The application of the surface energy balance system model to estimate evapotranspiration in South Africa

Doctoral Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

In a water scarce country like South Africa with a number of large consumers of water, it is important to estimate evapotranspiration (ET) with a high degree of accuracy. This is especially important in the semi-arid regions where there is an increasing demand for water and a scarce supply thereof. ET varies regionally and seasonally, so knowledge about ET is fundamental to save and secure water for different uses, and to guarantee that water is distributed to water consumers in a sustainable manner. Models to estimate ET have been developed using a combination of meteorological and remote sensing data inputs. In this study, the pre-packaged Surface Energy Balance System (SEBS) model was used for the first time in the South African environment alongside MODerate Resolution Imaging Spectroradiometer (MODIS) satellite data and validated with eddy covariance data measured in a large apple orchard (11 ha), in the Piketberg area of the Western Cape. Due to the relative infancy of research in this field in South Africa, SEBS is an attractive model choice as it is available as open-source freeware. The model was found to underestimate the sensible heat flux through setting it at the wet limit. Daily ET measured by the eddy covariance system represented 55 to 96% of the SEBS estimate, an overestimation of daily ET. The consistent underestimation of the sensible heat flux was ascribed to sensitivities to the land surface air temperature gradient, the choice of fractional vegetation cover formula as well as the height of the vegetation canopy (3.2 m) relative to weather station reference height (2 m). The methodology was adapted based on the above findings and was applied to a second study area (quaternary catchment P10A, near Grahamstown, Eastern Cape) where two different approaches for deriving surface roughness are applied. It was again demonstrated that the sensible heat flux is sensitive to surface roughness in combination with land surface air temperature gradient and again, the overestimation of daily ET persisted (actual ET being greater than reference ET). It was concluded that in complex environments, at coarse resolution, it is not possible to adequately describe the remote sensing derived input parameters at the correct level of accuracy and at the spatial resolution required for the accurate estimation of the sensible heat flux.

Includes abstract.

Includes bibliographical references.