In order to investigate the health conditions of the Swiss forests every year, a constant set of parameters is  assessed on permanent plots. Different sources of variation lead to erroneous data of needle leaf loss. An  approach to analysis of erroneous data by using self-organizing maps (T. Kohonen) is described.
Self-organizing maps preserve the topological characteristics of data and can be used for data compression with a minimal reconstruction error. The compression can be treated as a generalization of the whole data set and helps to derive cause - effect relationships.
The results of the analysis can be used to visualize varying external influences on the damage assessment and account for them in the design of further surveys. The detection of group - specific variation can be used for  quality control. The generalized spatial panerns can form the basis for an optimal allocation of plots.