Quantitative Multiresolution Characterization of Landscape Patterns for Assessing the Status of Ecosystem Health in Watershed Management Areas

Landscape ecology is a field that has grown from realizing that maintenance of ecological resources requires management at several spatial and temporal scales, including landscape-level ecosystems as whole units of study and management (Forman, 1995; Grumbine, 1994; Noss, 1983 and 1996). The subsequent need for characterizing landscape structure has led to a variety of measurements for assessing different aspects of spatial patterns; however, most of these measurements are known to depend on both the spatial extent of a specified landscape and the measurement grain. Therefore, measurements that incorporate a range of scales would be most informative. In response, this paper introduces a new method for obtaining a multi-resolution characterization of land cover fragmentation patterns within a fixed geographic extent. Our particular interest is in watersheddelineated extents. The method applies the concept of conditional entropy as one moves from larger “parent” land cover pixels to smaller “child” pixels that are heirarchically nested within the parent pixels. When applied over a range of resolutions, one obtains a “conditional entropy profile”. The conceptual and methodological development of conditional entropy profiles is presented, followed by current and future directions for evaluating and applying this methodology.