ESTIMATION OF EVAPOTRANSPIRATION OF GREEN LANDSCAPES IN AL AIN, UNITED ARAB EMIRATES USING NORMALIZED VEGETATION INDEX AND GROUND SURFACE METROLOGICAL DATA

Evapotranspiration is a major source of water depletion in arid and semiarid environments; and it is a poorly quantified variable in the hydrological cycle. Remote sensing is expected to have potential application to quantify this variable especially at large scale. The present study reported information and methodology useful to determine whether remotely sensed vegetation indices could be used to calculate evapotranspiration (Et). To achieve this goal, various regression analyses were conducted between data from satellites, climatological factors and Et values. In this study solar radiation and associate ambient temperature values were obtained from local metrological station. However, linear regressions between temperature and solar radiations were obtained to simulate the solar radiation in Al Ain, for some of the missing dates. The predicted and actual solar radiation values were comparable with each other and with the reported published data. The solar radiation values together with wind speed, dew point were obtained from online data of Al-Ain metrological station, maximum and minimum temperature were obtained onsite and then all the data were processed with ASCE-EWRI (2004) method to obtain Et. Selected sub-scenes of Landsat ETM+ images free of cloud were used to derive NDVI using ER-Mapper and JMP software packages. While extracting NDVI, some negative values were obtained. This indicated that the spectral profile extraction encountered areas with no vegetation. From the obtained relationship between NDVI and Et, it was observed that evapotranspiration increases sharply with the increase of NDVI. The predicted Et results obtained from the multiple regression function of field Et, NDVI, solar radiation and/ or temperature were comparable with the actual Et, the numerical results were reported. The results showed that a remotely sensed vegetation index could be used to determine Et. However, there is still considerable work to be done before simple and full automated extraction of Et from remotely sensed data can be achieved for large scale applications.