Modeling Biomass and Timber Volume by Using an Allometric Growth Model from Landsat TM Images

Qingmin Meng, Chris J. Cieszewski, Roger C. Lowe, Michal Zasada

Warnell School of Forest Resources, University of Georgia

Athens, GA 30602

Abstract:
Estimation of biomass and timber volume is important not only for forest management and planning but also for forest ecology. Most models used in the estimation of biomass or timber volume from remote sensing images are linear, fixed-effect models, and their simulations are based on sampling site information, such as pixel information. In this paper on integrating forest inventory data and Landsat TM data, we describe an estimation from the use of linear fixed-effects models and linear mixed-effects models based on the allometric growth model. The predicator, either the volume of the normalized difference vegetation index (NDVIvol) or the surface area of the normalized difference vegetation index (NDVIsa) extracted from the Triangulated Irregular Network (TIN), implies the amount of trees both for site and space. With the aid of image-processing and the 3-D analysis extensions of ArcView GIS software, three dimensional models are created using TINs for 144 counties in the state of Georgia. The performance of models shows that linear mixed-effects models are much better than linear fixed-effects models. The NDVIvol is better than NDVIsa for modeling biomass and timber volume, and both the NDVIvol and NDVIsa appear much better for indicating the overall index of biomass and timber volume than indicating any special index, such as sawtimber volume and volume of the sawlog portion. The mixed-effect models reveal that different regions have different allometric characteristics of biomass and timber volume compared with the increase of the amount of trees in the state of Georgia.

Author Keywords:

allometric growth models, linear mixed-effects models, the volume of NDVI, Landsat TM data, biomass, timber volume, Triangulated Irregular Network

KeyWords Plus:

Addresses:

Publisher:

IDS Number:

ISSN: