Can a vegetation index derived from remote sensing be indicative of areal transpiration?

Monthly, maximum-value-composited normalized difference vegetation indices (NDVI), calculated from NOAAAVHRR images, were correlated with annual transpiration (TR) estimates (i.e. annual precipitation minus runoff) for seven watersheds in Pennsylvania. A moderate relationship between the standardized variables was detected. The NDVI–TR relationship dramatically improved with improved watershed TR estimates. At the Little River watershed in Georgia, where the water balances of two sub-catchments could be reliably estimated over water cycles of variable length (about 2 months to 12 years), the correlation coefficient between NDVI and TR was found to be 0.94 (a sample size of 13). The present approach avoids the common practice of applying arbitrary hydrological models to validate the NDVI–TR relationship and attempts to minimize the effects of possible spurious correlations between the two variables that may stem from well-defined annual cycles in both the TR process and the foliage development of vegetation. It is concluded here that NDVI seems to reflect temporal changes in areal TR in a humid environment under well-vegetated conditions. © 2000 Elsevier Science B.V. All rights reserved.