Large area forest stem volume mapping using synergy of spaceborne interferometric radar and optical remote sensing : a case study of northeast chin
More than a decade of investigations on the use of the interferometric ERS-1/2 tandem coherence for forest applications have increased the understanding of the behaviour of C-band repeat-pass coherence over forested terrain. It has been shown that under optimal imaging conditions, ERS-1/2 tandem coherence can be used for stem volume retrieval with accuracies in the range of ground surveys. Large-area applications of ERS-1/2 tandem coherence are rare though. One of the main limitations concerning large-area exploitation of the existing ERS-1/2 tandem archives for forest stem volume retrieval is related to the considerable dependence of repeat-pass coherence upon the meteorological (rain, temperature, wind speed) and environmental (soil moisture variations, snow metamorphism) acquisition conditions. Conventional retrieval algorithms require accurate forest inventory data for a dense grid of forest sites to tune models that relate coherence to stem volume to the local conditions. Accurate forest inventory data is, however, a rare commodity that is often not freely available. In this thesis, a fully automated algorithm was developed, based on a synergetic use of the MODIS Vegetation Continuous Field product (Hansen et al., 2002), that allowed the training of the Interferometric Water Cloud Model IWCM (Askne et al., 1997) without further need for forest inventory data. With the new algorithm it was possible to train the IWCM on a frame-by-frame basis and thus to account for the spatial and temporal variability of the meteorological and environmental acquisition conditions. The new algorithm was applied to a multi-seasonal ERS-1/2 tandem dataset covering Northeast China that was acquired between 1995 and 1998 with baselines up to 400 m.