CHAPTER II
CONCLUSIONS

There are limits to computer simulations. This stumping study was not designed to examine spatial variability of bulk density or total nitrogen.. A study of spatial scales of variation of soil properties should be designed for that purpose. There are many caveats that should be noted before using GIS analysis, and these results are examples of GIS use that should be considered preliminary, pointing out areas for more precise analysis. Goovearts (1998) reviews proper usage of geostatistics for characterizing spatial variability of soil properties. The random grouping of three subplots is not the best arrangement for geostatistical analysis, which should be evenly distributed, but does provide a validation data set for predictions made from plot averages. The plots at the stumping sites are consistently spaced at an average of 26.8 m. The number of plots per site (39-56) is low for geostatistical analysis of variation.

The geostatistical assumption that points that are close together should be more similar than points farther apart will depend on the scale and the variable of interest. Bulk density and total nitrogen are normally fair candidates, as local conditions are likely to influence their levels. The idea of mean stationarity, equal means and variances over the site, or intrinsic stationarity, equal variance of data at similar distances over the site that are assumed for geostatistical analysis should be considered for each particular variable. In this study, with two different treatments, post-treatment stationarity is unlikely, and reduces the effectiveness of analysis across treatments for variability.

It can be seen, however, that the tools of GIS analysis allow insights into the spatial aspects of the results from this study that may not be obvious from traditional statistical analysis. There are inherent patterns in nitrogen at Hoodsport and Sweethome and in bulk density at Gates that may influence these results.

GIS facilitates mapping and presentation of the data and helps forest managers in their decision making. The concepts and applications of GIS are useful in advancing our understanding of the soil as an interactive component of an ecosystem at local and landscape scales.