Multitemporal analysis of PAL images for the study of land cover dynamics in South America

Abstract

Pathfinder AVHHR Land (PAL) database has been used for the retrieval of Land Surface Temperature (LST) over South America, which, along with NDVI parameter, will allow the studying of the evolution of the vegetation between July 1981 and September 2001. To this end, a classification has been built, based on PAL NDVI and Reanalysis air temperature at 2 in height data. This classification takes into account both vegetation and thermal patterns, and has been validated by a comparison with CAZALAC's map of and zones (Centro del Agua para Zonas Aridas y semiaridas de Latino-America y el Caribe), as well as with Global Land Cover Characteristics' classification built by the USGS (United States Geological Survey). The principal advantage of this new classification is that it is a dynamic classification, that considers the actual state of the cover, since no assumption on land occupation is made for its construction. LST and NDVI yearly and long-term evolutions are analyzed with the help of this classification. Yearly evolutions are compared with Reanalysis air temperature at 2 in height and precipitation, and show good concordance. LST long-term evolution shows to be strongly affected by satellite changes and orbital drift. These latter require an adequate correction to allow deeper study. On the other hand, NDVI does not show this trend, but aerosol absorption from Mount Pinatubo's eruption in June 1991 corrupts temporarily the data. These results also validate the above-mentioned classification.