Estimation of Large-Scale Evaporation Using Remote Sensing I Posters
Presiding: M Jasinski, NASA Goddard Space Flight Center; B Kustas, USDA Hydrology and Remote Sensing Laboratory
H23B-01 1330h
Estimating Land Surface Fluxes Using Microwave and Thermal Remote Sensing Data During SMEX02/SMACEX
A two-source (soil + vegetation) energy balance model using microwave-derived near-surface soil moisture (TSMSM) as input was applied to a corn and soybean production region in central Iowa. Six days of the Polarimetric Scanning Radiometer (PSR) derived soil moisture data and Landsat derived vegetation information as well as local meteorological data were used to run the model. These data were acquired during the Soil Moisture Experiment in 2002 (SMEX02) and the Soil Moisture Atmosphere Coupling Experiment (SMACEX). The PSR derived soil moisture maps at 800 m resolution which were resampled to 30 m. At this higher resolution, a flux footprint model was applied to weight pixels within the source area of the flux tower measurements. The root mean square difference (RMSD) values between TSMSM estimated and tower measured net radiation, Rn, and soil heat flux, G, were within 25 Wm-2. The TSMSM model also produced reasonable estimates of sensible heat (H) and latent heat flux (LE) with both RMSD values for H and LE being within 55 Wm-2. The TSMSM model output was also compared with estimates from the two-source model version using radiometric surface temperature observations (TSMTH) from Landsat. The results from two Landsat overpasses under partial canopy cover on July 1, 2002, and near full cover on July 8 indicate that TSMSM gave similar results (slightly better) to TSMTH for July 1, but did not perform as well for July 8. When the two models were in good agreement, the surface temperature estimated from TSMSM agreed closely with the Landsat radiometric temperature observations. By contrast, when TSMSM output gave greater differences with TSMTH flux estimates and tower measurements such as July 8, the TSMSM model generally computed lower LE and higher H than TSMTH and the tower measurements. This resulted in TSMSM producing significantly higher surface temperatures than Landsat radiometric temperature observations. Both soil moisture and fractional vegetation cover/leaf area greatly affect TSMSM output. However, it is not clear what factors caused TSMSM to overestimate H and underestimate LE for July 8. This investigation also showed that by using the higher resolution soil moisture and vegetation cover information, the model output more closely matched tower measurements. This is most likely the result of accounting for the effects of spatial variation in the vegetation cover and application of a flux footprint model for estimating the appropriate source area from model output to compare with tower based flux measurements.
H23B-02 1330h
Validating Patterns in Large Scale Sensible and Latent Heat Flux Estimates from a Remote Sensing-based Model and Aircraft-based Flux Measurements
A remote sensing field experiment conducted in the Southern Great Plains in 1997 (SGP97) in central Oklahoma, had aircraft-based flux observations as well as remotely sensed data collected over one of the main study sites in central Oklahoma. This agricultural region contains primarily grassland/pasture and winter wheat, which was recently harvested leaving a significant number of fields either as wheat stubble or plowed bare soil. Multi-spectral data obtained by aircraft provided high resolution (30 m) spatially-distributed vegetation cover and surface temperature information over an area approximately 10 km north-south by 30 km east-west. The spatial variations in these surface states strongly affect the partitioning of surface fluxes between sensible and latent heat. These data, together with coarser resolution (5 km) satellite data, are used in a remote sensing-based energy balance modeling system that disaggregates flux estimates to the 30 m resolution. From the aircraft-based measurements collected along the 15 km transect, "segmented" flux values over 1 km sampling intervals were computed, which were then sub-sampled using a 250 m moving window using a new scheme for estimating time-space dependence of aircraft surface fluxes. From these two estimates of the large scale heat flux patterns, a comparison is made for exploring consistency in flux distributions. This type of comparison involves estimation of the flux-footprint or source area for the aircraft flux observations in order to weight the upwind model pixels within the aircraft sensor footprint. Highest correlation between aircraft and modeled estimated heat and water vapor fluxes were obtained using different flux-footprint estimates with the source-area for heat estimated to be much closer to the aircraft flight line than for water vapor. Factors that may be contributing to these results are discussed.
H23B-03 1330h
Bulk Surface Momentum Parameters for Satellite-Derived Vegetation Fields
Most numerical atmospheric simulation models and land surface hydrology models used today require knowledge of aerodynamic roughness in the parameterization of surface fluxes. In this paper, the bulk parameters associated with the absorption of surface momentum by vegetated landscapes are theoretically estimated within the context of Raupach's roughness sublayer formulation and the canopy area index. The estimated parameters include the bulk plant drag coefficient, maximum u*/uh, sheltering coefficient, and canopy area density at onset of sheltering. The test case includes the U.S. Southern Great Plains, that includes four principal IGBP land classes; Evergreen needleleaf, grassland, crops, and savanna/shrub. The estimation approach uses the Method of Moments and roughness data from several international field experiments and other published data. The above procedure provides a physically based approach for estimating roughness length for seasonally variable vegetation fields using satellite data.
H23B-04 1330h
Estimating Evapotranspiration Using a Remote Sensing Driven Water Balance
Evapotranspiration is difficult to estimate on regional, climatic scales. One approach is to use a water budget equation, i.e., total precipitation minus the sum of evapotranspiration and net runoff equals the change in terrestrial water storage. Satellite observations from GRACE, TRMM, and future precipitation and surface water missions will enable evapotranspiration to be computed as a residual of a careful, remote sensing based water budget. Here we describe the method for estimating evapotranspiration using data from GRACE along with observation based precipitation and runoff, which takes into account the unique nature of the GRACE observations. Evapotranspiration is estimated over the Mississippi River basin and compared with output from the land surface model and two operational atmospheric modeling systems. Results suggest that the new technique provides skill in evaluating modeled evapotranspiration, particularly in terms of bias.
H23B-05 1330h
EFFECT OF SCALING TRANSFER BETWEEN EVAPOTRANPIRATION MAPS DERIVED FROM LANDSAT7 AND MODIS IMAGES
Remotely sensed images of the Earth's surface provide information about the spatial distribution of evapotranspiration. Since, the spatial resolution of evapotranspiration predictions depends on the sensor type; scaling transfer between images of different scales needs to be investigated. The objective of this study is to investigate the effect of the scaling transfer between evapotranspiration maps derived from Landsat7 and MODIS images. In this study, the Surface Energy Balance Algorithm for Land (SEBAL) was used to derive evapotranspiration maps from Landsat7 Thematic Mapper (TM) and Moderate Resolution Imaging Spectroradiometer (MODIS) images. LandSat7 has a spatial resolution of 60 m, MODIS of 1000 m. Two up-scaling procedures are evaluated. The first consists of averaging 60 by 60 m LandSat pixels of radiances to obtain 1000 by 1000 m pixels at the MODIS scale before SEBAL is applied. The second consists of first applying SEBAL and then to average evapotranspiration rates from 60 m to 1000 m spatial resolution. The averaging process (aggregation) will include calculating arithmetic and geometric means. In the down-scaling process (MODIS scale to LandSat scale), a 16 days earlier or later LandSat image will be used to characterize the fine scale variability within the large MODIS pixels. Two down-scaling procedures are evaluated. The first consists of down-scaling the original MODIS radiance data; the second of down-scaling the evapotranspiration maps at MODIS scale. The results of this study show that up- and down-scaled ET maps over the Middle Rio Grande Basin show a relatively good agreement with evapotranspiration maps directly derived from LandSat and MODIS images.
H23B-06 1330h
Estimating Regional Evapotranspiration in the Amazon Basin from the Atmospheric Water Budget
Estimates of regional evapotranspiration (ET) over the Amazon basin have been found to be highly dependent on the parametrization of transpiration. The inadequately understood and difficult to model vegetation control over ET under different conditions of energy and water availability leaves us with divergent estimates of the annual ET cycle in the basin. To avoid this problem, we estimate regional ET over the Amazon basin from the atmospheric water budget. The control volume is defined to be the atmosphere overlying the basin and the variables involved are area-averaged ET, precipitation (P), vertically integrated moisture convergence (Q), and change in the water vapor content of the atmospheric column (dW/dt). Significant differences exist between measures of these water budget components obtained from different data sources, reflecting the uncertainty of available data on precipitation, wind speed and atmospheric moisture. To reduce the uncertainty in the resulting ET estimate, we incorporate multiple estimates of each budget term in a constrained least squares estimator, weighting each estimate according to its uncertainty and constraining the optimization by the water balance equation: dW/dt= Q+ET-P. Data on windspeed and atmospheric moisture critical to this water budget study are obtained from various global reanalyses: NCEP-NCAR Reanalysis, NCEP-DOE Reanalysis, and the ECMWF Reanalysis ERA-40. Data on atmospheric moisture from the NASA Water Vapor Project (NVAP) is also utilized. TRMM and GPCP data products are used to obtain precipitation estimates. ET estimates from different land-surface models are also incorporated into the estimator. We explicitly account for the uncertainties associated with the available measures of the water budget components, and study their effect on the derived ET value and its error bounds. This approach is tested over the 5-year period 1997-2001. We evaluate our ability to resolve patterns of moisture fluxes over the basin and associated ET variations at the sub-monthly, seasonal and interannual timescales.
H23B-07 1330h
Mapping Energy Balance Fluxes in Riparian Areas Using SEBAL
Accurate information on the distribution of evapotranspiration in arid riparian areas is needed for sustainable management of water resources as well as for a better understanding of water exchange processes between the land surface and the atmosphere. In this study, the Surface Energy Balance Algorithms for Land (SEBAL) was selected to determine the energy balance fluxes in the riparian areas of the Middle Rio Grande Basin (New Mexico), San Pedro River (Arizona) and Owens Valley (California). The objective is to compare SEBAL energy balance fluxes derived from LandSat TM images with those measured on the ground with eddy covariance towers. Good agreement was found between energy fluxes determined with SEBAL from LandSat TM images and those measured on the ground.
http://www.ees.nmt.edu/EPSCoR/
H23B-08 1330h
Evapotranspiration on Western U.S. Rivers Estimated by Remote Sensing and Eddy Covariance Flux Tower Data
Evapotranspiration (ET) rates measured from eight eddy covariance flux towers on three western United States rivers were highly correlated with Enhanced Vegetation Index (EVI) values from Moderate Resolution Imaging Spectrometer sensors on the NASA Terra satellite. Sixteen-day composite values of EVI and maximum daily air temperature (Ta) were combined to predict ET across species and sites (r2 = 0.76). The relationship was then used to estimate ET for 2000-2004 over large river stretches on the Upper San Pedro River, the Middle Rio Grande, and the Lower Colorado River. EVI and ET values were similar across river systems. Measured and estimated ET values tended to be moderate when compared to earlier, and often indirect, estimates, and ranged from 850-1,060 mm yr-1. EVI for individual plant associations, used as a measure of relative ET rates, ranked cottonwood (Populus spp.) and willow (Salix spp.) highest, mesquite (Prosopis glandulosa) and saltcedar (Tamarix ramosissima) intermediate, and giant sacaton (Sporobulus wrightii) and arrowweed (Pluchea sericea) lowest in potential ET. However, saltcedar EVI had a high variance, as this species can grow in sparse to dense stands, depending on water availability. ET rates estimated by remote sensing in this study produced similar values as direct, ground-based measurements on the San Pedro River, but they were much lower than official values estimated for riparian water budgets using crop coefficient methods for the Middle Rio Grande and Lower Colorado River.
H23B-09 1330h
A Hybrid Surface Energy Balance Approach for Large Scale Evapotranspiration Estimation and Prediction in Agricultural Areas
Over the last few years, several surface energy balance methods for the estimation of latent heat fluxes from remotely sensed satellite imagery have been introduced and/or refined. These models have shown the ability of obtaining seasonal spatially distributed evapotranspiration fluxes at various scales and over large areas. In the arid western United States, water managers are challenged in balancing the high consumptive use of irrigated agriculture with competing urban and ecological uses of fresh water. Water managers from Irrigation Districts and Federal Agencies such as the US Bureau of Reclamation have a need for improved operational tools for the prediction of evapotranspiration and irrigation water demand on a five to ten day timeframe. The paper will present a hybrid model that couples the surface energy balance approach with a simple empirical reflectance-based crop coefficient model, for estimation and prediction of evapotranspiration over large agricultural areas. The model is applied to a rain-fed intensively cultivated agricultural area, close to Ames, Iowa during the summer of 2002. The satellite, airborne and ground fluxes were collected during the SMACEX 02 experiment. The model is run in both simulation and prediction mode and the derived latent heat fluxes are compared spatially and temporally to aircraft derived fluxes from the USU airborne system and ground measured fluxes at thirteen eddy covariance stations, using appropriate upwind footprint source area functions.
H23B-10 1330h
Actual and Apparent Evapotranspiration in the Environment: A Study Toward the Resolution of the Evaporation Paradox
Over the past 50 years evaporation from pans has been shown to be decreasing. Until recently, pan evaporation was widely accepted to be proportional to landscape evapotranspiration. This decrease stands out against data that show an increase in global precipitation. This paradox can be explained using Bouchet's complimentary relationship. At the root of the paradox is that evaporation pans measure apparent potential evaporation. Apparent potential evaporation is only equal to the actual potential evapotranspiration (and actual evapotranspiration) under moist conditions. According to the complimentary relationship, evapotranspiration from a landscape will decrease if the conditions drop below the potential level (i.e. the land dries out). This decrease will liberate available energy that will accelerate evapotranspiration from an evaporation pan (or any other moist surface). We present a comparison of actual evapotranspiration measured at the First ISLSCP Field Experiment (FIFE) in the Konza Prairie, Kansas and at the Atmospheric Radiation Measurement (ARM) program in the Washita River Basin, Oklahoma with apparent potential evaporation data from nearby evaporation pans compiled by the National Climactic Data Center (NCDC). The data generally support the theory underlying the complimentary relationship.
H23B-11 1330h
Variations of Oceanic Surface Latent Heat Flux from Remote Sensing Data Sets
We examined variations of ocean surface latent heat fluxes from three remote sensing techniques: the Goddard Satellite Surface Turbulence/Flux version 2 (GSSTF2), the Japanese Ocean Flux Data Set with Use of Remote Sensing Observations (J-OFURO) and the Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite Data version 2 (HOAPS2). The global average latent heat fluxes are 102.9 W/m2 and 96.7 W/m2 for GSSTF2 and HOAPS, for the period 1998-1992, respectively. A global increasing trend has been found for all three data sets for the period 1992-2000, the global averages are 102.7 W/m2, 114.3 W/m2, and 99.6 W/m2, with increases of 9.4%, 13.0% and 7.3%, for GSSTF2, J-OFURO and HOAPS2, respectively. Empirical Orthogonal Function analyses were performed to examine the nonseasonal variations. The second EOF of GSSTF2, J-OFURO, and HOAPS2 correspond to an ENSO mode, which correlates with an SOI at 0.74, 0.71 and 0.59, respectively. The first EOF mode of all the three data sets is characterized by positive anomalies in most areas, especially in the subtropics, with opposite changes in the equatorial Pacific and maritime continent. The pattern correlations between the first EOFs of these data sets are 0.50, 0.55, and 0.37 for GSSTF2 and J-OFURO, GSSTF2 and HOAPS2, and J-OFURO and HOAPS2 respectively. If only the area within 30 degree is considered, the corresponding correlations are 0.50, 0.53 and 0.52. This mode is interpreted as the decadal variation of the Hadley circulation which started in the early nineties. If the first EOF mode is removed, the corresponding increases from linear regression are 2.2%, 7.3%, and close to zero, respectively.
H23B-12 1330h
Reducing Uncertainties in Estimates of Evapotranspiration Using only Remotely Sensed Data, Three Gorges Region of China
Accurate water and energy budgets depend on high quality estimates of local and regional evapotranspiration. Ground measurements are rare for many under-gauged regions of the world, so the use of remote sensing data to determine these estimates is especially appealing. Remote sensing offers greater spatial coverage (i.e. global) than traditional measures, but has its limitations (e.g., may have less-frequent temporal sampling, and the length of such datasets is limited to the launch of appropriate satellites). Additionally, there are uncertainties associated with methods for deriving evapotranspiration estimates. In this study, we apply the method by Jiang and Islam (2001) of combining the Priestley-Taylor equation and NDVI-Surface Temperature relationship to estimate the evapotranspiration (ET) for a large area using remote sensing information (e.g., MODIS data). ET estimates from this approach are compared to the potential evapotranspiration data measured by two methods (small and large pans) on the ground. Evaluations of the validity of the method by Jiang and Islam and other relevant methods for the study region will be provided, with an effort to reduce methodological uncertainties. Estimates of evaporation at sixteen-day intervals and 1 km resolution are determined for over one year. Our estimates are unique by avoiding the use of flux towers, and can provide accurate predictions of evapotranspiration using only remotely sensed data. The large area to be studied is in the Three Gorges Region of China, a region undergoing rapid environmental, land-use, and hydrological change. As the Three Gorges Dam undergoes filling, accurate assessments of the region's hydrology have important implications for its water quality, public health, and economy.