoothed NDVI curves

Start of season

Length of Season

Season Max. NDVI

Season Integ. NDVI

11

ye

ars

ann

ual

ph

en

om

etr

ics

Decision rules for rangelands

identification

Decision rules for ranking rangelands

productivity

Rangelands Productivity map

NDVI regression on time

Intra-annual vegetation trend

Relation b/w NDVI and precipitation

Monotonic trend map

Rangelands productivity

calendar

Rangelands productivity

calendar map

Impedance Map

Understanding rangelands dynamics in relation to climate variability in the Upper Indus Basin

On going MS Thesis of Mr. Sawaid Abbas, AIT, Bangkok. Under the HICAP programme of ICIMOD

Identification of Rangelands

Rangelands Productivity in the Upper Indus Basin

Monotonic Trend of grasses/shrubs during last 11 years

The analysis is based on 16 days composite MOD13Q1 NDVI product.In total 253 images, covering 11 years, were analyzed.

P-Value of Monotonic Trend of grasses/shrubs

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0

50000

100000

150000

200000

250000

300000

350000

P V

alu

es

Are

a i

n h

ecta

res

Elevaton Range in meters

- 0.8 to - 0.7

- 0.7 to - 0.6

- 0.6 to - 0.5

- 0.5 to - 0.4

- 0.4 to - 0.3

- 0.3 to - 0.2

- 0.2 to - 0.1

0.1 to 0.2

0.2 to 0.3

0.3 to 0.4

0.4 to 0.5

0.5 to 0.6

0.6 to 0.7

0.7 to 0.8

0.8 to 0.9

- 0.1 to 0.0

0.0 to 0.1

Mean P Value

Elevation

-30000

-20000

-10000

0

10000

20000

30000

40000

50000

90

110

130

150

170

190

LOS

SOS

LINT

RF*10000

Day

of

year

Seaso

nal In

tegrated

ND

VI

Elevation

Thank you