Estimating Distributed Soil Evaporation to Account for Background Evaporation Between Satellite Image Dates

Satellite based surface energy balance models are now routinely operated to produce evapotranspiration (ET) products on an operational basis for use in water resources management. Landsat satellite imagery is commonly used to produce estimates of ET at field scale with energy balance methods because of the onboard thermal imager and the high spatial resolution. Two Landsat satellites are currently in operation, Landsat 5 and Landsat 7, each flying in the same orbit but eight days apart. Because of cloudiness not all images or portions of images are suitable for processing. Although areas having cloud cover vary between different dates, it is common to find only one good image per month suitable for processing. As a result, monthly and ultimately seasonal ET depths are generally based on only one “snapshot” of ET per month. In METRIC (Mapping Evapotranspiration at high Resolution using Internalized Calibration), monthly ET depths are determined by interpolating a relative ET fraction (ETrF, where ETrF = ET / ETr, and ETr is tall reference ET computed from weather data) determined by energy balance for the image dates to all days falling between image dates using a cubic spline function and multiplying by the corresponding daily reference ET. A potential shortfall in basing integrated ET averages on periodic snapshots from satellite is that local or regional precipitation events antecedent to the satellite images may unduly dominate the ETrF image and may not represent evaporation from rainfall averaged over the monthly period. In addition, some rain events may occur in between satellite images that are not ‘seen' in a subsequent image, and therefore those evaporation amounts are not fully accounted for. We have developed a soil-water balance procedure to adjust the ETrF derived from the satellite overpass date and the METRIC surface energy balance model to account for background evaporation from soil caused by rainfall over monthly or longer integration periods. The result of the adjustment is an ETrF image to be used during the spline procedure and consequently final ET map that better represents the average evaporative conditions over the period. The method currently uses the FAO-56 evaporation model, but with an enhancement for skin evaporation, on a grid covering the image area. As input, information about the distributed ETr and precipitation for the study area is needed. The ETrF for each image pixel is adjusted by the ratio of the average evaporation over the month from bare soil to the evaporation estimated for the image date. This ratio is moderated using the Normalized Difference Vegetation Index (NDVI) to account for shading effects of vegetation on reducing the signature of evaporation from soil. An example of the correction is shown in Figures 1. This procedure is simple and straightforward to apply and will be used until more sophisticated daily process models based on the Penman-Monteith model, fitted to each image pixel, and spun up and corrected each clear image date, are ready for routine application. Figure 1. ET estimates from METRIC from August 13 1997 before (left) and after (right) adjustment of background evaporation. The ET estimates are shown as a fraction of alfalfa-based reference ET, ETrF.