COVARIANCE OF BIOPHYSICAL DATA WITH DIGITAL TOPOGRAPHIC AND LAND-USE MAPS OVER THE FIFE SITE

Sampling design is critical in locating ground sampling stations for large-scale climatological field experiments. In the stratified sampling design adopted for the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE), the study region was stratified into 14 different terrain units based on land use/land cover and topographic variables that were hypothesized to have a strong influence on surface biophysical properties. Digital terrain maps were produced to facilitate ground data integration and extrapolation. This paper describes the biophysical stratification of the FIFE site, implementation of the stratification using geographic information system (GIS) techniques, and validation of the stratification with respect to field measurements of biomass, soil moisture, Bowen ratio (beta), and the greenness vegetation index (GVI) derived from thematic mapper satellite data. Maps of burning and topographic position were significantly associated with variation in biomass, GVI, and beta. The effects of burning and topography were stronger for the Konza Prairie Long-Tem Ecological Research (KPLTER) site than for the rest of the FIFE site, where cattle grazing was a major confounding effect. The stratified design did not appreciably change the estimated site-wide means for surface climate parameters but accounted for between 25 and 45% of the sample variance depending on the variable. The design was weakened by undersampling of several strata, by high within-station variance in soil and vegetation data, and by failure to account for diverse land management practices on private lands surrounding KPLTER. We recommend that future large-scale climatological studies include the development of a digital terrain data base well in advance of field campaigns and that multitemporal imagery be used to obtain preliminary estimates of spatial and temporal variance in surface biophysical properties. We also recommend that sampling for the most heterogeneous biophysical variables be conducted in the framework of a multistage estimation scheme incorporating remotely sensed data. Although this means that ground-based estimation of regional fluxes cannot be made independent of aircraft or satellite data, it may well be the only means of obtaining reliable estimates of these variables over large areas.