Applying high-resolution satellite images to estimate tree diversity of mixed broadleaf-Korean pine forest

The application of remote sensing technology to estimate forest biodiversity is a research hotspot and challenge in forest ecology. To explore the potential of high-resolution satellite data in estimating forest vegetation diversity, we extracted the spatial pattern of Korean pine (Pinus koraiensis) canopies, and estimated the arbor species diversity, supported by field large plot data in Jiaohe, Jilin Province. First, we extracted the Korean pine canopy distribution based on a GeoEye-1 winter image, which agreed well with individual tree position measured in situ. Then, we analyzed the spatial population distribution of Korean pine using landscape ecology and geostatistics index, which was found to have a maximum autocorrelation scale of 30--40 m. Finally, we used a RapidEye satellite image with a special red edge band in autumn to establish multiple linear regression model with plot diversity data. The model linked remote sensing image spectral values (5 m) and texture parameters (30 m window size) with species diversity. Results showed that red edge information did not significantly improve the ability of multi-spectral data in estimating vegetation biodiversity, but regional mapping results matched well with the stand age class.