NORMALIZED DIFFERENCE VEGETATION INDEX MEASUREMENTS FROM THE ADVANCED VERY HIGH-RESOLUTION RADIOMETER

The potential of computing normalized difference vegetation index (NDVI) measurements from the global, daily observations collected by the NOAA Advanced Very High Resolution Radiometer (AVHRR) has created great interest in the use of these data sets to study global biospheric dynamics. Initial qualitative studies demonstrated this potential but the move to quantitative assessments is hindered by limited understanding of the performance characteristics of the AVHRR observations relative to surface vegetation conditions. Factors related to instrument precision and calibration, atmospheric attenuation and off-nadir viewing create deviations of the NDVI observations unrelated to vegetation dynamics. Deviations in excess of 50% between the satellite and equivalent ground observations are possible if no effort is made to account for these effects. In addition, off-nadir viewing causes spatio-temporal variations in the measurements and cloud occurrence reduces temporal resolution below the AVHRR's daily repeat cycle. It appears possible to reduce these errors to approximately +/- 10% (+/- 0.1 NDVI) with at least a monthly time resolution if all of the observation attributes are addressed adequately. Much of the remaining error resides in atmospheric variability, uncertainties in off-nadir views and loss of sensor precision of large solar zenith angles. A global measure of vegetation green foliage dynamics with a measurement precision of +/- 10% and a monthly time resolution may not yet meet the exacting needs of some biospheric modelers, but it is, in fact, remarkable, considering that this possibility was not even conceived when the AVHRR was designed. Further refinement of global remotely sensed vegetation foliage measurement precision appears possible and should be the primary focus for terrestrial remote sensing research in the next decade.