Estimation of Evapotranspiration Using Fused Remote Sensing Image Data and Energy Balance Model for Improving Water Management in Arid Area

Remote sensing has proved to be very useful in the investigation of vegetation and hydrological monitoring, especially when studying vast areas. In this paper, the complement between two optical remote sensing data (Landsat TM and NOAA- AVHRR) and a Digital Elevation Model (DEM) is used to estimate hydrological parameters based on derived surface reflectance. These parameters which are used in the Modified Soil Energy Balance Algorithm for Land (M-SEBAL) model have been used to estimate net radiation, soil heat flux, sensible heat flux and evapotranspiration (ET) for Sanapsilaa Basin in Yemen. The area is known for arid and semi-arid weather conditions with undulating topography. Image data from AVHRR on-board NOAA satellites with a large areal coverage, good temporal and spectral resolution are found to be very useful in generating some parameters required for the above process. However, the data have poor spatial resolution. On the other hand, image data from the Thematic Mapper on-board the Landsat satellite, with a high spatial and spectral resolution should be able to provide values for the parameters involved, but the area coverage is significantly reduced. This study has been carried out, using a data fusion technique in order to exploit the respective advantages of these two disparate sources of image data. A general framework is then proposed to generate ET maps for arid and semi-arid regions. This is achieved by means of multi-temporal, multi-resolution remote sensing data. Taking into account topographic effects, an attempt has also been made to incorporate DEM information for estimating the net radiation of the areas involved. An application for computing a daily ET map over Sanapsilaa Basin, Yemen is presented. As a result, a daily ET map generated from meteorological observations was compared with estimated ET data simulated from remote sensing data. In conclusion, data from both remote sensing sources give reasonable values with the result from the TM being better than those obtained from the AVHRR. This is attributed to the differences in spatial resolution, in which TM data is higher than AVHRR. The fusion of the two shows improves spatial detail whilst maintaining the spectral signature close to the original.