A multiyear analysis of the relationship between surface environmental variables and NDVI over the Canadian landmass

Publisher Website
Google Scholar
Abstract
The relationships between the normalized difference vegetation index (NDVI) and surface environmental variables are studied across the Canadian landmass using data over a six year period (1982 to 1987). The NDVI images are 15 day composites over the growing season obtained from NOAA AVHRR Global Area Coverage data. Daily climatological surface data in combination with vegetation and soil characterization digital data sets are input into a multilayer soil water budget model to calculate daily amounts of potential evapotranspiration (PE), actual evapotranspiration (AE), and soil moisture. A preliminary study for the year 1986 addressing the above relationships was undertaken by Cihlar et al. (1991). In this study, the stability of the relationships and the effect of a wide range of climatic conditions are investigated. It is found that consistently over the 6 year period, seasonal NDVI trajectories followed PE trends (and AE where water supply was not limiting), lagging by approximately one compositing period (15 days). Classes with adequate moisture supply showed a higher NDVI correlation with AE than those suffering from water deficit. For water deficient classes, the correlation between NDVI and PE was considerably higher than for NDVI versus AE, suggesting that in these cases, NDVI reflects the thermal regime more than moisture availability. Multiple linear regressions of AE against NDVI and PE of individual compositing periods accounted for 70% to 85% of the total AE variance using class‐averaged values. Using seasonal totals, AE was most closely related to NDVI among the variables studied. These results indicate that for the geographic region under consideration, moisture does not severely limit leaf/plant growth in most cases and where it does, the impact is moderated by the energy environment. They also illustrate the very high variability in all four parameters for the wide geographic and temporal ranges involved. Therefore, environmental research aiming at developing techniques or conclusions applicable over large areas must take this variability explicitly into account. How to do this in regions with scarce accurate environmental measurements will continue to be a significant challenge to earth scientists in the forthcoming years.