Parameterization of an ecosystem light-use-efficiency model for predicting savanna GPP using MODIS EVI

The enhanced vegetation index (EVI) tracked the seasonal variations of GPPEC well at both site- and cross-site levels (R2 = 0.84). The EVI relationship with GPPEC was further strengthened through coupling with ecosystem light-use-efficiency (eLUE), defined as the ratio of GPP to photosynthetically active radiation (PAR). Two savanna landscape eLUE models, driven by top-of-canopy incident PAR (PARTOC) or top-of-atmosphere incident PAR (PARTOA) were parameterized and investigated. GPP predicted using the eLUE models correlated well with GPPEC, with R2 of 0.85 (RMSE = 0.76 g C m− 2 d− 1) and 0.88 (RMSE = 0.70 g C m− 2 d− 1) for PARTOC and PARTOA, respectively, and were significantly improved compared to the MOD17 GPP product (R2 = 0.58, RMSE = 1.43 g C m− 2 d− 1). The eLUE model also minimized the seasonal hysteresis observed between green-up and brown-down in GPPEC and MODIS satellite product relationships, resulting in a consistent estimation of GPP across phenophases. The eLUE model effectively integrated the effects of variations in canopy photosynthetic capacity and environmental stress on photosynthesis, thus simplifying the up-scaling of carbon fluxes from tower to regional scale. The results from this study demonstrated that region-wide savanna GPP can be accurately estimated entirely with remote sensing observations without dependency on coarse-resolution ground meteorology or estimation of light-use-efficiency parameters.