|Hain, Christopher -|
|Chavez, Jose -|
Submitted to: Advances in Water Resources
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: June 8, 2012
Publication Date: December 1, 2012
Repository URL: http://handle.nal.usda.gov/10113/59978
Citation: Anderson, M.C., Kustas, W.P., Alfieri, J.G., Gao, F.N., Hain, C., Prueger, J.H., Evett, S.R., Colaizzi, P.D., Howell, T.A., Chavez, J. 2012. Mapping daily evapotranspiration at Landsat spatial scales during the BEAREX'08 field campaign. Advances in Water Resources. 50:162-177. Interpretive Summary: Daily and seasonal information about evapotranspiration (ET) distributions over agricultural landscapes, generated at scales resolving individual fields, can promote more efficient use of water resources. The need for such information is critical in agricultural areas in the semi-arid southwestern U.S. – the site of the Bushland Evapotranspiration and Agricultural Remote sensing EXperiment of 2008 (BEAREX08). In this area, groundwater extractions to support irrigated agriculture are leading to significant and unsustainable declines in water levels in the Ogallala Aquifer. Satellite remote sensing can help to quantify rates of extraction and consumptive water use over large regions, and can benefit growers practicing water conservation techniques. The Landsat satellite provides the satellite information required to run ET models at 100m resolution, adequate for resolving water use on a field-by-field basis. However, the Landsat overpass frequency (16 days) is not optimal for developing estimates of cumulative water use over a growing season. In this paper, a simple technique for interpolating ET between Landsat imaging dates is explored, conserving the ratio of actual-to-potential ET. This simple technique is geared toward application in data sparse regions, requiring no local observations of weather or rainfall. The technique reproduces cumulative water use over the BEAREX08 campaign site to within 10% of the measured value. Next steps in this research will be to integrate coarse yet more frequent imagery from other satellites to improve assumptions about soil moisture between Landsat overpasses.
Technical Abstract: Robust spatial information about environmental water use at field scales and daily to seasonal timesteps will benefit many applications in agriculture and water resource management. This information is particularly critical in arid climates where freshwater resources are limited or expensive, and groundwater supplies are being depleted at unsustainable rates to support irrigated agriculture as well as municipal and industrial uses. Gridded evapotranspiration (ET) information at field scales can be obtained periodically using land-surface temperature-based surface energy balance algorithms applied to moderate resolution satellite data from systems like Landsat, which collects thermal-band imagery every 16 days at a resolution of approximately 100 m. The challenge is in finding methods for interpolating between ET snapshots developed at the time of a clear-sky Landsat overpass to provide complete daily time-series over a growing season. This study examines the efficacy of a simple interpolation algorithm that does not require local ground measurements of weather or rainfall, designed for applications in data-sparse regions. The algorithm relies on general conservation of the actual-to-potential ET ratio between overpass dates. Ratio maps are computed on clear Landsat dates, and then these ratios are applied to daily maps of potential ET (PET) during the interim. The algorithm is tested with ET retrievals generated using the Atmosphere-Land Exchange Inverse (ALEXI) surface energy balance model and associated DisALEXI flux disaggregation technique, which uses Landsat-scale thermal imagery to reduce regional ALEXI maps to a finer spatial resolution. Daily interpolated ET is compared with lysimeter and eddy covariance flux measurements collected during the Bushland Evapotranspiration and Agricultural Remote sensing EXperiment of 2008 (BEAREX08), conducted in an irrigated agricultural area in the Texas Panhandle. Two PET datastreams are evaluated: one based on micrometeorological measurements collected locally at a grass reference site, and another using purely remotely derived data from mesoscale analyses and satellite-based radiation products. Both datastreams performed reasonably, reproducing cumulative ET to within 10-15% over the growing period from emergence to peak biomass. In this case, the Landsat 5 overpass timings coincided with troughs in the observed ET time curve, which would lead to underestimates of cumulative ET even if the retrievals were perfect. Impact of frequency and phase of Landsat overpass was examined to assess optimal specifications for future water resource mapping missions.