Evaluating the long-term water balance of arid zone stream bed vegetation using evapotranspiration modelling and hillslope runoff measurements

The difference between long-term actual evapotranspiration (AET) and precipitation (P) provides a useful indication of the extent to which a site retains or loses water resources, and therefore of the likely occurrence of specific land degradation processes. Sink areas (AET ˃˃ P) receive lateral water inputs from other parts of the catchment. In arid and semi-arid environments these areas are frequently found in, or next to, the stream beds of ephemeral rivers and are often characterised by intensive land use or high conservation values. For both types of land use it is important to know if, and how much, AET exceeds P, and where the lateral water inputs come from. Thick sedimentary fills in the stream bed, variable climate conditions and ephemeral flow conditions pose specific difficulties to the evaluation of the water balance of these sites. The objective of this study was to develop an approach to explore the relative importance of lateral water inputs to shrub stands growing in thick sedimentary fills of semi-arid ephemeral rivers. The approach is based on (i) estimating long-term AET - P balances in the channel sediments and (ii) assessing whether these inflows originate mainly from surrounding hillslopes or from the upstream part of the catchment. A physically based evapotranspiration model for sparse vegetation was used to estimate the long-term AET rates. The relative importance of hillslope runoff and channel flow was evaluated in a semi-quantitative fashion from a combination of surface area estimates and mostly published values of soil hydrological parameters. The approach was developed and tested in a selected stand of Retama sphaerocarpa shrubs in a stream bed at the Rambla Honda field site (Tabernas, Almería, SE Spain). Predictions from the evapotranspiration model, which were found to be accurate during previous studies at Rambla Honda, show that actual evapotranspiration (AET) largely exceeds precipitation (P) at annual scales. The estimated deficit may be compensated by: (a) infiltration of local rainfall during extreme events; (b) runon from the surrounding hillslopes; or (c) infiltration of channel flow during flash floods originating from the upper part of the catchment. Results show that possibilities (a) and (b) cannot explain the water deficit. Deep storage of water during floods in the main channel, however, can be as much as 60–150 mm per event, and may have been 160–400 mm per year during the study period (1994– 1997). This amount is large enough to replenish the annual deficit of ca. 100 mm per year found in the R. sphaerocarpa stand. These results imply that under current climate conditions land use changes in the upper sections of the Rambla Honda basin are more important for the persistence of the stream bed vegetation of our site than the land cover and runoff from surrounding hillslopes