Oral Session - Remote Sensing Chair(s): Slonecker, Terrence¹, Jarnagin, Taylor¹, ¹ US Environmental Protection Agency, Reston, VA
Thursday, April 1, 2004 8:00 AM - 11:40 AM Apollo Room 3-4

Lifting the green veil: enhanced change detection using the Wide Dynamic Range Vegetation Index. Henebry, Geoffrey^*,1, Viña, Andrés¹, Gitelson, Anatoly¹, ¹ Center for Advanced Land Management Information Technologies (CALMIT), Lincoln, NE, USA

ABSTRACT- It has long been recognized that the Normalized Difference Vegetation Index (NDVI) suffers a rapid decrease of sensitivity even at moderate Leaf Area Index (LAI) values of 2 to 4 that are commonly encountered in croplands, woodlands, and productive grasslands. This decrease in NDVI sensitivity casts a green veil over the land surface that obscures vegetation dynamics across vast areas of the planetary surface during much of the growing season. This veil impacts the monitoring of vegetation dynamics, the development of land surface climatologies, and the detection of significant changes. The prevalence of the veil can be estimated from NDVI time series. The Wide Dynamic Range Vegetation Index (WDRVI) is a straightforward generalization of the NDVI that was recently developed to increase its sensitivity under higher green biomass conditions. Applied to a standard, widely available AVHRR dataset for the conterminous US, the WDRVI exhibited significant increases in sensitivity between for Omernik Level III ecoregions dominated by woodlands, croplands, and grasslands. Ecoregions with low aboveground net primary production showed no increase in sensitivity of the WDRVI over the NDVI. We present further results on how of this increase in sensitivity affects the observable spatio-temporal structures in the image time series and how the WDRVI performs under different uncertainty scenarios.

KEY WORDS: WDRVI, AVHRR NDVI, land surface phenology, change detection, remote sensing