Satellite Product Validation

The Need For Validation

Remote sensing is the process of collecting and interpreting information about an object without being in physical contact with the object. When we make estimates of net primary production from space, we do not measure changes in CO₂ concentrations, or C gas exchanges. Rather, we measure light, that has been reflected off an area of the earth's surface, and then infer canopy light harvesting as an input to approximate rates of production in that area. It is necessary to validate these estimates using ground based measurements, in order to define how accurate the estimates are likely to be. These tests help us assess the reliability and creditability of the products, as well as define any limitations the products have.

Product Comparisons and Validations

The two graphs below show comparisons of the MODIS NDVI product to NDVI derived from finer resolution Landsat and IKONOS imagery. We found that the MODIS product systematically overestimates NDVI relative to both Landsat and IKONOS products. This could be a result of sensor differences or variability introduced through changes in scale.

Comparison of MODIS NDVI to Landsat ETM+ NDVI	Comparison of Landsat ETM+ NDVI to IKONOS NDVI

A comparison of the MODIS FPAR product to our field-based measurements of FPAR demonstrates that MODIS overestimates FPAR in boreal fire disturbed regions. This has been documented by other investigators in other study sites, and is an area of active research. Part of the answer is likely related to differences in field versus satellite measurements of ground cover.

Comparison of MODIS FPAR data to field measurements of FPAR at a 1987 Burn Site in Delta Junction, AK

©Woods Hole Research Center, 2007