Forest Disturbances and Biomass Mapping in Northeastern Asia

Guoqing Sun, University of Maryland, guoqing.sun@nasa.gov (Presenting)
Anming Fu, Chinese Academy of Sciences, fam1234@126.com
Jeff Masek, NASA GSFC, jeffrey.g.masek@nasa.gov

Forests in the northeastern Asia undergo frequent forest fire and extensive logging in recent years. Forest cover and its changes from 2000 to 2006 in northeastern Asia were studied using 500m MODIS product and other satellite data. The study region (38-60N, 115-140E) was divided into four sub-regions based on different climate-environment regimes. In each sub-region, TM/ETM+ images were used as ancillary data for training and testing. A no-parametric 2D feature space split method was used to segment forest region according to its MODIS attributes in both red and MIR channels. A decision tree was used to split forest areas into different forest biomes by multi-temporal and multi-spectral MODIS data.

The detection of forest disturbance caused by forest fire and logging using MODIS time-series data requires representative signatures between two images to be compared. Data quality flags and various data smooth or correction methods were applied to reduce cloud and snow effects on signature, and the median value observed in summer was used to represent growing status of each year. The time-series NDVI was used to find when maximum dropping of NDVI happened. Then the change vectors were calculated, and the disturbance degrees of changed pixels were nominated by statistical threshold.

GLAS samples of canopy height and biomass within each category were used to give statistical estimations, and the current carbon storage and changes were mapped. The quantitative results will be given in this poster. Forest inventory and other published maps and statistical reports were used in results evaluation.