Photogrammetric Engineering & Remote Sensing · 2002 · 30 citations · 0 references
Natural SciencesGeoinformaticsUrban PlanningGeographic Information SystemsRemote SensingSpatial AnalysisLand Cover Change DetectionLand Cover MappingUrban ExpansionUrban Expansion DetectionUrban GrowthUrban Remote SensingUrban Change CharacterizationUrban Change DetectionUrban Change MappingUrban Expansion MonitoringUrban Expansion Pattern AnalysisUrban Growth AnalysisUrban Growth MonitoringUrban Mapping
This paper develops a texture analysis procedure to distinguish built from non-built pixels at a meaningful scale of resolution to serve planning needs, and applies the technique to imagery from two dates in order to monitor and measure urban patterns accurately. The methodology used digitally merged Landsat thematic mapping and SPOT-P data, resampled to 10 meters, for 1990 and 1995. For each date, a two-step texturing analysis resulted in a binary built/non-built map defining urban versus non-urban super pixels. Results from the study clearly defined growth pixels for the five-year time interval, with an accuracy of 92 percent. These pixels were then compared to a potential map produced by a geographic information systems analysis based on environmental inducements and constraints to growth. Results showed that the portions of the study area that were rated highest in potential did in fact experience the largest amount of urban expansion. Application of this technique over time is shown to provide an effective tool to identify, map, monitor and quantify patterns of urban and change.