MID-TERM STATUS OF THE FOREST DRAGON PROJECT

摘要

The objective of the Forest DRAGON Project is the development of algorithms for classification of synthetic aperture radar, SAR, data and interferometric SAR (InSAR) data, and the generation of forest maps at regional level for the main forested regions of China. For this purpose ERS-1/2 tandem coherence and ENVISAT ASAR data in Alternating Polarisation (AP) and Wide Swath (WS) mode are considered. ERS-1/2 tandem coherence has been chosen as primary candidate for the estimation of forest stem volume because of its strong sensitivity to forest biophysical parameters. Interferometric processing is carried out by means of a well-established chain at the end of which geocoded coherence, backscatter and local topography products are obtained. Forest base-maps are obtained for the mid-1990s making use of the SIBERIA Project algorithm. First investigations in Northeast China show the need of an improvement of the algorithm to take into account the availability of multi-seasonal data and misclassifications in areas with strong topography. To update the maps multi-temporal ENVISAT ASAR AP images in HH/HV mode are considered. A data processing chain, which includes data import, calibration, geocoding and mosaicing, has been set to process the ASAR data automatically. Preliminary results in Northeast China show that foreston-forest types can be discriminated well using one date ASAR AP HH/HV images. ScanSAR or Wide Swath mode represent an important source of data for updating global and continental maps, yet to be explored. In this study, ENVISAT ASAR WS data over Northeast China are analysed for foreston-forest mapping at regional scale. The results are compared with existing land cover maps, in particular the Vegetation Continuous Field (VCF) from MODIS.