Sensitivity of modelled gross primary productivity to topographic effects on surface radiation: A case study in the Cotter River Catchment, Australia

Gross primary productivity (GPP) is a critical response variable for many environmental problems, including terrestrial carbon accounting and the calculation of catchment water balances. Various approaches for modelling GPP have been developed and applied at continental and landscapes scales, but little attention has been given to the sensitivity of GPP to the spatial scale of its driving variables. A key driving variable is surface radiation (R-s) which is influenced by both meso-scale factors (latitude, time of year, cloudiness) and the topographic variables of slope, aspect and horizon shading. We compared the sensitivity of modelled GPP to two different sources of surface radiation (R-s): (1) the ANUCLIM method which only captures the meso-scaled factors; and (2) the SRAD method which incorporates the topographic effects GPP was calculated using the radiation use efficiency (RUE) model (Roderick et al., 2001) to discern general patterns of vegetation productivity at a sub-catchment (i.e. sub-water shed) scale. The radiation use efficiency approach uses the normalized difference vegetation index (NDVI) derived from satellite data (MODIS TERRA), along with estimates of solar radiation at the top of the atmosphere (R-o) and canopy (R-s). In this approach, R-o and R-s capture the influence of diffuse irradiance in canopy photosynthesis and vegetation productivity respectively. This research showed that R-s calculated using the SRAD program provides important discrimination of GPP regimes at a sub-catchment scale, as the result of minimum and maximum daily radiation varying between shaded and exposed surfaces. However, mean daily radiation at a whole-of-catchment scale did not differ between the two sources as the differences in the minimum and maximum daily values tend to cancel each other out. Applications of GPP models therefore need to consider whether topographic factors are important and select the appropriate source of R-s values. GPP models should also reflect understanding of radiation use efficiency. However, further research is required especially with respect to the influence of water stress on plant response. (C) 2010 Elsevier B.V. All rights reserved.