John, R; Chen, JQ; Noormets, A; Xiao, XM; Xu, JY; Lu, N; Chen, SP (2013). Modelling gross primary production in semi-arid Inner Mongolia using MODIS imagery and eddy covariance data. INTERNATIONAL JOURNAL OF REMOTE SENSING, 34(8), 2829-2857.
We evaluate the modelling of carbon fluxes from eddy covariance (EC) tower observations in different water-limited land-cover/land-use (LCLU) and biome types in semi-arid Inner Mongolia, China. The vegetation photosynthesis model (VPM) and modified VPM (MVPM), driven by the enhanced vegetation index (EVI) and land-surface water index (LSWI), which were derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) surface-reflectance product (MOD09A1), were used to model and validate the temporal changes in gross primary production (GPP) from the EC towers during the 2006 and 2007 growing seasons. The annual GPP predicted by the VPM model (GPP(VPM)) was predicted reasonably well in 2006 and 2007 at the cropland (coefficient of determination, R-2=0.67 and 0.71, for 2006 and 2007, respectively) and typical steppe (R-2=0.80 and 0.73) sites. The predictive power of the VPM model varied in the desert steppe, which includes an irrigated poplar stand (R-2=0.74 and 0.68) and shrubland (R-2=0.31 and 0.49) sites. The comparison between GPP obtained from the eddy covariance tower (GPP(tower)) and GPP obtained from MVPM (GPP(MVPM)) (predicted GPP) showed good agreement for the typical steppe site of Xilinhaote (R-2=0.84 and 0.70 in 2006 and 2007, respectively) and for the Duolun steppe site (R-2=0.63) and cropland site (R-2=0.63) in 2007. The predictive power of the MVPM model decreased slightly in the desert steppe at the irrigated poplar stand (R-2=0.56 and 0.47 in 2006 and 2007 respectively) and the shrubland (R-2=0.20 and 0.41). The results of this study demonstrate the feasibility of modelling GPP from EC towers in semi-arid regions.