Geostatistics in evaluating forest damage surveys: considerations on methods for describing spatial distributions

Mapping observed needle/leaf loss onto the sampling grid to obtain the spatial distribution of forest damage can easily lead to misinterpretations. These can arise if the needle/leaf losses on sample plots are not representative of the area surrounding the plot. Often it cannot be decided if the localization of a single needle/leaf loss is due to a local trend or a random observation in between the normal spatial variation. Geostatistical methods are applied in this paper to show the spatial distribution of needle/leaf losses based on the average needle/leaf loss of individual plots. Geostatistical methods are briefly described, with special emphasis on the construction of variograms. Applicability of the methods are demonstrated using a data set of the Swiss Forest Damage Assessment in 1986 and 1990. Geostatistical methods are an ideal tool for epidemiological studies. By combining the spatial distribution of forest damage with other spatially related data, important directions for causal research can be found. Comparison of spatial distribution between years can reveal significant short-term changes. (C) 1997 Elsevier Science B.V.