Water Productivity Mapping (WPM) at various Resolutions (scales) using Remote Sensing

WPMSSM &

Water Productivity Mapping (WPM) at various Resolutions (scales) using Remote Sensing

A proof of Concept Study in the Syr Darya River Basin in Central Asia

Cotton wet biomass (WBM)

WBMcotton = 71.18*(IRS TBVI32)3.96

R2 = 0.834

0

2

4

6

8

10

12

0.1 0.2 0.3 0.4 0.5 0.6 0.7IRS TBVI32

co

tto

n w

et

bio

ma

ss

2006

2007

Cotton dry biomass (DBM)

DBMcotton = 51.59*(IRS TBVI32)4.86

R2 = 0.821

0

1

2

3

4

5

6

7

0.1 0.3 0.5 0.7IRS TBVI32

co

tto

n d

ry b

iom

as

s

20062007

Cotton yield

Yieldcotton = 5.156*IRS NDVI - 0.964

R2 = 0.753

0.0

0.5

1.0

1.5

2.0

2.5

0.25 0.3 0.35 0.4 0.45 0.5 0.55IRS-NDVI (Sept 4, 2007)

fie

ld-a

ve

rag

e-y

ield

2007

Cotton Leaf Area Index (LAI)

LAIcotton = 10.37*(IRS TBVI31)1.915

R2 = 0.725

0

1

2

3

4

5

0.1 0.2 0.3 0.4 0.5 0.6IRS TBVI31

co

tto

n L

AI

20062007

Water Productivity Mapping and Soil Salinity Mapping Workshop, Tel Aviv, Israel, May 13-15, 2008

Xueliang Cai1, Prasad S. Thenkabail1, Alexander Platanov1, Chandrashekhar M. Biradar2, Yafit Cohen3, Victor Alchanatis3, Naftali Goldshlager4, Eyal Ben-Dor5, MuraliKrishna Gumma1, Venkateswarlu Dheeravath1, and Jagath Vithanage1

1= International Water Management Institute (IWMI), Sri Lanka; 2 = University of New Hampshire, USA; 3 = Institute of Agricultural Engineering, ARO, Israel; 4 = The Soil Erosion Research Station, Ministry of Agriculture, Israel; 5= Tel Aviv

University, Israel

0 2 41 km

IRS

0 2 41 km

QB

kg/m3 kg/m3

WPMSSM &

Need, Scope, Context for

WPM

WPMSSM &

• World Food Crisis: The world is suddenly facing a food crisis not known to it since 1970s and many have already pronounced that the era of cheap food is over;

• Need to produce more food: The world population is growing at nearly 100 million per year, irrigated and rainfed croplands that increased so rapidly in the 1970s to 2000s have almost stagnated, the era of “green revolution” has ended, populations of emerging markets are consuming more, croplands are diverted to, settlements, bio-fuels, and water is becoming limited for croplands with increasing demands from ever increasing urbanization and industrialization. Whereas, more food needs to be produced to address these issues, it is widely accepted that increasing the extent of croplands and utilizing more and more water is neither sustainable nor environmentally acceptable;

• Uncertainty from climate change yet to be considered: To adopt to CC and cope with water scarcity, agricultural sector has to enhance the productivity of land and water resources; and

• How to produce more food with same land and water: This implies an urgent need to increase water productivity (“more crop per crop”) leading to a “blue revolution” that will produce more food with existing levels of croplands and water.

Water Productivity Mapping using Remote Sensing (WPM)

Need, Scope, Context

WPMSSM &

Goals

WPMSSM &

The overarching goal of this study was to develop methods and protocols for agricultural crop water productivity mapping using satellite sensor data. The specific objectives were to produce:


the Goals

1. Crop productivity maps (CPMs);2. Crop Water use maps (WUMs);3. Water productivity maps (WPMs).

WPMSSM &

Definition of WP

WPMSSM &

Water productivity is “the physical mass of production or the economic value of production measured against gross inflow, net inflow, depleted water, process depleted water, or available water” (Molden, 1997, SWIM 1). It measures how the systems convert water into goods and services. This study calculated WP using equation:

Water productivity maps were produced by dividing crop productivity maps by water use maps.

)/m(m (ETa) useWater

)(kg/moductivityPrCrop)(kg/moductivityPrWater

23

23


Definition

WPMSSM &

Study Area

WPMSSM &

AralSea

Toktogul

Kirgizstan

Galaba Kuva

Kazakhstan

Tajikistan

Lakes

Basin boundaries

Administrative provinces

Rivers

Canals

# Test sites

Uzbekistan

Irrigated area

Water Productivity Mapping using Remote Sensing (WPM) The Syr Darya River Basin

Source: IWMI, 2007

WPMSSM &

AralSea

Toktogul

Kirgizstan

Galaba Kuva

Kazakhstan

Tajikistan

Lakes

Basin boundaries

Administrative provinces

Rivers

Canals

# Test sites

Uzbekistan

Irrigated area

GalabaKuva

Kuva

Scattered GT points in selected farm plots at Galaba study site

Scattered GT points in selected farm plots at Kuva study site

Water Productivity Mapping using Remote Sensing (WPM) Study Areas: Galaba and Kuva in the Syr Darya River Basin

Photo Credit: Alexander PlatanovPhoto Credit: Alexander Platanov

WPMSSM &

Dataset collection Dates: Satellite Sensor data and

Groundtruth data

WPMSSM &

Water Productivity Mapping using Remote Sensing (WPM) Data Collection dates: Satellite sensor data and Groundtruth Data

Source: IWMI, 2007

WPMSSM &

Characteristics of Satellite Sensor Data

WPMSSM &

Water Productivity Mapping using Remote Sensing (WPM) Characteristics of the Satellite Sensor Data

Source: IWMI, 2007

WPMSSM &

IRS P6 (@ 23.5m)

ETM+ (@ 14-30m)

Syrdariya Fergana

Quickbird (@ 0.6-2 m)IKONOS (@ 1-4m)

Hyperion (@ 30m)ALI (@ 30m)

MODIS (@ 250-500 m)

Water Productivity Mapping using Remote Sensing (WPM) Satellite sensor Data Coverage

Source: MODIS

WPMSSM &

Groundtruth Data Collection and

Synthesis

WPMSSM &

Item Unit valueSLNO (num) 65Unigue_Id (code) G-C-F5-V01-02Location GalabaCrop Name CottonNo.Field 2No.Visit 1No.Points 2Latitude (num) 40.39519Longitude (num) 68.84597English Date (num) 07-May-2006Elevation (m) 280Average NDVI from NDVI Camera (unitless) 0.017LAI Meter PAR (µmol m-2s-1) 1399LAI Meter LAI (ratio) 0.12Canopy Cover (%) 7Soil cover (%) 90Water cover (%) 0Weed cover (%) 3Crop Height (mm) 51Crop Density (plants/m2) 15Biomass Wet (kg/m2) 0.01125Biomass Dry (kg/m2) 0.0015Cropping Intensity 3Crop Growth Stage 1Crop Health (1-5) (grading) 2Crop Vigour (grading) 2Soil Moisture (%) *Soil Type (text) *Soil Salinity (%) 4Soil Temperature (EC meter) (0oC) *Electrical conductivity (EC Meter) * *Electrical conductivity (EM-38) * *Electrical conductivity (EM-38) * *Seed type 1Yield (ton/ha) 3.04Water application m3/ha 4394.283333Water Productivity kg/m3 0.69

Water Productivity Mapping using Remote Sensing (WPM) Illustration of Groundtruth Data Collection

Photo Credit: Alexander Platanov

Source: IWMI, 2007

WPMSSM &

Item Unit ValueSLNO (num) 84Unigue_Id (code) G-C-F5-V07-03Location GalabaCrop Name CottonNo.Field 2No.Visit 7No.Points 3Latitude (num) 40.39447Longitude (num) 68.84694English Date (num) 02-Aug-2006Elevation (m) 282Average NDVI from NDVI Camera (unitless) 0.579LAI Meter PAR (µmol m-2s-1) 830LAI Meter LAI (ratio) 1.907Canopy Cover (%) 40Soil cover (%) 53Water cover (%) 5Weed cover (%) 2Crop Height (mm) 400Crop Density (plants/m2) 12Biomass Wet (kg/m2) 1.104Biomass Dry (kg/m2) 0.496Cropping Intensity 3Crop Growth Stage 4Crop Health (1-5) (grading) 4Crop Vigour (grading) 3Soil Moisture (%) 1.5Soil Type (text) *Soil Salinity (%) 4Soil Temperature (EC meter) (0oC) 27.3Electrical conductivity (EC Meter) * 0.88Electrical conductivity (EM-38) * 154.4Electrical conductivity (EM-38) * 103.2Seed type 1Yield (ton/ha) 3.04Water application m3/ha 4394Water Productivity kg/m3 0.692

Water Productivity Mapping using Remote Sensing (WPM) Illustration of Groundtruth Data Collection

Source: IWMI, 2007Photo Credit: Alexander Platanov

WPMSSM &

Variable UnitCollecting method

Sample size

Mean value

Sample size

Mean value

Sample size

Mean value

Sample size

Mean value

A. General Cotton Cotton Wheat Wheat Maize Maize Rice RiceCoordinate degree Hand-held GPS 585 - 191 - 116 - 43 -Soil type - Eye observation 15 - 15 - 6 - 2 - B. Crop variables for spectro-biophysical\Yield modeling NDVI - NDVI camera 566 0.487 166 0.622 105 0.571 43 0.602PAR µmol m-2s-1 LAI meter 580 1060 174 1029 105 960.429 38 957.868Leaf area index - LAI meter 580 1.338 173 2.057 105 1.204 38 2.84Wet biomass kg/m2 Cut and counting 577 1.801 172 1.499 108 2.186 37 2.166Dry biomass kg/m2 Cut and counting 575 0.772 172 0.563 106 0.994 37 0.884Crop height mm Ruler 576 453 172 569.535 108 920.88 41 610.244Soil cover % Eye estimation 585 61.753 175 30.144 113 49.301 42 8.2Canopy cover % Eye estimation 585 34.087 173 58.035 113 36.451 42 69.78Yield ton/ha Laboratory 45 2.109 45 3.495 18 2.983 6 4.523 C. Variables to study the factors affecting Water Productivity EC dS/m EM-38a 315 106.567 48 91.077 62 110.279 26 79.933Soil moisture % Laboratory (weight) 36 12.55 9 16.9 15 11.95 6 18Crop density plant/m2 Cut and counting 577 21.133 172 253.837 97 18.213 39 343.077Weed cover % Eye estimation 585 5.025 173 12.922 108 14.426 42 10.595Water cover % Eye estimation 585 3.51 173 0.556 108 0.01 42 13.738Crop health grading Eye estimation 572 3.164 172 3.291 108 3.231 41 3.78Crop vigor grading Eye estimation 573 3.004 172 3.087 108 3.028 41 3.61 D. Meteorological variables for plant water use estimations or ET calculations Air temperature Selsius degree Automated weather 5798 22.1 Relative humidity % stationb 5798 50 Wind direction degree (February-October) 5798 169.8 Wind Speed KM/h 5798 1.38 Rainfall mm 5798 151.8 E. Water applied measurements Irrigation

applicationmm Weirs 5 293 2 80.57 4 158.9

4355.2

Note: a = Average value of vertical and horizontal EC. b = the "watchdog" automated weather station was set up in Galaba site and the weather data was used for all crops.

Water Productivity Mapping using Remote Sensing (WPM) Groundtruth Data: Variables, Sample size, and Mean Values of the Variables

Source: IWMI, 2007

WPMSSM &

WPM: methods

WPMSSM &

I. Crop productivity (kg\m2) maps (CPMs)• Crop type mapping;• Spectra-biophysical modeling;• Extrapolation to larger areas;

II. Water use (m3\m2) maps (WUMs) or ETactual

• simplified surface energy balance (SSEB) model;

III. Water productivity (kg\m3) maps (WPMs).*********************************************************A. Factors affecting water productivity.

The study involved 3 major steps leading to water productivity maps (WPMs):


Methods

WPMSSM &

1. Crop Type Mapping

WPMSSM &

Legend

Wheat

Settlements

Fallow land

1a. Crop Type Mapping

using IRS LISS 23.5 m time-series Data

Galaba

Source: IWMI, 2007

WPMSSM &

Alfalfa

Cotton

Fallow

Home garden

Orchard

Rice

Settlements

Legend

LULC Areas share

ha %

Alfalfa 858.5 8.7

Cotton 4414.9 44.5

Fallow 1853.8 18.7

Home garden 90.3 0.9

Orchard 1.4 0.0

Rice 361.8 3.6

Settlement 573.9 5.8

other 1769.9 17.8

Sum 9924.5 100.0

LULC and the areas in Galaba site

1b. Crop Type Mapping using Quickbird 2.44 m single date for Galaba

So

urce: IW

MI, 2007

Photo Credit: Alexander Platanov Photo Credit: Alexander Platanov Photo Credit: Alexander Platanov

WPMSSM &

2a. Spectro-Biophysical Models

WPMSSM &

Models and respective equations: (Dry biomass and IRS bands as example)

Linear: bioDry = a + b * IRSbx 　　　　Multi-linear: bioDry = a + b * IRSbx1 + c * IRSbx2 　　　Quadratic: bioDry = a + b * IRSbx + c * IRSbx^2 　　　Logarithmic: bioDry = a + b * log(IRSbx) 　　　　Exponential: bioDry = a * e ** (b* IRSbx) [ Log(BioDry) = Log(a) + b*IRSbx ]

Power: bioDry = a * IRSbx ** b [ Log(BioDry) = Log(a) + b* Log(IRSbx) ]

Images: (1) IRS, (2) Quickbird

Band reflectance as variable: (1) bands, (2) TBVI

Dependent variable: (1) Wet biomass, (2) Dry biomass, (3) Yield, (4) LAI

2a. Spectro-Biophysical\Yield Models Different Types of Models

WPMSSM &

2b. Best Models from all Model Types

WPMSSM &

Spectro-biophysical and yield models. The best models for determining biomass, LAI and yield of 5

crops using IRS LISS and Quickbird data (5-10% points sieved)

Best bands Best indices

Crop Parameter Sensorsample

size Best model band R-square Best modelband

combination R-squareCotton Wet Biomass IRS 140 Exp 2 0.697 Power 2, 3 0.834

QB 41 Multi-linear 1, 4 0.813 Multi-linear 1,4; 3,4 0.506 Dry Biomass IRS 136 Power 2 0.620 Power 2, 3 0.821 QB 41 Exp 2 0.521 Exp 1, 2 0.661 LAI IRS 135 Multi-linear 3, 4 0.634 Power 1, 3 0.725 QB 41 Multi-linear 2, 4 0.511 Quadratic 2, 4 0.574 Yield IRSA 14 Linear 2, 3 0.753 QBB 7 Linear 3, 4 0.610

Wheat Wet Biomass IRS 9 Quadratic 2 0.425 Quadratic 1, 3 0.678 Dry Biomass IRS 14 Quadratic 1 0.205 Quadratic 3, 4 0.309 LAI IRS 18 Quadratic 4 0.8 Multi-linear 1,3; 2,3 0.465 Yield IRS 12 Linear 2, 3 0.67

MaizeD Wet Biomass IRS 19 Power 2 0.815 Power 2, 3 0.871 Dry Biomass IRS 17 Exp 2 0.928 Power 2, 3 0.903 LAI IRS 19 Multi-linear 1, 3 0.777 Multi-linear 1,2; 2,3 0.839

RiceE Wet Biomass QB 10 Multi-linear 1, 2 0.535 Multi-linear 1,2; 2,4 0.600 Dry Biomass QB 10 Multi-linear 1, 2 0.395 Multi-linear 1,3; 2,3 0.414 LAI QB 10 Multi-linear 2, 4 0.879 Quadratic 2, 3 0.234

Alfalfa Wet Biomass IRS 21 Power 2 0.838 Quadratic 1, 2 0.853 QB 8 Multi-linear 2, 4 0.772 Multi-linear 1,2; 2,3; 3,4 0.887 Dry Biomass IRS 21 Power 2 0.817 Exp 1, 2 0.812 QB 8 Multi-linear 2, 4 0.732 Multi-linear 1,2; 2,3; 3,4 0.867 LAI IRS 21 Power 3 0.499 Exp 3, 4 0.639 QB 8 Multi-linear 1, 3, 4 0.927 Multi-linear 1,3; 3,4 0.858

A, Yield model using 2007 dataB, Yield model using 2006 dataC, ∑NDVI camera is the accumulated NDVI derived using the hand hold NDVI camera for field data during 2006 D, Sample points of data from Quickbird for maize was inadequateE, Sample points of data from IRS for rice was inadequate

Note:

2b. Spectro-Biophysical\Yield Models using IRS LISS 23.5m and Quickbird 2.44m Data

Best Model R2 values and Waveband combinations

Source: IWMI, 2007

WPMSSM &

2c. Illustrative Plots of Best Models

WPMSSM &

Cotton wet biomass (WBM)

WBMcotton = 71.18*(IRS TBVI32)3.96

R2 = 0.834

0

2

4

6

8

10

12

0.1 0.2 0.3 0.4 0.5 0.6 0.7IRS TBVI32

co

tto

n w

et

bio

ma

ss

2006

2007


DBMcotton = 51.59*(IRS TBVI32)4.86

R2 = 0.821

0

1

2

3

4

5

6

7

0.1 0.3 0.5 0.7IRS TBVI32

co

tto

n d

ry b

iom

as

s

20062007

Cotton yield

Yieldcotton = 5.156*IRS NDVI - 0.964

R2 = 0.753

0.0

0.5

1.0

1.5

2.0

2.5

0.25 0.3 0.35 0.4 0.45 0.5 0.55IRS-NDVI (Sept 4, 2007)

fiel

d-a

vera

ge-

yiel

d

2007

Cotton Leaf Area Index (LAI)

LAIcotton = 10.37*(IRS TBVI31)1.915

R2 = 0.725

0

1

2

3

4

5

0.1 0.2 0.3 0.4 0.5 0.6IRS TBVI31

co

tto

n L

AI

20062007

Note: * the cotton yield model uses September 4, 2007 IRS LISS image.

2c1. Best Spectro-Biophysical\Yield Models Illustrative Examples for Cotton Crop Variables versus IRS

LISS 23.5 m Data

Source: Xueliang Cai, 2009

WPMSSM &

Wheat Leaf Area Index (LAI)

LAIwheat = 0.005*(IRS b4)2 - 0.509*(IRS b4) + 7.175

R2 = 0.804

0

1

2

3

4

8 10 12 14 16IRS b4

wh

ea

t L

AI

20062007

Maize wet biomass* (WBM)

WBMmaize= 34.664*(IRS TBVI32)4.196

R2 = 0.870

0

2

4

6

8

10

0.1 0.3 0.5 0.7IRS TBVI32

ma

ize

WB

M (

kg

/m2 )

20062007

Wheat yield**

Yieldwheat = 6.192*IRS NDVI - 0.47

R2 = 0.666

1

2

3

4

0.3 0.4 0.5 0.6IRS NDVI (Apr 18, 2007)

Wh

ea

t y

ield

(to

n/h

a)

2007

Cotton Wet biomass (WBM)

WBMcotton = 9.610*(QB TBVI43)1.490

R2 = 0.680

0

1

2

3

4

5

6

7

0.1 0.2 0.3 0.4 0.5 0.6QB TBVI43

Co

tto

n W

BM

(kg

/m2)

2007

Note: * The best model for cotton wet biomass using Quickbird image

** The wheat yield model uses April 18, 2007 IRS LISS image.

is WBMcotton = 17.416 - 0.685 * QBb1 - 0.163 QBb4

2c2. Best Spectro-Biophysical\Yield Models Illustrative Examples for Cotton, wheat, Maize versus Qucikbird and

IRS indices


WPMSSM &

2c. Model Validation

WPMSSM &

Cotton Wet biomass (WBM)

(kg/m2)

Actual = 1.042*Modeled

R2 = 0.790

0

2

4

6

8

0 2 4 6 8

modeled value

actu

al v

alu

e


(kg/m2)


R2 = 0.777

0

1

2

3

0 0.5 1 1.5 2 2.5 3

modeled value

actu

al v

alu

e

Cotton leaf area index (LAI)


R2 = 0.732

0

1

2

3

4

0 1 2 3 4modeled value

actu

al v

alue

Note: Twenty-five percent points were reserved for validation

2d. Spectro-Biophysical\Yield Model Validation Models evaluated for Cotton Crop using IRS LISS 23.5 m

Data


WPMSSM &

2d. Frequency of Occurrence of Bands

WPMSSM &

11

25

3028

6

0

10

20

30

40

blue green red nir swir

Fre

quen

cy o

f oc

curr

ence

(%

)

Note: blue band only occur in Quickbird and swir only in IRS data

An overwhelming proportion of the best models involved the use of indices rather than wavebands. The red and the near-infrared (NIR) bands, which are also required for computation of NDVI, were most frequently occurring. The green band follows closely. Blue band is not very critical in modeling crop variables. SWIR band 1 (see IRS in Table 1) was found, surprisingly, less important.

2e. Frequency of Occurance of Bands in Best Models

Source: IWMI, 2007

WPMSSM &

3. Extrapolation to Larger Areas

WPMSSM &

Wet biomass map (18-Jul-2006)

0 2 41 km 0 2 41 km

0 2 41 km 0 2 41 km

Yield map

Dry biomass map (18-Jul-2006)

LAI map (18-Jul-2006)

19.8

0

Legend (kg/m2) 10.7

0

Legend (kg/m2)

4.1

0

Legend (kg/m2) 3.2

0

Legend (ton/ha)

3. Extrapolation to Larger Areas using IRS LISS 23.5 m using best models of cotton crop variables

Source: IWMI, 2007

WPMSSM &

4a. Water use (ETactual) using simplified surface energy balance Model (SSEBM)

WPMSSM &

ETact – the actual Evapotranspiration, mm.

ETfrac – the evaporative fraction, 0-1, unitless.

ET0 – reference ET, mm.

Tx – the Land Surface Temperature (LST) of pixel x from thermal data.

TH/TC – the LST of hottest/coldest pixels.

CH

xHfrac TT

TTET

fracact ETETET 0

4a. Water use (ETactual) Water use is determined by multiplying Evaporative fraction by reference ET

WPMSSM &

4b. ETfrac.

WPMSSM &

2006-04-24 2006-05-10 2006-06-11

2006-07-29 2006-08-14 2006-10-01

Legend unitless

0 2 41 km 0 2 41 km 0 2 41 km

0 2 41 km 0 2 41 km 0 2 41 km

4b. ETfraction Evaporative fraction using Landsat ETM+ thermal data during crop growing season

Source: IWMI, 2007

WPMSSM &

4c. ETref.

WPMSSM &

4c. ETreference Using Penman-Montieth Equation based on meteorological data

Source: IWMI, 2007

WPMSSM &

4d. ETact.

WPMSSM &

0 2 41 km

ET map for cotton field, average ET=512mm (Apr 15 – Oct 15, 2006)

905

150

Legend mm

0 2 41 km

ET map for rice field, average ET=619mm (May 21 – Sept 28, 2006)

769

165

Legend mm

4d. ETactual

ETactual = ETfraction * ETreference

Water use of Irrigated Cotton versus Irrigated Rice crops

Source: IWMI, 2007

WPMSSM &

5. Water Productivity Maps (WPM)

WPMSSM &

(1))/m(m (ETa) useWater

)(kg/moductivityPrCrop)(kg/moductivityPrWater

23

23

5a. Water Productivity

WP calculated in this study

WPMSSM &

WPM using IRS LISS 23.5 m and Quickbird 4 m

IRS Quickbirdgroup area (ha) share (%) area (ha) share (%)0-0.1 167.9 4.1 85.1 1.8

0.1-0.2 695.5 16.8 970.2 21.00.2-0.3 1421.0 34.4 1550.7 33.50.3-0.4 1381.7 33.4 1370.0 29.60.4-0.5 414.7 10.0 542.7 11.70.5-0.6 50.3 1.2 93.0 2.0

>0.6 2.4 0.1 17.9 0.4

5b1. Water productivity maps (WPMs) at 2 Resolutions (Scales)

Mean Min Max Range No. Pixels

IRS 0.285 0 0.70 0.7 74850

QB 0.289 0 1.06 1.06 7776175

Unit: kg/m3

0 2 41 km

IRS

0 2 41 km

QB

Kg/m3 Kg/m3

Source: IWMI, 2007

WPMSSM &(a) (b)

Within field variability of water productivity derived from (a) Indian Remote Sensing (IRS) P6 and (b) Quickbird images

Mean Min Max Range Area (ha) No. Pixels

IRS 0.334 0.08 0.60 0.52 81 1296

QB 0.339 0.07 0.68 0.61 80.2 134642

0 200 400100 m 0 200 400100 m0.68

0.07

Legend kg/m3

0.68

0.07

Legend kg/m3

Unit: kg/m3

5b2. Water productivity maps (WPMs) at 2 Resolutions (Scales) IRS LISS 23.5 m versus Quickbird 2.44 m

Source: IWMI, 2007

WPMSSM &

Within field variability of water productivity derived from (a) Indian Remote Sensing (IRS) P6 and (b) Quickbird images

0.58

0.09

Legend kg/m3

0.58

0.09

Legend kg/m3

(a) (b)

Mean Min Max Range Area (ha) No. Pixels

IRS 0.334 0.12 0.50 0.38 6.7 107

QB 0.339 0.09 0.58 0.49 6.6 11083 Unit: kg/m3

5b3. Water productivity maps (WPMs) at 2 Resolutions (Scales) IRS LISS 23.5 m versus Quickbird 2.44 m

Source: IWMI, 2007

WPMSSM &

5c. Water Productivity

Scope for improvement

Legend

<0.3

0.3-0.4>0.4

Kg/m3

With an average value of 0.3 kg/m3, the water productivity map shows explicit scope for improvement: where and how much.

Source: IWMI, 2007

WPMSSM &

6. Factors Affecting Water Productivity

WPMSSM &

6a. Factors Affecting WP Degree of influence of Various factors on WP variations within and between field as

measured during field work

Land leveling

Weeds 6%

Salinity

Water deficit

43%

31%

14%

7%

Water logging






WPMSSM &

100 0 10050 Meters

0.69

0.06

Legend unitless

Quickbird NDVI (26-July-2006)

0.69

0.06

Legend unitless

100 0 10050 Meters

IRS NDVI (18-July-2006)High salinity

Low salinity

6c. Factors Affecting WP Degree of influence of Various factors on WP variations within and between field as

measured during field work



Source: IWMI, 2007

WPMSSM &

6d. Factors Affecting WP Distinguish different levels of factors affecting WP using crop variables and spectral

vegetation indices – soil salinity as example

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

A B C D

Group

Ave

rag

e T

BV

I val

ue

TBVI32

TBVI41

EC values were grouped into four groups indicating different levels of soil salinity. Mean TBVI values of the respective groups were plotted.

TBVI32 TBVI41

AB,AC,AD, BC,BD

AD,BD

Test of differences:

Source: IWMI, 2007

WPMSSM &

Note: # = Statistical tests of significance were conducted to determine significant differences between two groups of data. For example, A-B# indicates that IRS band 2 mean spectral reflectivity is significant different between moisture level A and B at 95% confidence level or higher

Grade CriterionNo. of

samplesNDVI (

- )LAI (

- )

Wet biomass (kg/m2)

Dry biomass (kg/m2) IRSb1 IRSb2 IRSb3 IRSb4 TBVI21 TBVI31 TBVI32 TBVI41 TBVI42 TBVI43 SAVI

Soil A x=1 417 0.487 1.337 1.811 0.781 13.387 13.323 24.891 11.348 0.012 0.299 0.305 0.081 0.069 0.371 0.452 moisture B x<=1.2 75 0.480 1.272 1.816 0.802 12.230 11.261 27.406 10.778 0.053 0.380 0.419 0.063 0.009 0.433 0.620

[descriptive] C 1.2<x<=1.7 42 0.564 1.547 1.600 0.649 12.813 12.776 22.966 10.422 0.007 0.280 0.284 0.106 0.100 0.376 0.420 D x>1.7 33 0.581 1.650 2.342 0.943 11.046 10.086 21.181 8.659 0.051 0.307 0.350 0.123 0.072 0.415 0.517

Test of statistical significance (TSS)#

AB#,AD, CD

AB,AD, CD

AB,AC, AD,BC,

BD

AC,AD,BD,CD

AB,AD,BC,CD

AB,BC AB,BC AC,AD,BC,BD

AB,BC,BD

AB,BC AB,BC

Crop A x<=10 77 0.443 1.062 0.790 0.360 13.554 13.926 22.611 11.001 0.009 0.250 0.240 0.104 0.112 0.345 0.355 density B 10<x<=20 261 0.524 1.488 2.059 0.835 12.656 12.166 24.625 10.675 0.029 0.318 0.340 0.085 0.056 0.393 0.503

[plant/m2] C 20<x<=40 200 0.495 1.393 2.052 0.940 13.408 13.333 25.476 11.448 0.014 0.308 0.315 0.077 0.063 0.376 0.467 D x>40 34 0.347 0.720 0.822 0.308 13.901 14.165 25.299 12.052 0.004 0.287 0.281 0.070 0.074 0.351 0.416

TSS

BD,CD AB,AC,

BD,CDAB,AC, BD,CD

AB,AC, BD,CD

BC AB,BC AB,AC BC AB AB,AC AB,AC AB,AC AB,AC AB AB,AC

Water A x=0 537 0.491 1.335 1.829 0.787 13.133 12.955 24.846 11.136 0.017 0.306 0.317 0.081 0.064 0.378 0.469 cover B x<=5 18 0.562 1.756 1.598 0.605 12.158 11.894 21.218 9.219 0.020 0.265 0.280 0.139 0.120 0.391 0.413 [%] C x>=5 9 0.597 1.586 1.935 0.799 11.213 9.955 19.882 7.547 0.059 0.279 0.333 0.195 0.138 0.450 0.491

TSS AB AB,AC AB,AC Weed A x<=2 191 0.477 1.290 1.785 0.728 13.075 12.811 24.830 11.080 0.020 0.306 0.319 0.082 0.062 0.380 0.473 cover B 2<x<=5 159 0.459 1.269 1.689 0.695 12.583 12.161 24.180 10.511 0.027 0.313 0.333 0.090 0.063 0.393 0.493 [%] C 5<x<=10 50 0.397 1.007 0.903 0.395 14.172 14.550 23.481 11.679 0.007 0.248 0.239 0.097 0.104 0.336 0.354

D x>=10 45 0.408 0.919 0.945 0.425 15.111 16.099 23.743 12.639 0.029 0.222 0.193 0.089 0.118 0.304 0.285

TSS

AD,BD AC,AD,

BC,BDAC AD,BC,

BDAD,BC,

BD AD,BC,

BDAD,BC,

BDAD,BD AD,BD AD,BD AD,BD AD,BD

ECEM38 A x<=6 44 0.669 2.181 2.961 1.421 12.389 11.769 27.221 10.751 0.035 0.370 0.394 0.070 0.035 0.429 0.584 [dS/m] B 6<x<=10 90 0.557 1.523 1.966 0.919 13.650 13.771 25.005 11.616 0.004 0.291 0.291 0.080 0.076 0.363 0.431

C 10<x<=14 104 0.503 1.301 1.404 0.677 14.184 14.725 23.882 11.795 0.014 0.253 0.238 0.092 0.105 0.337 0.352 D x>=14 70 0.355 0.926 0.777 0.345 14.641 15.428 22.638 11.805 0.024 0.215 0.192 0.109 0.132 0.316 0.284

TSS

AB,AC, AD,

BD, CD

AB,AC, AD,BD,

CD

AB,AC, AD,BD

AB,AC, AD,BD,

CD

AB,AC, AD, BD,

AB,AC,AD,BD

AB,AC, AD,BD

AB,AC,AD

AB,AC,AD,BC,BD

AB,AC,AD,BD

AB,AC,AD,BC,BD

AD,BD AB,AC,AD,BC,BD

AB,AC,AD,BD

AB,AC,AD,BC,

BD

6d. Factors Affecting WP Distinguish different levels of factors affecting WP using crop variables and spectral

vegetation indices

Source: IWMI, 2007

WPMSSM &Note: # = Statistical tests of significance on crop vigor and health were conducted to determine significant differences between two groups of data.

For example, A-B# indicates that IRS band 2 mean spectral reflectivity is significant different between moisture level A and B at 95% confidence level or higher

Grade CriterionNo. of

samplesNDVI

(-)LAI (-)

Wet biomass (kg/m2)

Dry biomass (kg/m2) IRSb1 IRSb2 IRSb3 IRSb4 TBVI21 TBVI31 TBVI32 TBVI41 TBVI42 TBVI43 SAVI

Crop vigor A wrost 40 0.280 0.568 0.399 0.201 15.164 16.226 22.235 12.291 0.033 0.189 0.157 -0.107 -0.139 -0.290 0.233 [descriptive] B 117 0.329 0.860 0.624 0.255 14.362 14.941 23.066 11.736 0.016 0.231 0.216 -0.101 -0.116 -0.325 0.319

C 246 0.466 1.189 1.642 0.720 12.945 12.645 24.587 10.978 -0.020 0.308 0.323 -0.083 -0.063 -0.382 0.478 D 131 0.666 1.987 3.099 1.258 12.000 11.139 26.524 10.453 -0.047 0.376 0.411 -0.068 -0.021 -0.433 0.608 E best 35 0.814 2.708 3.844 1.786 10.458 8.718 30.227 9.536 -0.091 0.480 0.547 -0.047 0.044 -0.516 0.810

Test of statistical

significance#

AC,AD, AE,BC, BD,BE, CD,CE,

DE

AC,AD,AE,BC,BD,BE,CD,CE,

DE


DE


DE


DE


DE


DE

AC,AD, AE,BC, BD,BE,

CE


DE


DE


DE

AC,AD, AE,BC, BD,BE, CD,CE


DE


DE


DE

Crop health A wrost 30 0.304 0.619 0.307 0.153 15.085 16.105 22.120 12.030 0.032 0.189 0.158 0.116 0.147 0.297 0.233 [descriptive] B 110 0.323 0.775 0.600 0.279 14.217 14.797 22.667 11.569 0.015 0.228 0.213 0.104 0.119 0.325 0.315

C 199 0.410 1.068 1.332 0.607 13.524 13.500 24.448 11.393 0.009 0.285 0.290 0.087 0.078 0.363 0.430 D 202 0.645 1.834 2.893 1.163 12.043 11.233 25.908 10.480 0.044 0.364 0.398 0.069 0.025 0.423 0.588 E best 31 0.783 2.772 3.593 1.715 11.509 10.318 29.452 10.162 0.063 0.430 0.474 0.060 0.003 0.479 0.703

Test of statistical

significance


DE


DE

AC,AD, AE,BC, BD,BE, CD, CE


DE

AC,AD,AE,BD,BE,CD,

CE

AC,AD,AE,BC,BD,BE,CD,CE


DE

AD,AE, BD,BE,

CD




AC,AD, AE,BC, BD,BE,

CD




6e. Factors Affecting WP Distinguish different levels of factors affecting WP using crop variables and spectral

vegetation indices

Source: IWMI, 2007

WPMSSM &

• Soil salinity was well captured by all IRS bands, TBVIs and field measured LAI, WBM and DBM.

• The water cover levels were distinguishable only using IRS band 4 reflectivity, TBVI41, and to lesser extent band 3 reflectivity.

• Moisture differences were best captured by IRS band reflectivity in band 3, band 4, TBVI21 and TBVI41.

• Crop density results in significant differences in LAI and biomass, Certain groups of crop density can be distinguished using IRS indices. But crop density together with water cover, were not differentiated as well as other factors.

• Conventional NDVI32 is much less sensitive to differences in water cover, soil moisture and weed cover. This highlights the need to use multiple indices rather than a single NDVI.

6d. Factors Affecting WP Results of the Statistical Tests of Significance

WPMSSM &

Conclusions

WPMSSM &

Water Productivity Mapping (WPM) using Remote Sensing

Conclusions1. Methods and protocols for WPM using remote sensing was well established and

demonstrated. The process involved 3 broad steps: (a) crop productivity (kg\m2) maps (CPMs), (b) water use (m3\m2) maps, and (c) water productivity (kg\m3) maps;

2. Crop productivity maps (CPMs) were produced by: (a) crop type mapping, (b) spectro-biopysica\yield modeling, and (c) extrapolation of model results to spatial domains using remote sensing;

3. Water use maps (ETactual) (WUMs) were produced using simplified surface energy balance model (SSEBM);

4. Water productivity maps (WPMs) were then produced by dividing CPMs by WUMs.

5. Results showed that WP of the irrigated cotton crop (the most dominant crop in the Syr Darya river basin) varied between 0-0.6 kg\m3. Of this only 11 percent of the cotton crop area was in 0.4 kg\m3 or higher WP. About 55% of the cotton area had <0.3 kg\m3. The results had similar trends for rice, maize, and wheat.

6. The results imply that there is highly significant scope to increase WP (to grow “more crop per drop”) through better management practices. The challenge is to increase land and water productivity of the 55% of the areas. If we can achieve that, food security of future generations can be secured without having to increase croplands and\or greater water use.

WPMSSM &

Thanks

www.iwmi.org

Education

Water Productivity Mapping (WPM) at various Resolutions (scales) using Remote Sensing