干旱区生态环境知识资源中心

Shallow groundwater discharge in arid environments and subsequent loss by evapotranspiration (ET) often represents a significant component of the water balance. Evaporation is most reliably measured at the point scale. However, due to the vast distances over which this process can occur, field measurements of discharge are not viable. Remote sensing offers potential in mapping surface information which can be used to estimate ET. Various models have been developed to estimate ET from remote sensing imagery, though these have largely been applied to agricultural areas. Two studies, Kampf & Tyler (2006) and Bastiaansen et al. (1998a,b), have applied a surface energy balance modelling approach for estimating ET using multispectral satellite imagery. The two studies have conflicting conclusions, with the Bastiaanssen et al. study claiming such an approach could provide useful estimates. The Kampf & Tyler study suggested that the uncertainty within the model well exceeded the acceptable error margin of 50 W/m(2). Both studies tested their methods in an arid groundwater discharge zone though the landforms were slightly different. In this study, we employ eddy covariance instrumentation to obtain field measurements of the surface energy balance to apply the more rigorous validation method used by Kampf & Tyler (2006). We use these data to test the Surface Energy Balance for Land Algorithm (SEBAL) which was developed and validated by Bastiaanssen et al. (1998a, b), by applying it to the Landsat 5 data on groundwater discharge zones along the margin of the Great Artesian Basin (Figure 1). SEBAL is a widely used algorithm, but has mainly been tested in agricultural settings. Our results show that the model structure is seriously limited in its application in this environment. SEBAL is not able to adequately represent the Bowen ratio, which it assumes is related to the surface temperature. As a result, we see the spatial distribution of latent energy (LE) flux estimates being controlled by net radiation. Applying SEBAL to a summer image where the surface temperature appears to be strongly related to the Bowen Ratio, we see that SEBAL performs much better at representing the spatial distribution of LE fluxes. However, the uncertainty is very large in the net radiation calculations and when considering the low LE fluxes, this is significant. Further more, it remains to be investigated if the low surface temperature found in the discharge zones is caused by evaporative cooling or a high albedo.

[GRAPHICS]

.