Canadian Forest Service Publications
Data and model validation methods used in developing the STIM growth model. 1993. Bonnor, G.M.; De Jong, R.J.; Boudewyn, P.A.; Flewelling, J.W. Pages 122-128 in J.K. Vanclay, J.P. Skovsgaard, and G.Z. Gertner, Editors. Growth and yield estimation from successive forest inventories, Proceedings: IUFRO conference. June 14-17, 1993, Copenhagen. Danish Forest and Landscape Research Institute, Lyngby.
Issued by: Pacific Forestry Centre
Catalog ID: 3357
Availability: Order paper copy (free)
To produce the STIM (Stand and Tree Integrated Model) growth model, data from approximately 1400 permanent sample plots were obtained from eight forestry agencies in the Pacific Northwest. A major portion of the model development time was spent in data and model validation.
Data validation procedures included:
- screening tree and plot data to identify anomalies and missing data;
- plotting variables to identify outliers and artificially generated data;
- comparing growth rates and other derived variables to assess trends;
- discarding plots with deficient data.
Model validation procedures included:
- assessing the quality of individual growth predicting equations using various statistical analyses;
- checking individual model prediction runs against real-life data;
- comparing stand projections with the sum of tree projections;
- analysing yield curves to assess long-term trends.