Integrating spatial statistics into a multi-resolution classification framework to improve the mapping of urban land use/cover

The fundamental problem involved in producing accurate urban land use maps from remotely sensed data is that urban areas comprise a complex spatial assemblage of land cover types. With the recent increasing availability of 1–5 m high-resolution satellite and airborne digital images, the problem of spatial variability in each urban land use/cover type is more acute, making it more difficult to use the traditional approach of single-resolution classification. This paper presents a multi-resolution analysis and classification framework for selecting and integrating suitable information from different resolutions and analytical techniques into classification routines. Information on the spatial structure of the image should be acquired from spatial statistics for each land use/cover class prior to and during the classification. The multi-resolution approaches were tested using simulated multi-resolution images for a portion of the rural-urban fringe of the San Diego Metropolitan area. We demonstrate that the multi-resolution classification approaches integrating techniques of spatial analysis can significantly improve the accuracy of urban land use/cover classifications as compared with single-resolution approaches.