The Assessment of Tree-Volume Distribution in East Texas Ecosystem Using Probabilistic Kriging and Sequential Gaussian Simulation

P. P. Siska 1 and P. Goovaerts 2

College of Forestry, Stephen F. Austin State University (1)
TerraSeer Inc. Ann Arbor, MI (2)

The last decade has witnessed the growing importance of geospatial applications in forest inventory and analysis. The need to accurately evaluate forest resources prompts forest researchers and practitioners to adopting modern tools for accurate mapping and spatial analysis. This project demonstrates the application of probabilistic kriging and sequential gaussian simulation (SGS) in evaluating the timber volume, its spatial continuity and also the uncertainty in volume estimates. Accounting for four hundred ground measurements together with remote sensing stand classification geostatistical simulation was used to generate thirty realizations of the spatial distribution of the timber volume in East Texas. These realizations were averaged and the accuracy of the SGS-based estimates was compared to the results of inverse distance weighting method and ordinary kriging of ground measurements. SGS leads to more detailed assessment of natural resources while providing an empirical model of the uncertainty attached to the final estimates. In addition, the probability for the timber volume to exceed 700 cubic feet per acre was mapped using sequential simulation and indicator kriging of ground samples. The results were evaluated using standard statistical parameters.

Keywords: kriging, sequential gaussian simulation, uncertainty

(presentation)