Relation between the normalized difference vegetation index and ecological variables

Abstract

The objectives of this study were to analyze the relationship between the AVHRR data [transformed into a normalized difference vegetation index (NDVI)] and ecological variables and, secondly, to evaluate the possibility of estimating actual evapotranspiration using AVHRR and other readily available data. The study was undertaken across a range of conditions in Canada. Meteorological and composited NDVI data were compiled for 15-day periods over the 1986 growing season and were supplemented by digital ecological and soil data. The effect of soil and meteorological conditions on vegetation growth was modeled using the Versatile Soil Moisture Budget model, which accounts for additions and losses of water from various layers in the root zone. The various data sets were integrated using a geographic information system. We found that individual vegetation/soil combinations exhibited different NDVI trajectories, most closely related to the trends in potential evapotranspiration. There was a highly significant association between coincident NDVI and actual evapotranspiration values ( r = 0.77) or between NDVI and potential evapotranspiration for the previous 15-day period ( r = 0.86). High correlation was found between cumulative NDVI and cumulative actual evapotranspiration over the growing season ( r = 0.96). Actual evapotranspiration for individual 15-day periods could be estimated from NDVI and potential evapotranspiration values with a residual error of about 10–15%. The results show that the NDVI trajectories provide important information on the seasonal development of vegetation at northern latitudes. Results in estimating actual evapotranspiration were also encouraging but further work is needed to determine the accuracy limits for this geographic area. The representativeness of the 1986 results for the region is subject of continuing study with 6 years of NDVI data.