University of Twente Student Theses


Correction of Photogrammetrically Derived DSM’s for Hydrodynamic Modelling Using Object Oriented Analysis

Shrestha, Nawaraj (2010) Correction of Photogrammetrically Derived DSM’s for Hydrodynamic Modelling Using Object Oriented Analysis.

[img] PDF
Abstract:Hydrodynamic models (HDM) have been widely used in predicting flood events. Digital elevation models (DEM) are important in HDM as they strongly control the water movement. As such, the topographic features represented in DEMs play vital role in accurate prediction of any flood event by HDMs. Photogrammetrically derived digital surface models (DSM) represent earth’s topography along with the elevated features on it. In HDM all the elevated features represented in DSM block the water flow although in reality only the features impermeable to water (e.g. house) block the water flow and permeable features (e.g. tree) allow the flow. As such, the impermeable features have to be present while the permeable features have to be removed from DSM. In addition, the artefacts due to shadow and occlusion also lead to erroneous terrain representation which have significant effect in HDM and have to be corrected for accurate prediction. Thus, use of corrected DSM for accurate prediction from HDM is necessary and is termed as ‘Pseudo-DTM’. This research describes a method that integrates Object oriented approach (OOA) with common tools available in remote sensing and GIS software to get Pseudo- DTM. OOA is used to identify and extract the features from orthophoto and DSM. The extracted features are used to selectively remove the terrain values from DSM. The removed terrain values are then predicted using the interpolated height values. Two approaches are used to remove the permeable features and artefacts from the DSM. The object based approach (OBA) uses shape of features to define the area to be removed from DSM, while the object based buffer (OBBA) approach considers formation of object based buffer (OBB) along with the features. The features such as trees, humps, groynes, embankments were extracted with accuracies greater than 85% while houses were extracted at accuracy of 70%. The misidentification between trees and houses is most sensitive during removal process. The results from simple and compound cases showed that the OBA approach was able to remove the features leaving behind the artefacts while OBBA approach was able to remove the remaining artefacts. The applicability of the method however, has not been tested on areas where trees occur on slope, embankment and depressions. The described approach shows the integration of OOA with GIS operation to get correct Pseudo-DTM for HDM. The correction procedure however, largely depends on proper identification of the features. The use of Pseudo-DTM thus should provide with reliable local and bulk flow prediction from HDM’s.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page