Evaporative water use of different land uses in the upper-Thukela river basin assessed from satellite imagery

Evaporative water use of various land use classes in the upper-Thukela river basin was estimated using the public domain version of the Surface Energy Balance Algorithm for Land (SEBAL) and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images. Twenty eight images were analyzed covering the period between June 2005 and September 2006. The South Africa land use map developed in the year 2000 was used to compute the evaporative water use of the various land uses in 13 Quaternary Catchments (QCs) in the upper-Thukela river basin.There was a good correlation between the SEBAL estimates of total evaporation and ground measurements from a Large Aperture Scintillometer installed at a site in one of the QCs in the study area. It was observed that the land uses that generated relatively large volumes of evaporative water were forestry (i.e. Eucalyptus, Pine, mixed species & indigenous), “water bodies” (i.e. water supply reservoirs, farm reservoirs) and wetlands. Total evaporation rates for all land use classes were high during the summer season (wet), with Eucalyptus ranging between 3 mm d−1 during the winter season (dry) and about 5 mm d−1 during the summer season. Bare rocks and eroded soil surface land use class had the least evaporative water use i.e., less than 1 mm d−1. The relatively low evaporation rates over “water bodies” during the dry winter season could be attributed to the averaging of mixed pixels (wet and dry) due to change in areal coverage and which could not be captured in the 1 km × 1 km low resolution MODIS images. Thus, the potential for applying remote sensing techniques, using low resolution satellite images, to quantify water use by various land uses in the Thukela river basin was explored with promising results.

Research highlights▶ Each land use has different evaporative water use. ▶ Remote sensing is a good platform for addressing scale issues in hydrology. ▶ Low spatial resolution of satellite images affects the accuracy of remotely sensed flux estimates.