Date of Award

8-2008

Level of Access Assigned by Author

Campus-Only Thesis

Degree Name

Master of Science (MS)

Department

Forest Resources

Advisor

Steven A. Sader

Second Committee Member

Kathleen P. Bell

Third Committee Member

Katherine K. Carter

Abstract

Conversion of forestland is a reality in Maine as growing human populations are exerting development pressures on privately owned forests. Specifically, forestland in the Lower Penobscot and Lower Kennebec watersheds is predicted to face high conversion pressures. Knowledge of forest cover dynamics and human pressures on the forest are vital to making decisions concerning future land use. Modern remote sensing techniques and geographic information systems enable the accurate tracking of forest change and the integration of social, demographic, and biophysical data. This study is designed to detect and map forest cover disturbance in 81 townships across Midcoast Maine using Landsat Thematic Mapper satellite imagery in a three-date time series (2000-2002-2006). Visual interpretation of change detection data using high resolution aerial photography for a 24-township sample provides the validation of forestto- developed conversion. Discrete events of conversion are then modeled using logistic regression and boosted tree regression to understand drivers of forest conversion, and develop predictions where development pressures may impact forestland in the future. Scenarios are constructed using population projections to simulate alternative futures that predict the spatial allocation of development within the study area. Change detection results indicate that -34,580 acres of forest cover were disturbed during the study period, representing 3.32% of the total forest within the study area. Disturbance resulted from harvest, natural decline, and conversion to developed uses. Conversion accounted for 754 of 11,000 acres in the 24-township sample. These data reinforce the overall practical usefulness and accuracy of the forest change detection methods and provide input for the modeling procedures. We identified a set of key drivers of conversion using empirical model fitting methods. Distance to: roads, water, and conservation lands; and population density, surrounding forest, and surrounding developed area were the most influential independent variables included in the final model application. Model calibration and analysis results indicate that people have tended to settle in rural townships, but more so where there is easy access to open space and recreational opportunities, and also in relative proximity to developed areas and the services they provide. Model projections suggest an increase of 12,500 new households across all towns through 2030. Alternative future scenarios estimate that 3,000 to 60,000 acres of forestland could be permanently converted or impacted, offering some insight into potential outcomes of increased population. Identification of forestland at risk from conversion to developed uses could enable decision makers to assess losses in timber productivity and investment, habitat loss, and estimate fiscal burdens on town budgets which often result from low density rural development. We succeeded in gaining a baseline understanding of how the forest in the Midcoast Maine study area has changed, and what role the relative influence of a variety of factors played in conversion to developed uses. Identification of primary drivers of conversion at the landscape scale is relevant for our approach, but future studies should focus on local level analysis using land use zoning and other restrictions to improve planning for sustainable growth and municipal budgets needed to support additional infrastructure and services.

Download

Files over 10MB may be slow to open. For best results, right-click and select "save as..."

Share