|Title||Quantification of Cartographic Generalization in Land Cover Maps Using Spatial Pattern Index Measurements Derived from Digital Satellite Imagery|
|Year of Publication||1992|
|University||University of California|
This thesis presents a methodology for detecting potential errors in vegetation maps developed by interpretation of remotely sensed imagery. Land cover maps derived by photo interpretation of remotely sensed satellite imagery suffer from analyst-dependent cartographic errors such as over-generalization, poor boundary placement and misclassification. Image interpretation is subjective and generally inconsistent among maps prepared by different interpreters. The mapping process represents a great simplification of the spatial and spectral information in the imagery. The hypothesis of this research is that spatial pattern indices derived from satellite imagery and retained as polygon attributes help to preserve some of the original spatial and spectral information, and can be used to detect cartographic errors due to misclassification, boundary misplacement and excessive generalization. The approach involves establishing a distribution of by-class pattern index values to detect outliers in the distribution. Results indicate that the procedure is promising for enforcing cartographic consistency. In addition to the potential for error detection, information on within-polygon heterogeneity may be of ecological or socioeconomic interest.