Development of an object-based framework for classifying and inventorying human-dominated forest ecosystems

Abstract

This paper presents the development of a framework for classifying and inventorying Eastern US forestland based on the level of anthropogenic disturbance and fragmentation using high spatial resolution remote sensing data and a multiscale object-based classification system. We implemented the framework using a suburban area in Baltimore County, Maryland, USA as a case study. We developed a three-level hierarchical scheme of image objects. The object-based, multiscale classification and inventory framework provides an effective and flexible way of showing different mixes of human development and forest cover in a hierarchical fashion for human-dominated forest ecosystems. At the finest scale (level 1), the classification nomenclature describes basic land cover feature types, which are divided up into trees and individual features that fragment forests. The overall accuracy of the classification was 91.25%. At level 2, forest patches were delineated and classified into different categories based on the degree of human disturbance. At level 3, major roads were used to segment the study area into larger objects, which were classified on the basis of relative composition and spatial arrangement of forests and fragmenting features. This study provides decision makers, planners and the public with a new methodological framework that can be used to more precisely classify and inventory forest cover. The comparisons of the estimates of forest cover from our analyses with those from the 2001 National Land Cover Dataset (NLCD) show that aggregated figures of forest cover are misleading and that much of what is mapped as forest is highly degraded and is more suburban than natural in its land use.

Acknowledgements

This research was funded by the MacIntire-Stennis Federal Research Funds, the Northern Research Station, USDA Forest Service, and the National Science Foundation LTER program (grant DEB 042376). The authors would like to thank the two anonymous reviewers for their helpful comments and suggestions.