Estimating forest biomass with GLAS samples and MODIS imagery in Northeastern China

Abstract

The forest ecosystem in Northeastern China (NEC) is approximately 25% proportion of total forested area of China, which has been undergoing dramatic changes due to massive loggings and forest fires in the last several decades and successively intensive manual afforestation and closing protective recovery since 1990s. It is a hot region for scientific research in carbon balance. In this paper, national land cover GIS data, moderate resolution imaging spectroradiometer (MODIS) imagery, and vertical waveform of Geoscience Laser Altimeter System (GLAS) onboard the Ice, Cloud, and Land Elevation Satellite (ICESAT) were combined together to map forest aboveground biomass (AGB) in the NEC. Firstly, GLAS waveform has the advantage of three dimensional observations and can play the role as sampling footprints for forest biomes. The estimation algorithm was developed between field survey samples and height profile indices of GLAS waveform to predict forest AGB by neural net regression model. The correlation coefficient R2 between GLAS forest AGB and field-investigated ones was 0.73. Secondly, MODIS data affords spatially continuous images and can be used to stratify forested regions as statistical districts. one hundred of spectral clusters were derived from MODIS phenological curve of enhanced vegetation index (EVI) and near infrared (NIR) channel by K-Means method and stratified for the statistics of GLAS forest AGB samples. The result illustrates spatial pattern forest AGB and explores its total amount in the NEC.