A soil-plant-atmosphere model was used to estimate gross primary productivity (GPP) and evapotranspiration (ET) of a tropical savanna in Australia. This paper describes model modifications required to simulate the substantial C4 grass understory together with C3 trees. The model was further improved to include a seasonal distribution of leaf area and foliar nitrogen through 10 canopy layers. Model outputs were compared with a 5-year eddy covariance dataset. Adding the C4 photosynthesis component improved the model efficiency and root-mean-squared error (RMSE) for total ecosystem GPP by better emulating annual peaks and troughs in GPP across wet and dry seasons. The C4 photosynthesis component had minimal impact on modelled values of ET. Outputs of GPP from the modified model agreed well with measured values, explaining between 79% and 90% of the variance and having a low RMSE (0.003-0.281gCm -2day -1). Approximately, 40% of total annual GPP was contributed by C4 grasses. Total (trees and grasses) wet season GPP was approximately 75-80% of total annual GPP. Light-use efficiency (LUE) was largest for the wet season and smallest in the dry season and C4 LUE was larger than that of the trees. A sensitivity analysis of GPP revealed that daily GPP was most sensitive to changes in leaf area index (LAI) and foliar nitrogen (N f) and relatively insensitive to changes in maximum carboxylation rate (V cmax), maximum electron transport rate (J max) and minimum leaf water potential (ψ min). The modified model was also able to represent daily and seasonal patterns in ET, (explaining 68-81% of variance) with a low RMSE (0.038-0.19mmday -1). Current values of N f, LAI and other parameters appear to be colimiting for maximizing GPP. By manipulating LAI and soil moisture content inputs, we show that modelled GPP is limited by light interception rather than water availability at this site.
- SPA modelling