| Abstract: | The paper is prompted by apparent deficiencies in the design of plot studies in regional erosion surveys. The principal shortcomings of observational erosion research have been poor sampling design and inadequate analyses of data. The paper identifies various sources of bias which must be taken into account before plot data can be extrapolated to land units in a regional survey. Judging from soil loss data of a case-study in the Ardèche rangelands one may conclude that even accurate plot measurements can still be rather a rough basis for regional erosion assessment. Finally, the paper highlights strategies that might be used to improve erosion sampling. |
