METRIC: High Resolution Satellite Quantification of Evapotranspiration

Transcription

1 Copyright, R.G.Allen, University of Idaho, 2005 METRIC: High Resolution Satellite Quantification of Evapotranspiration Rick Allen University of Idaho, Kimberly, Idaho Overview Co-developers and Collaborators: M. Tasumi, R. Trezza, W. Bastiaanssen, T. Morse, W. Kramber, J. Wright

2 ET mapping with SEBAL and METRIC tm Surface Energy Balance Algorithm for Land Dr. Wim Bastiaanssen, WaterWatch, The Netherlands beginning in 1990 SEBAL is commercially applied in the U.S.A. by SEBAL-North America Mapping EvapoTranspiration with high Resolution and Internalized Calibration Allen and Tasumi, University of Idaho, Kimberly beginning in 2000 rooted in SEBAL 2000 METRIC tm is energy-balance-based ET mapping tied down and partly calibrated using ground-based reference ET (from weather data) METRIC tm is designed to work well in advective conditions of the western U.S.

3 Why Energy balance? ET is calculated as a residual of the energy balance Rn (radiation from sun and sky) H (heat to air) ET Basic Truth: Evaporation consumes Energy ET = R - G - H n G (heat to ground) The energy balance includes all major sources (R n ) and consumers (ET, G, H) of energy

4 Energy balance gives us actual ET Therefore, we can account for impacts on ET caused by: water shortage disease crop variety planting density cropping dates salinity management (these effects can be converted directly into a crop coefficient)

5 Seasonal ET for SE Idaho Idaho from Landsat Major Irrigated areas in Idaho and areas of METRIC application

6 Imperial Valley, CA via Landsat 7 Imperial Valley ET (mm) ET during January March, 2003

7 Path 35 ET (mm/yr) ETrF Jemez Alcalde Espanola NWS Pena Blaca Santa Fe Airport Angostura Albuquerque Candelaria Farms Albuquerque Intl. Airport LosLunas Jarales Boys Ranch Path 34 San Acacia Middle Rio Grande of New Mexico Imperial Valley, CA via Landsat 7

8 Why use High Resolution Imagery? 8/14/00 ET from individual Fields is Critical for Water Rights, Water Transfers, Farm Water Management Minidoka County, Idaho

9 Why use High Resolution Imagery? Landsat False Color (8/26/2002) Landsat False Color (8/26/2002) Daily ET (8/26/2002) Monthly ET (Aug. 2002) Daily ET (8/26/2002) Annual ET (2002) Monthly ET (Aug. 2002) Annual ET (2002) ET (mm/yr) 0 ETrF Minidoka County, Riparian vegetation and small fieldsidaho along the Middle Rio Grande, New Mexico

10 Why use High Resolution Imagery? Landsat vs MODIS Landsat False Color (MRG) 8/26/ :33am MODIS False Color (MRG) 8/26/ :02am

11 True Color southcentral Idaho August 14, 2000 Thousand Springs North Dairy area (lots of corn, alfalfa) Twin Falls recent burn 100 miles basalt Wood River Valley Burley Craters of the Moon Lake Walcott

12 False Color southcentral Idaho August 14, 2000 Red indicates active vegetation Thousand Springs North Dairy area (lots of corn, alfalfa) Twin Falls recent burn 100 miles basalt Wood River Valley Burley Craters of the Moon Lake Walcott

13 Surface Temperature southcentral Idaho August 14, 2000 Thousand Springs North Twin Falls recent burn basalt Wood River Valley Temperature ( o C) Burley Craters of the Moon Lake Walcott

14 Net Radiation southcentral Idaho August 14, 2000 Thousand Springs North R n H ET G Twin Falls recent burn basalt Wood River Valley Net Radiation (W/m 2 ) Burley Lake Walcott Craters of the Moon

15 Ground Heat Flux southcentral Idaho August 14, 2000 R n H ET Thousand Springs North G Twin Falls recent burn basalt Wood River Valley Soil Heat Flux (W/m 2 ) Burley Craters of the Moon Lake Walcott

16 Heat Flux to Air southcentral Idaho August 14, 2000 R n H ET Thousand Springs North G Twin Falls recent burn basalt Wood River Valley Sensible Heat (W/m 2 ) Burley Craters of the Moon Lake Walcott

17 Instantaneous ET southcentral Idaho August 14, 2000 R n H ET Thousand Springs North G Twin Falls recent burn Wood River Valley Latent Heat (W/m 2 ) 0 basalt Burley Craters of the Moon Lake Walcott

18 24-hour ET southcentral Idaho August 14, 2000 Thousand Springs North Evapotranspiration (mm/day) Twin Falls recent burn basalt Wood River Valley ETr Fraction Burley Craters of the Moon Lake Walcott

19 Need for ET Maps in Idaho Quantify Net Depletion from Groundwater Pumping (unmeasured) Compare actual ET with Water Right Calculate Natural and Irrigation-Induced Recharge to Aquifers (via water balance to calibrate MODFLOW) Determine Actual ET for Developing better Crop Coefficient Curves

20 What Landsat sees Transmissivity of atmosphere Visible Near Infrared Wavelength in Microns: (Band 6 is the surface temperature band (not shown)) Various amounts of reflection Land Surface

21 Disposition of Solar Radiation in the Atmosphere H 2 O, O 2, O 3, N 2 O H 2 O, O 2, O 3, N 2 O Indirect Direct Solar

22 Longwave (Infrared) Radiation in the Atmosphere 4 H 2 O, CO 2, CH 4, CFC s

23 Broadband Surface Albedo Albedo by EBT-BBT NDVI less than 0.65 NDVI 0.65 or more Albedo by updated m ethod NDVI less than 0.65 NDVI 0.65 or more MODTRAN based albedo MODTRAN based Albedo EBT-BBT = old method used in METRIC until 2004 (based on broad-band transmissivity) Updated method: Transmissivity of individual bands τ C P W + C 2 air 3 4 in,b = C1 exp + C5 K t cosθh cosθh C

24 ERDAS Imagine Image Processing and Modeling System

25 METRIC Energy Balance METRIC is a sort of hybrid between pure remotelysensed energy balance and weather-based ET methods Combines the strengths of energy balance from satellite and accuracy of ground-based reference ET calculation: satellite-based energy balance provides the spatial information and distribution of available energy and sensible heat fluxes over a large area (and does most of the heavy lifting ) reference ET calculation anchors the energy balance surface and provides reality to the product.

26 Weather Data In METRIC, Weather Data are used for: Wind speed for sensible heat flux calculation Reference ET for Calibrating the Cold Pixel Reference ET to Extrapolate ET over: 24-hour period Days between Images

27 Weather Data from USBR AgriMet In METRIC tm applications, Alfalfa Reference ET r is computed using the hourly ASCE Standardized Penman-Monteith Equation

28 Sensible Heat Flux (H) H = (ρ c p dt) / r ah dt = floating near surface temperature difference (K). r ah = the aerodynamic resistance to heat transport (s/m). r ah = ln z z 2 ns Ψ u * h( z k ) + Ψ h( 2 z 1 ) z 2 z 1 dt r ah H u * = friction velocity k = von karmon constant (0.41)

29 Aerodynamic Equations used in METRIC tm H = ( ρ c p )(T a1 T a1 ) / r ah r Ψ ah = Ψ Ψ z ln z 2 ns Ψ u * h( z k ) + Ψ h( 2 z 1 ) u * = T a1 T a2 = f (T s ) ln 200 z 0m u 200 k Ψ Correction for atmospheric instability h 1 + = 2ln ( z2 ) x 2 ( z 2 2 ) 2 1+ x( 200m) 1 x 200m 2ln ln + ( ) = + 2ARCTAN x( 200m) 2 2 Ψ h ( z2 ) = z L 5 2 m( 200m) ( ) 0. 5π m + ( 200m) m ( 200m) = z L 5 2 and x ( height) ( height) = 1 16 L 3 T0 L Cpairu = ρ air * k g H for stable

30 Near Surface Temperature Difference (dt) To compute the sensible heat flux (H), define near surface temperature difference (dt) for each pixel z 2 z 1 dt r ah H Classical: dt = T surface T air SEBAL/METRIC: dt = T z1 T z2 T air is unknown and unneeded SEBAL and METRIC tm assume a linear relationship between T s and dt: dt = b + at s T s is used only as an index and can have large bias and does not need to represent aerodynamic surface temperature

31 METRIC tm -ERDAS submodel for sensible heat and ETrF

32 How METRIC tm is Trained METRIC tm is trained for each image by defining dt at the 2 anchor pixels: At the cold pixel: H cold = R n G -LE cold where LE cold = 1.05 λ ET r dt cold = H cold r ah / (ρ c p ) (in classical SEBAL, H cold = ~0 and T cold ~ T s is for water) At the hot pixel: H hot = R n G -LE hot where LE hot ~ 0 (if indicated by water balance) dt hot = H hot r ah / (ρ c p ) ( 1.05 is reduced and estimated as f(ndvi) during winter, etc.)

33 Hot Pixel Samples from an Agricultural Area SEBAL / METRIC Cold Pixel

34 ET r F at the Hot pixel: (is it really zero?): The operator must direct METRIC concerning any residual ET at the hot pixel. ET r F can be estimated using the FAO-56 surface evaporation estimation procedure Bare soil water balance, MRG, /1/2002 2/1/2002 3/1/2002 4/1/2002 5/1/2002 6/1/2002 7/1/2002 Kc based on ETr 8/1/2002 9/1/ /1/ /1/ /1/ Precipitation (mm)

35 Soil Heat Flux (G) via Bastiaanssen (1995): G/R n = T s ( α)(1 -.98NDVI 4 ) via Tasumi et al., (2003) from USDA-ARS data at Kimberly: G/R n = exp ( LAI) for LAI > 0.5 G/R n = 1.80 (T s 273)/ R n for LAI < 0.5 (~bare soil) and G = G/R n R n

36 Water Heat Flux (G) (Appendix 10) For clear, deep water Monthly average G vs. Rn measured in a deep Japanese lake (mean depth of 21m) (data from Yamamoto and Kondo, 1968). Monthly evaporation from three Great Lakes (Derecki, 1981)

37 Water Heat Flux (G) American Falls, Reservoir, Idaho 2004 midday readings based on combination of eddy covariance and REBS Bowen ratio measurements. Tasumi and Allen, 2004, preliminary data, University of Idaho units for R n, H, LE, G are W/m 2. Wind speed, WS, is in m/s. Kc = Evap / ET r

38 Water Heat Flux (G) American Falls, Reservoir, Idaho hr averages based on combination of eddy covariance and REBS Bowen ratio measurements. Tasumi and Allen, 2004, preliminary data, University of Idaho units for R n, H, LE, G are W/m 2. Wind speed, WS, is in m/s. Kc = Evap / ET r

39 Mr. SEBAL Wim Bastiaanssen

40 Interpolation to the rest of the day (and month) (and year)

41 ET r F = Fraction of ET r = K c Assumption: ET r F is consistent through the day ET, mm/hour Kimberly Lysimeters - August 4, Time of Day Sugar Beets ET ETrF ETr ETrF ET, mm/hour Kimberly Lysimeters - August 5, Time of Day Sugar Beets ET ETrF ETr ETrF ETo Daily Average ETrF ETo Daily Average ETrF ET, mm/hour Kimberly Lysimeters - August 6, Time of Day Sugar Beets ET ETrF ETr ETrF ET, mm/hour Time of Day Sugar Beets ET Kcr ETr ETo Kimberly Lysimeters - August 7, 1989 Daily Average Kcr Kcr ETo Daily Average ETrF

42 Extrapolation of Inst. to 24-hour ET : ET r F method - METRIC Sugar Beets and Potatoes ETrF for 1988 and 199 satellite dates 1.20 ETrF for Sugar Beets May to September, 1989, Data from Dr. J.L Wright 1.20 y = x R 2 = : ETrF at satellite time hour ETrF :1 line Average Daily ETrF Instantaneous ETrF at 11:00 am ET r F based on Lysimeter Data by Dr. J.L. Wright, USDA-ARS

43 24-Hour Evapotranspiration (ET 24 ) ET = ET F ET 24 r r _ 24 Path 39: Am. Falls - 24-hour ET 8/07/00 (in EF method, ET 24 = EF x R n24 )

44 Seasonal ET > 1000

45 Comparison with Lysimeter Measurements: Lysimeter at Kimberly (Wright) 12/17/01

46 Kimberly, Idaho Periods between Satellites ET during period, mm Impact of using Kc from a single day to represent a period: Kimberly 1989 Sugar Beets, 1989 Kimberly, Idaho 0 04-May 21-Jun 20-May 07-Jul 18-Apr 05-Jun 23-Jul 25-Sep Lys. Kc on Sat. date x sum ETr Sum. all lysimeter meas. (Truth) SEBAL METRIC ET ET for for period Lysimeter data by Dr. J.L. Wright, USDA-ARS

47 Seasonal ET Cumulative ET in 1989 for Sugar Beets 800 Error = 2.5% /1/89 4/15/89 4/29/89 5/13/89 5/27/89 6/10/89 6/24/89 7/8/89 7/22/89 8/5/89 8/19/89 9/2/89 9/16/89 9/30/89 Cumulative ET (mm) from 4/1/89 SEBAL-ID Estimation Lysimeter Measurement

48 Comparison of Seasonal ET by METRIC tm with Lysimeter ET (mm) - April-Sept., Kimberly, 1989 Sugar Beets Lysimeter 718 mm Total METRIC 714 mm Lysimeter SEBAL METRIC

49 Comparison of Seasonal ET by SEBAL 2000 with Lysimeter ET (mm) - July-Oct., Montpelier, ID Lysimeter 388 mm Total SEBAL 405 mm Lysimeter SEBAL

50 Annual ET for all of California Created by SEBAL- North America for 2002 using MODIS satellite imagery (resolution = 1 km)

51 Comparison to K c Curves

52 Use to Refine Local K c Curves Potato 717 fields in the Twin Falls area Kc Average curve NDVI Potato Vegetation Index Day of Year Day of Year

53 S.Beet 516 fields K c near 1.0 indicating high production agriculture Kc Day of Year

54 W.Grain 564 fields Kc Day of Year

55 Alfalfa 325 fields Kc Day of Year

56 1.2 1 Comparison with Local K c Curves -- south-central Idaho Agrimet K c s are based on Wright (1981) (lysimeter) Potato Sugar Beet Agri Met Corn 1 Bean /27/00 10/27/00 3/1/00 3/31/00 4/30/00 5/30/00 6/29/00 7/29/00 8/28/00 9/27/00 10/27/00 3/1/00 3/31/00 4/30/00 5/30/00 6/29/00 7/29/00 8/28/00 Kc 0.6 Kc 3/1/00 3/31/00 4/30/00 5/30/00 6/29/00 7/29/00 8/28/00 9/27/00 10/27/00 3/1/00 3/31/00 4/30/00 5/30/00 6/29/00 7/29/00 8/28/00 Kc /27/00 10/27/00 Kc EB model Allen&Brockway Little need to adjust K c Little need to adjust K c Little need to adjust K c

57 Crop Coefficient (ETrF) Potato Fields, June NDVI

58 Crop Coefficient (ETrF) Potato Fields, June 19 June NDVI

59 Crop Coefficient (ETrF) Potato Fields, June 19 June 21 July NDVI

60 Crop Coefficient (ETrF) Potato Fields, 2000 basal K c June 19 June 21 July NDVI

61 Crop Coefficient (ETrF) Potato Fields, 2000 mean K c basal K c June 19 June 21 July NDVI

62 mean K c vs. NDVI Kc (ETrF) May to September 2000 Magic Valley, Idaho Averages of 100's of fields each satellite date Well-watered fields NDVI (toa) Alfalfa Sugar Beet Corn Potato S.Grain W.Grain ETrF = 1.25 NDVI

63 Application of ET maps

64 Actual ET from wetlands and riparian systems Wetland Boise Valley Seasonal ET 2000

65 Boise River Valley, Idaho ET BY LAND USE CLASS Benefit: New Information Cost: +$70,000 Class Name ET in mm Petroleum Tank Yards 237 Rangeland 242 Unclassified 298 Barren 335 Commercial / Industrial 380 Transportation 420 Idle Agriculture 436 Abandoned Agriculture 459 Junk Yard 467 Feedlot 479 Dairy 524 Other Agriculture 536 Public 548 Sewage 552 New Subdivision 606 Farmstead 609 Rural Residential 657 Urban Residential 684 Canal 731 Irrigated Crops 812 Perennial 820 Recreation 826 Water 924 Wetland 1,025

66 Boise River Valley, Idaho ET BY LAND USE CLASS Benefit: New Information Cost: + $70,000 Class Name ET in mm Petroleum Tank Yards 237 Rangeland 242 Unclassified 298 Barren 335 Commercial / Industrial 380 Transportation 420 Idle Agriculture 436 Abandoned Agriculture 459 Junk Yard 467 Feedlot 479 Dairy 524 Other Agriculture 536 Public 548 Sewage 552 New Subdivision 606 Farmstead 609 Rural Residential 657 Urban Residential 684 Canal 731 Irrigated Crops 812 Perennial 820 Recreation 826 Water 924 Wetland 1,025

67 Middle Rio Grande Region of New Mexico Consumptive use of irrigated agriculture Small holdings Less than full water supply in areas Some areas of water logging Interest in nonbeneficial consumptive use Interest in ET by riparian systems

68 Landsat true-color image (left) and landuse map (right) showing distribution of cottonwood and salt cedar land-use types along the MRG river near Isleta. The bottom set is a close-up of the top. Water Natural Vege. Agri. Vege. City Forests Dark Soil Bright Soil Cottonwoods Salt Cedar Willow Russian Olive Basalt Rock High Resolution Classification courtesy of Dr. Christopher Neale Utah State University

69 Middle Rio Grande Region of New Mexico Alfalfa* - MRG, ETrF ETrF Open Water - MRG, /1 2/1 3/1 4/1 5/1 6/1 7/1 8/1 9/1 10/1 11/1 12/1 * Sample fields were alfalfa in 2004, but might be other crops in /1 2/1 3/1 4/1 5/1 6/1 7/1 8/1 9/1 10/1 11/1 12/1 Alfalfa #1 Alfalfa #2 Alfalfa #3 Alfalfa #4 Alfalfa #5 Alfalfa #6 Alfalfa #7 Alfalfa #8 Cochiti Reservoir Heron Reservoir Abiquiu Reservoir Salt Cedar - MRG, 2002 ETrF ETrF Cottonwood - MRG, /1 2/1 3/1 4/1 5/1 6/1 7/1 8/1 9/1 10/1 11/1 12/ /1 2/1 3/1 4/1 5/1 6/1 7/1 8/1 9/1 10/1 11/1 12/1 cottonwood #1 cottonwood #2 cottonwood #3 cottonwood #4 cottonwood #5 cottonwood #6 cottonwood #7 cottonwood #8 Salt Cedar #1 Salt Cedar #2 Salt Cedar #3 Salt Cedar #4 Salt Cedar #5 Salt Cedar #6 Salt Cedar #7 Salt Cedar #8

70 Frequency Distribution of ET 15,000 acres of cottonwood and salt cedar June ET ( m m /m o.) A n n u a l E T ( m m ) Relative area Relative area Cottonwoods Saltcedar Cottonwoods Saltcedar

71 Estimated water consumption by class of riparian vegetation within the riparian area between San Acacia and Cochiti during 2002* Total area (acres) Annual ET r F (K c ) Annual ET (mm) Annual Water Consumption (AF) Cottonwood 10, ,000 Salt Cedar 4, ,000 Willow R. Olive *Assumes constant ETrF (i.e., ET/ETr) during the day From report by University of Idaho (Allen et al., 2004) to Keller-Bliesner Engineering, Logan, UT for U.S. Department of Justice High Resolution Classification courtesy of Dr. Christopher Neale Utah State University

72 ET r F for Albuquerque 1.2 Albuquerque City Area Average ETrF Month

73 Performance of Irrigation Projects

74 Irrigation Project Performance -- Idaho 0.7 Project wide Crop Coefficient -- METRIC Twin Falls Tract ,000 acres -- Alfalfa Reference Basis 0.6 K c Mar Apr May Jun Jul Aug Sep Oct March, Sept., and Oct. unavailable for 2003

75 Irrigation Project Performance -- Idaho 1.0 Twin Falls Canal Company, Idaho Evapotranspiration as a Ratio of Diversion plus Precipitation 0.8 Ratio Apr May Jun Jul Aug Apr-Aug

76 Conclusions ET maps are valuable for: Determining Actual ET Refining Crop Coefficient Curves (f.e. for a new ET reference) Water Rights Conflicts Ground-water Management Consumption by Riparian Vegetation ET maps by METRIC tm have good accuracy and consistency with the Reference ET approach Kc 1.2 Sugar Beet /1/00 3/31/00 4/30/00 5/30/00 SEBAL Allen&Brockway (1983) 6/29/00 7/29/00 8/28/00 9/27/00 10/27/00

77 Requirements for SEBAL or METRIC tm Satellite images with Thermal Band Higher resolution (Landsat) is needed for field scale maps Good quality weather data if local calibration is desirable Experienced, thinking human at the controls

78 More information at: (METRIC tm ) (SEBAL tm )

79

80 METRIC with MODIS

81 Landsat5: 8/31/2003 MODIS: 8/31/2003 (57 o ) ET r F= 0.40 std.dev.= 0.36 ET r F= 0.46 std.dev.= 0.20 MODIS: 8/9/2004 (2 o ) ET ET r F = ETrF ET r ET r F= 0.37 std.dev.= 0.25 ET r F by METRIC

82 MODIS: 8/31/2003, Level2 5min LST Level 2 5min LST represents a daily instantaneous LST image. White locations are missing LST due to missing thermal emissivity due to mis-categorization of pixel as a cloud near irrigated/desert boundaries. Daily LST products have these fall-outs also, but the number of the holes is less, primarily because of the smearing effect of resampling. Currently, METRIC uses Level 1 radiance (using calibration of LST vs. radiance by regression of good pixels in LST image). Level 3 8-day LST may also be useable

83 Another problem with MODIS Sensors: Band 5 (2 nd NIR) has been having striped output. Stripes are present in all Level 1 3 products. This problem has been reported for many bands as known issues. MODIS: 8/29/2003~8day MODIS: 8/9/2004 (2 o ) These two images false color using Bands Solution: Omit band 5 from albedo computations.

84 MODIS (Terra and Aqua) 1400 km 60 o 705 km 2400 km Earth 6400 km

85 Landsat 7 o max 705 km 160 km Earth 6400 km

86

87 Scan Angle (from Zenith), Twin Falls, Idaho August 1-16, MODIS Sensor Scan Angle, Deg from Zenith Aqua Terra Day of Year

88 Scan Angle Effect on Reflectance, S. Idaho

89 Estimated 24 hour ET (mm/day), 7/21/2000, path 40/30, Agr. Area Only Model Sensitivity To Correction of Surface Temperature for Atmosphere Estimated ET using uncorrected Ts y = x R 2 = Estimated ET using corrected Ts (Predictions are not sensitive due to calibration at hot and cold pixels)

90 to Aerodynamic Roughness, zom, (of all areas) Model Sensitivity --- Agricultural Areas (mm/day) if zom is double Est. ET Estimated 24 hour ET (mm/day), 7/21/2000, Agricultural Area y = x R 2 = (mm/day) if zom is half Est. ET Estimated 24 hour ET (mm/day), 7/21/2000, Agricultural Area y = x R 2 = Estimated ET with original z om (mm/day) Estimated ET with original z om (mm/day) (Predictions are not sensitive due to calibration at hot and cold pixels)