Evapotranspiration Derived from Satellite Observed Surface Temperatures

Abstract

Evapotranspiration is calculated from surface temperatures using an energy balance method. This method is sensitive to the temperature difference between the surface and the air above, and somewhat to the windspeed. In this study we consider the influence of the spatial variability of air temperature on the interpretation of surface temperatures. It is argued that small-scale atmospheric variations can be corrected by using temperature and wind data at a height of 50 m. Three models have been used to calculate these 50 m data from standard weather observations. The first model uses a very simple concept of constant temperature and wind over the test area (zero-dimensional). In the second model windspeed is also taken constant, but air temperature is evaluated from the initial vertical temperature in the atmosphere with a one-dimensional slab layer model. The third model is a two-dimensional primitive equations model in which wind velocity is calculated from the geostrophic wind and air temperature similar to model 2. A homogeneous grassland area was selected in the north of the Netherlands close to the sea. Several days in the summer of 1983 with clear skies and winds from the sea were selected. Surface temperatures were derived from the NOAA-7 satellite overpass in the early afternoon using the split-window technique. On most days an almost linear increase of both surface and air temperature is found with increasing distance to the sea. This study reveals that model 1 results in an unrealistic decrease of the calculated evapotranspiration with increasing distance to the coast. Furthermore evapotranspiration is underestimated. The evapotranspiration as calculated with models 2 and 3 is almost constant in the test area and agrees well with measurements. Model 3 gave more scatter, probably due to the fact that uncalibrated wind velocities were used. For practical calculation of evapotranspiration the Priestley-Taylor parameter α is often used. This study shows how this parameter can be derived from satellite observations of surface temperature.