water productivity of irrigated crops using hyperspectral remote sensing michael marshall u.s....

53
Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ 86001 Ph: (928) 556-7215, Fx: (928) 556 7169 [email protected] August 15, 2013 1

Upload: molly-walker

Post on 17-Jan-2016

215 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

1

Water Productivity of Irrigated Crops

using Hyperspectral Remote Sensing

Michael Marshall U.S. Geological Survey (USGS)

2255 N. Gemini Dr.Flagstaff, AZ 86001

Ph: (928) 556-7215, Fx: (928) 556-7169

[email protected]

August 15, 2013

Page 2: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

2

California water crisis Crop water productivity defined Remote sensing methods Project objectives Study area Field methods Current progress Near future

Presentation Overview

Page 3: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

3

California Water Crisis

15% of national receipts for crops

Water supply deficit 2 million acre-ft

Irrigated agriculture 75-80% of annual water budget (USDA, 2009)

(NRDC 2010)

Page 4: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

4

Bos (1985) expresses WP as follows:

Crop Water Productivity (WP)

Water productive crops assimilate more carbon and increase biomass/yield, while losing less water to atmospheric demand.

𝑊𝑃=𝑐𝑟𝑜𝑝𝑏𝑖𝑜𝑚𝑎𝑠𝑠 (𝑘𝑔𝑜𝑟 $ )

𝑐𝑟𝑜𝑝𝑒𝑣𝑎𝑝𝑜𝑡𝑟𝑎𝑛𝑠𝑝𝑖𝑟𝑎𝑡𝑖𝑜𝑛(𝑚3)

Page 5: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

5

Hyper- spatial/spectral Remote Sensing

Page 6: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

6

Field Spectral Data

LAI = 0.13

LAI = 6.89

LAI = 4.36

Page 7: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

7

Biomass/yield for key crops in CA‒ Hyper-spectral bands and techniques (field)‒ Hyper- and multi- spatial/spectral image fusion of

biomass/yield

Evapotranspiration (ET) for key crops in CA‒ Irrigation specific ET parameterization, hyper- spectral

canopy fraction, and CIMIS EP

‒ Hyper- and multi- spatial/spectral image fusion of ET

Crop Water Productivity (WP) for key crops in CA‒ Determine contributing factors and sensitivity to lower

WP‒ Potential climate driven water costs and savings

alternatives

Primary Research Objectives and Deliverables

Page 8: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

8

12 farms spanning Kern, Kings, Fresno, Sacramento, Solano, Yolo, and Yuba counties

Alfalfa, corn, cotton, and rice

Drip, pivot-line, and furrow

1200 spectra per visit 3 visits (rapid-growth,

flowering, and grain-filling)

Includes ET fields (rice x 2, corn x 3, alfalfa, and cotton)

Study Area (2011-2012)

Page 9: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

9

Field Visit 2012

Page 10: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

10

Field Methods (biomass)

60 m*

10 x 1m2 samples per frame‒ Spectra (ASD FieldSpec Pro3)‒ Height‒ # of plants per sample‒ Vegetation fraction (fc)‒ LAI/PAR (Decagon AccuPAR)‒ Biochemical

Aboveground wet biomass sample Allometric equation

Page 11: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

11

Agriculture and Agri-Food Canada (http://www.flintbox.com/public/project/5470/)

RGB photos taken above canopy Histogram-threshold approach to estimate (fc) and LAI RGB indices

Vegetation Fraction

(Liu and Pattey 2010)

Page 12: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

12

Allometric Equations

Cotton (Calibration) Cotton (Validation)

Adj R2 = 0.67p<0.001

Adj R2 = 0.67p<0.001

Page 13: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

13

Rice (Calibration) Rice (Validation)

R2 = 0.75p<0.001

R2 = 0.70p<0.001

Page 14: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

14

R2 = 0.67p<0.001

R2 = 0.62p<0.001

Page 15: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

15

R2 = 0.62p<0.001

R2 = 0.43p<0.001

Maize (Calibration) Maize (Validation)

Page 16: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

16

Biomass Methods with Field Spec

Spectra was collected from 350-2500nm between ±2 hours of solar noon

Inter-sensor calibration Atmospheric noise Hyperion aggregation (1nm —› 10nm) 1st derivative, 2nd derivative, exponential,

inverse logarithmic [log(1/R)] transformations

Page 17: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

17

Raw Spectra (per crop type)N = 65

Rice, best predicted with highest correlations in the NIR. High correlations around 428nm, but not consistently through the visible.

Mx = 0.73 @ 895 nmMn = -0.54 @ 428 nm

Page 18: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

18

N = 85

Alfalfa, probably best predicted overall. Very high correlations around the red-edge and SWIR, but lower in the NIR compared to rice.

Mx = 0.55 @ 763 nmMn = -0.50 @ 651 nm

Page 19: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

19

N = 109

Cotton, probably next best predicted. Very high correlations in the visible and over a larger range than the other crops. Characteristic biomass bands

Mx = 0.43 @ 895 nmMn = -0.54 @ 448 nm

Page 20: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

20

N = 109

Maize, the worst predicted. Very low correlations overall. Highest in the NIR

1st derivative considerably better

Mx = 0.32 @ 1094 nmMn = -0.14 @ 428 nm

Page 21: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

21

log(1/R)Used in the literature to indicate absorption (i.e. negative correlations with biomass now become positive). Overall, no significant improvements, however, slight improvements in the visible for all crops:

Page 22: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

22

Page 23: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

23

Page 24: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

24

Page 25: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

25

1st Derivative Much higher correlations in the NIR than raw spectra, but not the SWIR

Mx = 0.87 @ 1155nmMn = -0.79 @ 1124 nm

Page 26: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

26

Correlations higher than the raw spectra overall, particularly around the red-edge

Particularly high correlation at 1437nm (water)

Mx = 0.62 @ 773 nmMn = -0.63 @ 763 nm

Page 27: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

27

Correlations higher than the raw spectra in the NIR, but raw spectra much higher in the visible

Mx = 0.63 @ 1054 nmMn = -0.62 @ 943 nm

Page 28: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

28

Correlations higher than the raw spectra overall (visible, red-edge, NIR, and SWIR).

Mx = 0.60 @ 773 nm Mn = -0.42 @1679 nm

Page 29: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

29

2nd Derivative

Page 30: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

30

Page 31: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

31

Page 32: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

32

Page 33: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

33

2nd derivative results are very noisy, however strong inflection points exist at 468 (cotton), 529, 733, 1034, and 1235 nm

At each inflection point, a pseudo integral was performed over a selected window. In each window, the 2nd derivatives were summed and divided by the difference in wavelengths. The windows were centered at 468, 529, …

The window width was done iteratively from 25 – 300 nm at 25 nm increments

Some examples…

Page 34: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

34

Page 35: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

35

Stepwise RegressionCall:lm(formula = ident$Biomass ~ ident$Constant + ident$R895 + ident$R1599)

Residuals: Min 1Q Median 3Q Max -1596.0 -522.0 -118.3 397.8 2213.6

Coefficients: (1 not defined because of singularities) Estimate Std. Error t value Pr(>|t|) (Intercept) 154.8 309.2 0.501 0.618974 ident$Constant NA NA NA NA ident$R895 8634.1 1029.2 8.389 0.000000000112 ***ident$R1599 -10647.7 2625.4 -4.056 0.000201 ***---Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 795.5 on 44 degrees of freedomMultiple R-squared: 0.6449, Adjusted R-squared: 0.6288 F-statistic: 39.96 on 2 and 44 DF, p-value: 0.0000000001281

Page 36: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

36

Call:lm(formula = ident$Biomass ~ ident$Constant + ident$R1145 + ident$R1740)

Residuals: Min 1Q Median 3Q Max -1244.73 -438.06 -98.34 280.62 1949.94

Coefficients: (1 not defined because of singularities) Estimate Std. Error t value Pr(>|t|) (Intercept) 588.1 221.7 2.652 0.0111 * ident$Constant NA NA NA NA ident$R1145 -773402.7 83200.0 -9.296 0.00000000000605 ***ident$R1740 1799079.6 786673.4 2.287 0.0271 * ---Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 715.2 on 44 degrees of freedomMultiple R-squared: 0.713, Adjusted R-squared: 0.7 F-statistic: 54.66 on 2 and 44 DF, p-value: 0.000000000001183

1st Derivative

Page 37: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

37

2nd DerivativeCall:lm(formula = ident$Biomass ~ ident$Constant + ident$R1155 + ident$R1104)

Residuals: Min 1Q Median 3Q Max -1031.74 -251.20 -53.93 138.28 1724.47

Coefficients: (1 not defined because of singularities) Estimate Std. Error t value Pr(>|t|) (Intercept) 259.6 143.5 1.809 0.077235 . ident$Constant NA NA NA NA ident$R1155 8186742.3 625948.1 13.079 < 0.0000000000000002 ***ident$R1104 -14968697.3 3724062.7 -4.019 0.000225 ***---Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 575.1 on 44 degrees of freedomMultiple R-squared: 0.8144, Adjusted R-squared: 0.806 F-statistic: 96.56 on 2 and 44 DF, p-value: < 0.00000000000000022

Page 38: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

38

Inverse LogCall:lm(formula = ident$Biomass ~ ident$Constant + ident$R895 + ident$R743)

Residuals: Min 1Q Median 3Q Max -1197.7 -489.6 -150.0 350.7 3086.4

Coefficients: (1 not defined because of singularities) Estimate Std. Error t value Pr(>|t|) (Intercept) 1193.5 636.2 1.876 0.067281 . ident$Constant NA NA NA NA ident$R895 -12295.7 2344.5 -5.245 0.00000428 ***ident$R743 11050.1 2853.8 3.872 0.000354 ***---Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 830.3 on 44 degrees of freedomMultiple R-squared: 0.6132, Adjusted R-squared: 0.5956 F-statistic: 34.88 on 2 and 44 DF, p-value: 0.0000000008411

Page 39: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

39

Best index at 1215 and 1256 nm (R2 = 0.68). Three potential outliers (sresid > 2,<-2): RIBIG0101.3.2011, RIMIR0109.2.2012, RIWIL0101.3.2012. Heteroskedastic

2 Band HS Indices

Rice (Scatterplot – Raw Spec)

Page 40: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

40

Rice (2D λ-λ plots)

1215 & 1256 R2 = 0.68 1145 & 1155 R2 = 0.71

1225 & 1447 R2 = 0.67 692 & 1155 R2 = 0.52

Page 41: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

41

Alfalfa (Scatterplot – Raw Spec)

Best index at 651 and 692 nm (R2 = 0.55). Three potential outliers (sresid > 2,<-2): ALSHA1010.1.2011, ALSHA0901.3.2011, ALSHA1001.3.2011. Heteroskedastic

These residuals were identified in the original biomass analysis with photos, LAI meter, etc. I suspect that I double sampled the alfalfa by mistake

Page 42: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

42

Rice (2D λ-λ plots)

651 & 692 R2 = 0.55 631 & 705 R2 = 0.42

845 & 1195 R2 = 0.62 651 & 682 R2 = 0.46

Page 43: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

43

Cotton (Scatterplot – Raw Spec)

Best index at 1023 and 1084 nm (R2 = 0.42). Five potential outliers (sresid > 2,<-2): COSHA0609.2.2011,COWSR0309.2.2011,COSHA0801.3.2011,COSHA0901.3.2011, COWSR0801.3.2011. Heteroskedastic

Page 44: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

44

Cotton (2D λ-λ plots)

1023 & 1084 R2 = 0.42 973 & 1114 R2 = 0.37

428 & 1054 R2 = 0.45 1155 & 1447 R2 = 0.40

Page 45: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

45

Maize (Scatterplot – Raw Spec)

Best index at 1669 and 1699 nm (R2 = 0.33). Five potential outliers (sresid > 2,<-2): MATYL0401.3.2011, MASTA1010.1.2012, MADAV0909.2.2012.

Page 46: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

46

Maize (2D λ-λ plots)

1669 & 1699 R2 = 0.33 763 & 845 R2 = 0.26

794 & 1205 R2 = 0.39 692 & 733 R2 = 0.28

Page 47: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

47

Next steps…

Identify best 2-band hyperspectral index PCA and piecewise regression Estimate biomass and extrapolate

Page 48: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

Biomass modeling (REGFLEC)

1) PROSPECT (leaf radiative transfer model)1) Specific leaf area2) Leaf water content3) Chlorophyll a,b4) # of layers

2) SAIL (canopy radiative transfer model)1) Clumping index2) View zenith angle3) Solar illumination specular ratio4) Solar incident zenith angle5) Canopy “hotspot” parameter

3) Forward direction: spectral indices4) Inverse direction: biomass

48

Page 49: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

49

Remote Sensing Data FusionSensor Wavelength

range (μm)Spatial resolution (m)

# of bands Revisit (days) Digitization (bit)

E0-1 Hyperion 0.43-2.40 30 196 16 16

E0-1 ALI 0.43/2.35 30 10 16 16

Landsat TM 0.45-2.35 30 7/8 16 8

Landsat TIR 10.40-12.50 120/60 1 16 8

ASTER 0.52-2.43 30 14 16 8

ASTER TIR 8.125-11.65 15/30/90 14 16 8

SPOT 5 0.5-1.75 10/20 5 2-3 days 8

GeoEye-1,2 0.45-0.90 1.65 5 <3 11

IKONOS 0.45-0.93 4 5 3 11

Quickbird 0.45-0.90 2.44 5 1-6 11

Rapideye 0.44-0.85 5-6.5 5 1-6 16

ISAAC Same as Landsat Bands 2,3,4

Varies 3 Weekly* 8

Page 50: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

50

LE Latent heat LEs + LEc + LEI(Fisher et al., 2008)

LEc Canopy LE (Fisher et al., 2008); (Priestley and Taylor, 1972)

LEs Soil LE (Fisher et al., 2008); (Priestley and Taylor, 1972)

LEi Wet canopy LE (Fisher et al., 2008); (Priestley and Taylor, 1972)

fwet Relative surface wetness RH4 (Fisher et al., 2008)

fg Green canopy fraction (Fisher et al., 2008)

fT Plant temperature constraint (June et al., 2004)

fM Plant moisture constraint (Fisher et al., 2008)

fSM Soil moisture constraint RHVPD/β (Fisher et al., 2008)

fAPAR Fraction of PAR absorbed by green vegetation cover

m1SAVI + b1(Gao et al., 2000), (Huete, 2006)

fIPAR Fraction of PAR intercepted by total vegetation cover

m2NDVI + b2(Fisher et al., 2007)

fc Fractional total vegetation cover fIPAR(Campbell and Norman, 1998)

Topt Optimum plant growth temperature Tmax @ max{PARfAPARTmax / VPD} (Fisher et al., 2007)

NCMTgWETRffff

)1(

)())1(( GRfffNSWETSMWET

NCWETRf

IPARAPARff

2

OPTMAX TT

emaxAPARAPAR

ff

PT-JPL model (ET)

Page 51: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

𝜃𝑛=∑ (𝑃+𝐼 ) h𝑤 𝑒𝑟𝑒 𝐼=0.5 𝜃 𝑓𝑐 if 𝜃𝑛− 1<0.5𝜃 𝑓𝑐

Soil evaporation (Es)

{∑ 𝐸𝑠=∑ (1−𝑒−𝑘𝐿𝐴𝐼)𝐸0∑ 𝐸0<¿ 𝛽2 ¿∑ 𝐸𝑠=𝛽 [∑ (1−𝑒−𝑘𝐿𝐴𝐼)𝐸0 ]12∑ 𝐸0≥ 𝛽

2

𝐸𝑠=∑ 𝐸𝑠 ,𝑛−∑ 𝐸𝑠 ,𝑛−1

𝛽=𝜃𝑛−𝜃𝑤𝜃 𝑓𝑐−𝜃𝑤

51

The following formula puts Es in terms of precipitation and irrigation…

(Ozdogan et al. 2010)

(Mintz et al. 1992)

(Ritchie 1972)

Page 52: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

52

1) Daily reference ET (E0)

2) 2 km resolution3) Solar radiation (GOES)4) Air temperature, relative humidity,

and wind speed (2m) are interpolated from CIMIS network

CIMIS spatial data (http://wwwcimis.water.ca.gov/)

E0

June 19, 2011

𝐸0=∆ (𝑅𝑛−𝐺 )

λ [∆+𝛾 (1+𝐶𝑑𝑢2 ) ]+𝛾

37𝑇𝑎+273

𝑢2 (𝑒𝑠−𝑒𝑎 )

∆+𝛾 (1+𝐶𝑑𝑢2 )

Page 53: Water Productivity of Irrigated Crops using Hyperspectral Remote Sensing Michael Marshall U.S. Geological Survey (USGS) 2255 N. Gemini Dr. Flagstaff, AZ

1) Addresses critical research areas outlined in USGS Science Strategy 2017: Water Census, Climate Variability/Change, and Ecological Change

2) Farmers, scientists, and decision-makers will have the ability to pinpoint areas of lower/higher WP

3) Stakeholders will be able to explore factors contributing to lower WP for improved management

4) Evapotranspiration is a key component of land surface and atmospheric hydrologic models

5) Biomass is a key component of ecological models6) Climate change research7) Techniques developed here can be applied to other parts of the United

States and to developing countries where water scarcity is an ever-present reality

53

Relevance/Implications