Variability analysis of the transitory climate regime as defined by the NDVI/Ts relationship derived from NOAA-AVHRR over Canada

This research work outlines an original method for climate observation by remote sensing based on the local combination of normalized difference vegetation index (NDVI) and land surface temperature (T_s) measurements acquired by the NOAA-AVHRR sensor. It explores the phenomenon of linearity observed between T_s and the NDVI, which varies from positive to negative according to the conditions of the land surface energy budget regime and the vegetation type. Over vegetation, the decreasing relationship of T_s in relation to the NDVI (negative regression) due to vegetation cover transpiration is well known. However, over soils with sparse vegetation, bare soil, lichens or tundra, the relationship is reversed (positive regression) due to the high surface albedo which influences T_s values. The method is first demonstrate using full spatial and temporal resolution HRPT images over the BOREAS area corrected for atmospheric effects and screened for cloud cover in comparison with temperature and precipitation data. The method is then applied to composite images from the PAL multi-annual database at a resolution of 8 km and for Canada overall. It permits the determination of the ecotone position separating the forest from the tundra and the monitoring of the inter-annual fluctuations related to climatic variations and global warming.