ABSTRACT
Carbon density is one of the important indicators for carbon sequestration capacity in forest ecosystems. The analysis of spatial variation in forest carbon density can provide scientific basis for management of forest carbon sink. We investigated the carbon density of Pinus massoniana forest in Jiangxi Province. Based on plot investigation and meaurement of sample carbon content, the spatial autocorrelation and distribution of carbon density were analyzed and explored respectively by using global Moran I, local Moran I and geostatistics. Results from the global Moran I showed that ecosystem carbon density had significant positive spatial autocorrelation and its autocorrelation decreased with the increase of distance. Ecosystem carbon density had no spatial autocorrelation when the distance exceeded 400 km. The spatial distribution of ecosystem carbon density changed from aggregated distribution to random distribution with the increase of distance. Results from the local Moran I showed that the local spatial distribution of ecosystem carbon density was not uniform, and the difference of local spatial distribution gradually increased with the increase of distance. The fitting results of semi-variogram models showed that the spherical model was the best fitting model for the estimation of ecosystem carbon density. The ratio of nugget to sill was 0.36, indicating mode-rate spatial correlation of carbon density. The ecosystem carbon density based on kriging spatial interpolation mainly concentrated in the range of 85.14-153.52 t·hm-2. The spatial distribution regularity was generally low in middle region and high in peripheral region, which was consistent with the terrain characteristics of the study area.
