Evaluation of spatiotemporal patterns and uncertainty of multi-source remote sensing GPP products in the Southwest karst region of China

Accurate estimation of gross primary productivity (GPP) in the karst regions of southwestern China is important for understanding regional carbon-cycle dynamics. We evaluated five GPP products (BESS, GOSIF, GLASS, MOD17, and VPM) during 2001–2018 by comparing site-scale performance, spatial patterns, temporal variations, and relative uncertainties in karst and non-karst regions. GOSIF generally showed the strongest correlations with flux-tower-derived GPP, whereas VPM exhibited relatively low estimation errors at several sites. All products estimated lower annual mean GPP in karst regions than in non-karst regions. Monthly mean GPP was generally highest in summer and lowest in winter, with smaller karst–non-karst differences in summer. BESS, GOSIF, MOD17, and VPM showed significant increasing trends in annual mean GPP, whereas GLASS showed only a weak and non-significant increase. MOD17 had the lowest TCH-derived mean relative uncertainty, followed by VPM. Overall, product performance varied across the evaluation criteria, indicating that product selection should depend on the specific application and evaluation objective rather than on a single overall ranking. These findings provide useful support for carbon-cycle assessment and ecosystem monitoring in karst regions.