Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Zambie → Remote sensing estimation of spatial and temporal variability of actual evapotranspiration using SEBs Algorithm in the semi-arid Barotse Sub-Basin, South-Western Zambia

University of Zambia (2013)

Remote sensing estimation of spatial and temporal variability of actual evapotranspiration using SEBs Algorithm in the semi-arid Barotse Sub-Basin, South-Western Zambia

Phiri, Kakusa Wilson

Titre : Remote sensing estimation of spatial and temporal variability of actual evapotranspiration using SEBs Algorithm in the semi-arid Barotse Sub-Basin, South-Western Zambia

Auteur : Phiri, Kakusa Wilson

Université de soutenance : University of Zambia

Grade : Master of Science (MS) 2013

Résumé
Evapotranspiration (ET) is a dominant hydrologic flux in the water budget of semi-arid areas.Thus, accurate estimation of its dynamics in such environments is critical for improvingN water resources management. In this study the physically-based Surface Energy Balance System (SEBS) model was applied to estimate the spatial and temporal variability of actual ET (AET) in the semi-arid Barotse Sub-basin, South-Western Zambia. The model was run using atmospherically rectified Moderate-resolution Imaging Spectroadiometer (MODIS) satellite imagery on clear-sky warm-wet, cool-dry and hot-dry days. Furthermore, based on sunshine hours and daily AET, monthly fluxes were generated. The modelled evaporative fluxes were evaluated against Penman-Montieth potential ET (PET) and independently modelled AET from the Global Circulation Model (GCM) of the European Centre for Medium-Range Weather Forecast (ECMWF).Results showed that actual evaporative fluxes were 64.3% and 29.4% of PET on cool-dry and hot-dry days respectively. However, these fluxes were 104.2% of PET on warm-wet days. The systematic lack of physical agreement on these days implied that SEBS estimates were not necessarily implausible but that the assumptions on which PET is based differed from the surface conditions. This highlighted the uncertainties of evaluating AET against PET. The comparison with ECMWF estimates showed better agreement on many days at Sesheke weather station than at Kamanga. At a monthly time-step, however, this comparison showed lack of good agreement ascribed to input data and surface parameterisation. Sensitive analysis showed that model outputs varied by up to 3 mm day-1 when estimated air temperature in the term D(T0-Ta) was varied by 8 K whereas the use of NDVI versus landusebased surface roughness revealed a reduction of ET of up to 1.5 mm day-1 on forests when the latter was used. Flux analysis showed that water bodies and regularly flooded vegetation had the highest rates of 6.9 and 5.9 mm day-1 on warm-wet days respectively. The lowest rates occurred over mosaic vegetation/croplands and closed to open grassland with a high variation of up to 64.1 and 71.1% respectively between warm-wet and hot-dry days.On the overall, this study showed that the SEBS model can be successfully used to estimate evaporative fluxes in heterogeneous areas and improve water resources management. In order to accurately apply this model in such areas, however, there is need to use spatial input data and robust ways of estimating surface roughness

Mots clés : Algorithms ; Remote Sensing ; Evapotranspiration

Présentation

Version intégrale (30,33 Mb)

Page publiée le 7 juillet 2015, mise à jour le 15 février 2018