A Hybrid Approach to Detecting Impervious Surface at Multiple Scales
Investigators: Marina Alberti (PI), Stefan Coe, Yan Jang, Robin Weeks
Funding: National Oceanic and Atmospheric Administration
Detecting impervious surface in urban areas is critical to understanding the effects of urbanization on ecological processes. However, it presents unique challenges due to the spatial and spectral heterogeneity of the urban surfaces and the rapid changes in land cover that occur over short time periods. In this project, we develop a hybrid approach that combines an object-oriented and a pixel-based classification approach. Our approach integrates remotely sensed data – Landsat, Ikonos, and Lidar – and parcel data to develop a spatial database featuring urban object information at multiple spatial scales and class resolutions. Towards these objectives, eCognition™ software is used to perform image segmentation, nearest neighborhood classification, and the development of semantic rules incorporating object attribute information.