Applying airborne γ-ray and DEM-derived attributes to the local improvement of the existing individual-tree growth model for diameter increment

- We used airborne γ-ray and DEM-derived attributes for improving the tree growth model.
- Airborne γ-ray and DEM-derived attributes were effective in reducing RMSE and BIAS.
- Species responses differed: the substantial improvements were in broadleaved trees.
- Predictions derived for Norway spruce improved only slightly.
- The improved predictor was found to be the most accurate on less fertile sites.

Airborne gamma-ray data (γ-ray), measuring gamma radiation naturally emitted from the earth's crust has proved useful for predicting the character and distribution of soil properties in forested landscapes. In addition, digital elevation models (DEMs) provide a reliable source of information regarding the hydrological properties of soils. The growth potential of a forest stand is an important parameter in forest management and planning, so that accurate prediction of growth is needed. The present study looked into applying γ-ray data in combination with DEM-derived attributes for improving the existing individual-tree growth model. To adapt the national model to local conditions, the data of 1118 sample trees and 9987 tally located within 197 sample plots in South-eastern Finland were used. Trees were distributed subjectively to reflect the stands' composition and structure. The main aim was to test the suitability of airborne γ-ray in combination with DEM-derived attributes for localizing a general parametric prediction model for the trees' growth (diameter at breast height increment for the period of five years). Linear mixed effect procedures were used to fit models derived from the γ-ray and DEM. Of the various models constructed for comparison purposes, the best result was obtained with broadleaved trees, followed by Scot pine (Pinus sylvestris L.) while Norway spruce (Picea abies L.) revealed little improvement. The improvement was found to be more accurate in less fertile site types (Vaccinum type (VT) and Calluna type (CT)) as well as on mineral soils. The result was found to be effective in reducing the root mean square error (RMSE) and the bias.