Mapping and monitoring net primary productivity with AVHRR NDVI time-series: statistical equivalence of cumulative vegetation indices

In the last two decades, numerous investigators have proposed cumulative ve getation indices (i.e., functions which encode the cumulative effect of NDV I maximum value composite time-series into a single variable) for net prima ry productivity (NPP) mapping and monitoring on a regional to continental b asis. In this paper, we investigate the relationships among three of the mo st commonly used cumulative vegetation indices, expanding on the definition of equivalence of remotely sensed vegetation indices for decision making. We consider two cumulative vegetation indices as equivalent, if the value o f one index is statistically predictable from the value of the other index. Using an annual time-series of broad-scale AVHRR NDVI monthly maximum valu e composites of the island of Corsica (France), we show that the pairwise l inear association among the analysed cumulative vegetation indices shows co efficients of determination (R-2) higher than 0.99. That is, knowing the va lue of one index is statistically equivalent to knowing the value of the ot her indices for application purposes. (C) 1999 Elsevier Science B.V. All ri ghts reserved.