A satellite-based approach for evaluation of the spatial distribution of evapotranspiration from agricultural lands.

Abstract

The ultimate goal of this dissertation was to produce maps of surface evaporation for agricultural areas based on Landsat Thematic Mapper (TM) spectral data. This achievement was dependent upon successful attainment of four intermediate goals: (1) Enhancement of TM thermal spatial resolution; (2) Atmospheric correction of TM visible and near-IR spectral data; (3) Atmospheric correction of TM thermal data; and (4) Remote estimation of crop aerodynamic properties. A statistical technique was developed to combine low-resolution (120 m) TM thermal data (TM6) with higher resolution (30 m) TM reflective data based on the relation between TM6 and the TM red and near-IR wavebands. This method was successful in improving the visible appearance of the TM6 image and retaining the original thermal spectral information over diverse agricultural landscapes. Several atmospheric correction procedures were examined to determine which techniques could provide the ease and accuracy necessary for the remote ET model. The Lowtran7 radiative transfer code was chosen for correction of TM visible and near-IR data (TM1-TM4) because it provided adequate accuracy (±0.02 reflectance, 1 σ RMS) and easy application. For TM6, results using the Lowtran7 code with a variety of atmospheric models were unsatisfactory. However, a simple linear regression of measured surface temperatures (T(s)) and TM6 digital numbers provided estimates of T(s) to within ±1.2°C of measured values. Though the procedure was accurate, it required concurrent ground-based measurements of T(s) and would obviously be inconvenient if it were used on an operational basis. Reasonable estimates of aerodynamic parameters were made for an alfalfa canopy from remote measurements of red and near-IR reflectance. The uncertainty in sensible heat flux density associated with the error in remote estimates of aerodynamic resistance was ±25%. Since these results were probably crop-specific and possibly site-specific, more data sets of this nature will need to be collected for other crops to determine a universal relation between remotely sensed data and aerodynamic properties. Data from the satellite-based TM sensor and ground-based meteorological instruments were combined to produce maps of latent heat flux density (LE: a function of evaporation rate (E) and heat of vaporization (L)) for Maricopa Agricultural Center, Arizona. The satellite-based estimates of LE differed from coincident ground-based measurements, using a Bowen-ratio apparatus, by 4% in cotton and -6% in alfalfa. These results were within the suggested accuracy goal of ±11%.