Predicting the site index of sugi plantations from GIS-derived environmental factors in Miyazaki Prefecture

Abstract

We constructed sugi site index models using digital-terrain-analysis-based environmental factors for Miyazaki Prefecture. We selected 18 sugi plantation stands which were pure, undisturbed, and over 40 years old, planted with the same sugi-cutting cultivar, and managed by normal forest operations. The dominant tree in each stand was felled for stem analysis. Site index, defined here as dominant tree height at 40 years old, was estimated by stem analysis for each stand. Five types of DEMs were used: 100- and 50-m resolution derived from DEMs published by the Geographical Survey Institute, and 50-, 25-, and 12.5-m resolution derived from digital contour map manually generated from a 1:25,000 topographic map. A total of 14 indices categorized into solar radiation index, wetness index, and topographic exposure index were used to model the site index by multiple linear regression analysis. Through model selection procedures, the best-fitted site index models were selected for each type of DEM. The most precise model was that of the 12.5-m-resolution DEM (R2 = 0.692), following the model of the published 50-m-resolution DEM (R2 = 0.665). Site productivity of sugi was severely limited by direct solar radiation. Soil wetness also affected sugi site productivity; however, it can only be represented using a high-resolution DEM derived from fine-scale data. Our results suggest that the 50-m-resolution DEM published by the Geographical Survey Institute can be used for site index modeling.