Spatiotemporal dynamics of forest net primary production in China over the past two decades

Forest ecosystems play an important role in global carbon cycle regulation. Clarifying the dynamics and mechanism of carbon sink is of both scientific and political importance. In this paper, we have investigated the spatiotemporal change of forest net primary production (NPP) in China for recent two decades based on the geographically weighted regression (GWR) with a cumulative remote sensing index, the maximum normalized difference vegetation index (NDVI_max). GWR is a recently developed regression method with special emphasis on spatial non-stationarity. Outputs of forest NPP at three different stages was generated by the GWR model with NDVI_max for the 1980s, early and late 1990s which were consequently analyzed. Our results indicated a wave-like pattern of change in forest NPP in the three stages with a trough-like depression for the early 1990s. The average forest NPP increased by about 0.72% from the 1980s to the late 1990s. A continuously increasing trend at a pace of 0.07% and 0.22% yr^− 1 was observed in the tropical and subtropical zones from the 1980s to late 1990s respectively, while a continuously decreasing trend (− 0.05% yr^− 1) was noted for the temperate zone. From forest type perspective, only the deciduous broadleaf forests exhibited a continuously decreasing trend of 0.18% yr^− 1. The complex spatiotemporal patterns revealed by this study suggest the need for further research in this direction in order to build in-depth insights into the revealed complexities.