Unplanned Natural Experiments: The Case of Remote Sensing of Primary Production and Its Environmental Correlations in the Negev

Abstract

Studies of the correlations of environmental factors with vegetation growth using remotely sensed measurements are necessarily made against a background of biophysical and anthropogenic factors, such as local fertility, microclimate, and the effects of human land use, in addition to the factors of interest. This is an inevitable outcome of a natural (unplanned) design where the effects of the factors of interest are confounded with other, often unknown factors, possibly rendering the results inaccurate or poorly-constrained. The problems associated with a natural design would be reduced if sites could be identified in which uncontrolled variables had no impact. However, rarely are such sites known a priori. Here, a component of the net primary production (NPP) local scaling (LNS) method was used to estimate the potential NPP in the absence of confounding factors. Subsequent analyses of the effects of the selected environmental variables were carried out using the potential NPP. The method was tested in relation to NPP along the transitional ecotone from desert to semiarid conditions in the northern Negev, Israel. The effects of four environmental factors were tested: precipitation, topography, land cover, and interannual variability. While precipitation is generally the only environmental variable that is considered in drylands, the other factors were found to be significant. The results provided unambiguous evidence of the value of the method.