Multiscale analysis of the relationship between topography and aboveground biomass in the tropical rainforests of Sulawesi, Indonesia

Abstract

This article aims to explore spatial and altitudinal non-stationarity in the relationship between aboveground biomass (AGB) of tropical rainforest in Sulawesi (Indonesia) and topography. An autoregressive model through a geographically weighted regression (GWR) framework was used to study the relationship between ground-measured values of AGB and altitude above sea level at 85 sampling plots. The relationships between AGB and altitude were found to be significantly spatially variable and scale-dependent. The results also suggested high altitudinal variability in the examined relationship. Both the strength of the AGB–altitude relationship (ρ) and the altitudinal gradient (α) showed a high changeability in the horizontal and vertical dimensions. The complex spatio-altitudinal patterns in the GWR-based local estimates of the ρ and α parameters gave rise to both spatial and altitudinal variations in the scale effects. The approach presented in this study enables finding the most appropriate scale for data analysis within different altitudinal bands. The study found that the changes of the gradient α along altitudinal transects relate to prevalent environmental conditions observed at different altitudes, whereas the optimal bandwidth was related to the terrain surface heterogeneity.

Keywords:

• biomass
• altitudinal gradient
• topography
• GWR
• non-stationarity
• scale- dependency

Acknowledgements

This study is part of the German–Indonesian collaborative research project STORMA (Sonderforschungsbereich 552: ‘Stability of Rain Forest Margins in Indonesia’, subproject D6) funded by the German Research Foundation (DFG) and the financial support is gratefully acknowledged. The author gratefully thanks C. Zsyschka, BSc, for her great help during the field survey in 2007. Acknowledgements are also extended to S. Erasmi, Ph.D. (Leader of subproject D6), as well as to M. Kessler, Ph.D., and J. Clough, Ph.D., who helpfully provided forest inventory data from their field surveys.