A practical method for simulating AVHRR-consistent NDVI data series using narrow MODIS channels in the 0.5-1.0 μm spectral range

Abstract

Over the past two decades, a key indicator of climate change has been the long time series of global maps of the normalized difference vegetation index (NDVI), derived from remotely sensed data acquired with a series of NOAA advanced very high resolution radiometer (AVHRR) instruments from space. These NDVI values are calculated from relatively broad AVHRR channels in the red and near-infrared regions. Continuation of this long term data set is extremely valuable for climate-related research, However, sometime in the coming decade, the AVHRR time series measurements will no longer be continued. Instead, the measurements will be made using newer generation satellite instruments having narrower channels and improved spatial resolution. For example, the moderate resolution imaging spectroradiometer (MODIS) onboard the Terra spacecraft has several narrow channels in the 0.4-1.0 spectral range. The NDVI values derived from the MODIS red channel and near-IR channel will be biased compared to those derived from the broader AVHRR channels because of differences in channel positions and widths for the two instruments. The narrow MODIS near-IR channel is only slightly affected by atmospheric water vapor absorption, while the broad AVHRR near-IR channel is strongly affected by water vapor absorption. As a result, the largest bias comes from the near-IR channels on the two instruments. To a lesser extent, the bias also comes from the differences between the red channel positions and the widths of MODIS and AVHRR instruments. In this paper, the authors describe a practical method for simulating AVHRR NDVI values using several narrower MODIS channels in the 0.4-1.0 /spl mu/m spectral range, including the MODIS green channel and the water vapor absorption channel.