A Three-Step Approach to Model Tree Mortality in the State of Georgia

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

Abstract:

Tree mortality is one of the most complex phenomena of forest growth and yield. Many types of factors affect tree mortality, which is considered difficult to predict. This study presents a new systematic approach to simulate tree mortality based on the integration of statistical models and geographical information systems. This method begins with variable preselection using multiple linear regression models and logistic models and employs spatial autocorrelation detection and random sampling. Three random sampling methods are applied and compared to reduce the affects of spatial autocorrelation, and systematic random sampling significantly reduces the spatial autocorrelation among the observations and is used for the final variable selection and model fitting. Using Forest Inventory and Analysis (FIA) data for the State of Georgia, this systematic approach provides significant implications for future tree mortality studies and other spatial analysis in forestry or geography.

Author Keywords:

KeyWords Plus:

Addresses:

C. CIESZEWSKI. Warnell School of Forest Resources,  University of Georgia   Athens, GA 30602

Publisher:

IDS Number:

ISSN: