A GIS-based spatial pattern analysis model for ecoregion mapping and characterization.

摘要

The growing concerns on global climate change, biodiversity maintenance, natural resource conservation, and long-term ecosystem sustainability have been converting traditional single resource management into integrated ecosystem management. Ecoregions are large ecosystems of regional extent that contain smaller ecosystems of similar response potential and resource production capabilities. They can be used as a geographical framework for organizing and reporting resource information, setting bio-ecological recovery criteria, extrapolating site-level management experience to regions, and monitoring global changes.; The objective of this research is to develop a quantitative, multivariate regionalization model which is able to delineate ecoregions of multiple levels from remotely sensed data, and/or other in-situ data of environmental and land resource characteristics. The Spatial Pattern Analysis Model developed in this research uses a seeded and a non-seeded region-growing algorithm to generate spatially contiguous regions from polygonal primitive land units. Seeded region-growing assigns a land unit to a pre-defined seeded region at each step if the merge satisfies similarity criteria. The non-seeded method merges the most similar pair of neighboring units at each step until all units are grouped into one.; An ecoregion map of Nebraska with three hierarchical levels is developed with the model. In the mapping process, the STATSGO data set was used to build primitive map units. 66 clusters were initially grouped as Level III regions, 20 Level II regions and two Level I regions were generalized after further clustering. Environmental parameters include multi-temporal AVHRR data, soil rooting depth, organic matter content, available water capacity, growing degree days, and annual moisture deficit. Ecological characteristics of Level I and Level II regions are summarized from the GIS database and other natural resource and agricultural reports of the area.; Development of the Spatial Pattern Analysis Model provides a new approach of ecoregion mapping to resource managers and researchers. The mapping procedure is automated and efficient. Judgment biases and uncertainty of manual analysis are reduced. The method can be replicated for other regions or regionalization of other themes.