THE USE OF RADARSAT-1 IMAGES FOR CLASSIFICATION OF BOREAL PEATLANDS AND NON-PEATLANDS ENVIRONMENTS IN THE LA GRANDE RIVER AREA, JAMES BAY, QUEBEC

Peatlands ecosystems have been storing carbon through photosynthesis processes for thousands of years. As peatlands store more carbon than they release, they are qualified as « sinks of carbon ». In the context of global climatic change, peatlands could reverse their role and become a source of greenhouse gas influenced by decreased precipitation and increased evaporation under northern latitudes. To simulate CO2 and NH4 fluxes by models such as PCARS model (Peatlands Carbon Simulator; Frolking et al, 2002, 2001), realistic estimation of surface peatlands is required. The main goal of this project is to develop a methodology to characterize and classify peatlands (forested peatlands, ribbed peatlands, etc.) and non-peatlands environments (mineral forested, outcrop, abandoned land, etc.) using Radarsat-1 satellite imagery. The methodology is applied to boreal peatlands in the La Grande river area, James Bay, Québec, Canada (77°44'; 53°38'). The surface land classification from Radarsat-1 images is also compared with results from LANDSAT imagery. The defined methodology uses textural analysis for land cover classification. The accomplishment of this project has required a field campaign to validate, caracterize and classify the present-day land occupation. A further objective will be the use of this latter methodology to classify and interpret the entire peatlands (forested and non forested) on the whole drainage basin of the La Grande river. This project is part of an integrated research project supported by NSERC and Hydro-Québec and is realized with the financial support of « Fonds québécois de la recherche sur la nature et les technologies (NATEQ : # 73090).