Global continuous fields of vegetation characteristics: a linear mixture model applied to multi-year 8 km AVHRR data

As an alternative to the traditional approach of using predefined classific ation schemes with discrete numbers of cover types to describe the geograph ic distribution of vegetation over the Earth's land surface, we apply a lin ear mixture model to derive global continuous fields of percentage woody ve getation, herbaceous vegetation and bare ground from 8 km Advanced Very Hig h Resolution Radiometer (AVHRR) Pathfinder Land data. Linear discriminants for input into the mixture model are derived from 30 metrics representing t he annual phenological cycle, using training data derived from a global net work of scenes acquired by Landsat. We test the stability and robustness of the method by assessing the consistency of results derived independently f or each year in the 1982 to 1994 AVHRR data set. For those forested locatio ns where land cover variability would not be expected, the percentage woody estimates displayed standard deviations over the 12 years of less than 10% . Problems with the method occur in high latitudes where snow cover in some years and not others produces inconsistencies in the continuous fields. Ov erall, the results suggest that the method produces fairly consistent resul ts despite apparent problems with artifacts in the multi-year AVHRR data se t due to calibration problems, aerosols and other atmospheric effects, bidi rectional effects, changes in equatorial crossing time, and other factors. Comparison of continuous fields with other land cover data sets derived fro m remote sensing suggests 69% to 84% agreement in the per cent woody field, with the highest agreement when per cent woody is averaged over the 12 yea rs. In comparison with regional data sets for the US and Bolivia, the metho d overestimates per cent woody vegetation for grassland and sparsely wooded locations. We conclude that the method, with possible refinements and more sophisticated methods to include multiple endmembers, improved estimates o f endmember values and nonlinear responses of vegetation to proportional co ver, can potentially be used to indicate changes in land cover characterist ics over time using multi-year data sets as inputs when perfect calibration and consistency between years cannot be assumed.