Epsilon Archive for Student Projects

Abstract

The digital map product kNN-Sweden contains nationwide data of the country’s forest land. The estimated forest variables are mean age, mean tree height, volume per tree species and total volume. Today Södra Skog’s assessments of timber potentials are mainly based on experience rate and subjective assumptions. Several of the company’s important functions and activities can probably be improved by a comprehensive and updated description of forest conditions in the operation area.

The purpose of this study was to investigate how data from kNN-Sweden (with reference years 2000 and 2005 respectively) relate to forest management plans created by Södra Skog per evaluated variable and area size. Another intention of the study was to identify user needs in operational activities that can be supported by data from kNN-Sweden. The study began with a survey of user needs, which was later on linked to the reports discussion by proposals for implementation options.

The results show relatively large estimation errors on the stand level. Aggregated areas, however, show substantially reduced errors. Large differences also exist between the individual variables in terms of both root mean square error and bias, expressed in per cent of true mean. These differences are similar in kNN-Sweden 2000 and 2005.

Södra Skog should be able to use data from kNN-Sweden to support a number of functions and activities. As expected, reliability of the data decreases with increased geographical resolution, therefore overviews of large areas are initially most appropriate. Evaluating map products of more operational nature i.e. higher resolution, however, indicates together with the survey study that Södra Skog can use data from kNN-Sweden also for this purpose. Practical implementation probably requires Södra Skog to summarize data in more suitable classes, definitions and layout.
These conclusions are based on assuming kNN-Sweden to be a free product requiring only low processing costs.