Model validation


The model has been validated by (i) comparing its growth functions against growth functions of other models (Sterba 2003), (ii) comparing its projections against other projections carried out for the same forests (e.g. Nilsson et al 1992; Böttcher et al. 2012), and (iii) running the model on historic data and comparing the output to present day forest state. Separate validation studies have been conducted for Finland with EFISCEN 1.0 (Nabuurs et al. 2000) and for Switzerland with EFISCEN 2.2 (Thürig and Schelhaas 2006).

EFISCEN 1.0 was applied to Finland for the period 1921-1990. The initial situation was based on the results of the first national forest inventory, carried out in the years 1921-1924. Simulation results were then compared to corresponding later inventories. This validation study showed that projected forest resource development was able to capture observed forest resource development for 50-60 years. Differences in increment level were found after 30-40 years, with visible consequences in growing stock levels after 50-60 years. Another conclusion of the validation was that the results at a national level were fairly close to observed levels, but that results at a lower level (per species and region) showed much more deviation. An important reason for this seemed to be differences in management intensity between regions and between tree species.

EFISCEN 2.2 was validated for Switzerland. Firstly, the matrix as initialised by EFISCEN was compared with the original plot data from the second Swiss National Forest Inventory. The result of this comparison was in general satisfying. The largest deviations occurred in poor sites in the Alps region. This was attributed to the fact that forests on such sites usually have a protective rather than a productive function. Such stands are generally managed in an uneven-aged way, leading to a different distribution of growing stocks over age than in case of forests that are managed in a truly even-aged way. Secondly, the forest resource development was projected for the canton Bern for different sets and time spans of inventory plots where repeated measurements were available. On the aggregated (national) level the model produced results comparable to the observed values. However, at a more detailed (forest type) level results deviated sometimes considerably, due to differences in management intensity between regions and between tree species. Moreover, it was concluded that the current structure of the model is not suitable to simulate the uneven-aged, selective management as practised in the Alpine region.

References:

Böttcher, H., Verkerk, P.J., Gusti, M., Havlík, P., Grassi, G. 2012. Projection of the future EU forest CO2 sink as affected by recent bioenergy policies using two advanced forest management models. GCB Bioenergy 4(6): 773-783.
Nabuurs, G.-J., Schelhaas, M.-J., Pussinen, A., 2000. Validation of the European Forest Information Scenario Model (EFISCEN) and a projection of Finnish forests. Silva Fennica 34, 167-179.
Nilsson, S., O. Sallnäs and P. Duinker. 1992. A report on the IIASA forest study: Future forest resources of Western and Eastern Europe, The Parthenon Publishing Group, UK
Thürig, E., Schelhaas, M.-J., 2006. Evaluation of a large-scale forest scenario model in heterogeneous forests: a case study for Switzerland. Canadian Journal of Forest Research 36, 671-683.