Print Email Facebook Twitter Mapping Urban Surface Infiltration Capacity: Segment-based land cover classification with VHR imagery for urban water management and design. Title Mapping Urban Surface Infiltration Capacity: Segment-based land cover classification with VHR imagery for urban water management and design. Author Lee, D.J. Contributor Gorte, B. (mentor) Menenti, M. (mentor) Van De Ven, F. (mentor) Faculty Architecture and The Built Environment Department Geomatics Programme Geomatics Date 2013-12-06 Abstract Effective urban water management requires certainty about surface conditions such as the surface infiltration capacity (SIC). A SIC map of an urban catchment area could be a useful input for evidence-based urban design of water-sensitive/low-impact neighbourhoods and multitiered water management schemes for reducing flood vulnerability. Methods for mapping SIC are underdeveloped. Typically, infiltration rates are derived from topographic or urban extent maps. Instead, if a land cover map can precisely identify hydrologically relevant land cover classes, then a more accurate SIC map can be derived. The research explored whether an accurate SIC map could be derived from VHR multi-spectral imagery of Amersfoort, Netherlands, using specific hydrologically relevant land cover classes and segment-based land cover classifiers to achieve meaningful object resolution. The impact of different similarity metrics, rule sets, and data types on classification accuracy were explored. In particular, the impact of lower class specificity (generic classes) was tested. The SIC map was assessed based on impact in a pluvial flood model. Results indicate a high degree of spectral and textural confusion between impermeable and semi-permeable surface types in land cover classification. The analysis of the SIC and land cover maps illustrates that with generic classes the SIC is under-predicted for the catchment area despite a higher overall accuracy in land cover classification. Generic classes also reduce object resolution (detail). Using specific classes produces a less accurate land cover map but improves SIC prediction. In a pluvial flood model, the generic classes increased runoff volume and reduced peak runoff time, validating the conclusion that class specificity and object resolution plays an important role in mapping SIC. Subject remote sensingimage segmentationland cover classificationVHR imageryurban water managementtexture featurespermeability map To reference this document use: http://resolver.tudelft.nl/uuid:c7bd121a-1378-4ca1-b33c-129763fb0287 Part of collection Student theses Document type master thesis Rights (c) 2013 Lee, D.J. Files PDF P5_FinalReport_DanbiLee.pdf 59.68 MB Close viewer /islandora/object/uuid:c7bd121a-1378-4ca1-b33c-129763fb0287/datastream/OBJ/view