Quantification of net primary production of Chinese forest ecosystems with spatial statistical approaches

Abstract

Net primary production (NPP) of terrestrial ecosystems provides food, fiber, construction materials, and energy to humans. Its demand is likely to increase substantially in this century due to rising population and biofuel uses. Assessing national forest NPP is of importance to best use forest resources in China. To date, most estimates of NPP are based on process-based ecosystem modeling, forestry inventory, and satellite observations. There are little efforts in using spatial statistical approaches while large datasets of in-situ observed NPP are available for Chinese forest ecosystems. Here we use the surveyed forest NPP and ecological data at 1,266 sites, the data of satellite forest coverage, and the information of climate and topography to estimate Chinese forest NPP and their associated uncertainties with two geospatial statistical approaches. We estimate that the Chinese forest and woodland ecosystems have total NPP of 1,325 ± 102 and 1,258 ± 186 Tg C year−1 in 1.57 million km2 forests with a regression method and a kriging method, respectively. These estimates are higher than the satellite-based estimate of 1,034 Tg C year−1 and almost double the estimate of 778 Tg C year−1 using a process-based terrestrial ecosystem model. Cross-validation suggests that the estimates with the kriging method are more accurate. Our developed geospatial statistical models could be alternative tools to provide national-level NPP estimates to better use Chinese forest resources.