New research comparing five satellite-derived gross primary productivity (GPP) products across southwestern China's karst region finds no single dataset is best for every use. GOSIF matched flux-tower observations most closely, VPM showed low site-level errors, and MOD17 had the lowest relative uncertainty. All five products agreed that karst areas have lower annual productivity than non-karst areas, but long-term trends differed, with GLASS showing no significant increase. The study, published in Scientific Reports, evaluated BESS, GOSIF, GLASS, MOD17 and VPM from 2001 to 2018. It tested products against tower-based carbon flux data and used the three-cornered hat method to estimate uncertainty. The authors say product choice should depend on the research question, not a single metric. For carbon cycle studies and monitoring in complex landscapes, the practical result is a more transparent view of where satellite maps agree and where they don't.
