Glennie, Craig L.2019-09-10December 22018-12December 2https://hdl.handle.net/10657/4396Impervious surfaces are land covers that do not allow water penetration. Water runoff from impervious surfaces can cause major flooding in extreme climates; therefore, mapping such surface covers in urban areas is of great importance for water resources, climatology, and urban studies. Automated land cover classification of the remotely sensed data is a necessary step before impervious surface maps are compiled. Very high resolution, passive satellite imagery is the modern source of data for land cover classification and impervious cover mapping. However, shadowed areas and relief displacement in urban areas significantly limit impervious surface mapping accuracy. Recently, multispectral airborne lidar sensors have become available which have the ability to scan the ground at three different laser wavelengths simultaneously. The multispectral point cloud has the capability of being used as the sole source of data for land cover classification because they can simultaneously provide both spectral and geometric information. This dissertation proposes a machine learning classification approach to classify the land cover into diverse classes that could be used for urban change studies, using multispectral lidar points directly such that the 3D information of the points is retained. Two methods to mitigate the multi-echo effect are proposed to reduce the influence of this effect on the spectral information of ALS returns. Furthermore, hybrid intensity correction schemes are devised and tested to improve the classification accuracy. Next, the impervious surface map product is created using the classified points and it is shown that this map yields higher accuracies compared to a similar map created from hyperspectral passive imagery in shadowed areas (by ~21%), areas obstructed by relief displacement (by ~19%), and areas obstructed by tree canopies (by ~40%.) This demonstrates the advantage of multispectral ALS data for automated, high-accuracy, and high-resolution impervious surface mapping.application/pdfengThe author of this work is the copyright owner. UH Libraries and the Texas Digital Library have their permission to store and provide access to this work. Further transmission, reproduction, or presentation of this work is prohibited except with permission of the author(s).Impervious surface mappingLiDARMultispectral LiDARLand cover2019-09-10Thesisborn digital