University of Twente Student Theses

Login

Compare UAV laser and image data for flood modelling

Sakala, Katebe (2020) Compare UAV laser and image data for flood modelling.

Link to full-text:https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2020/msc/gfm/sakala.pdf
(only accessible for UT students and staff)
Abstract:Floods are a significant challenge that can cause considerable damage and impede development. There is a need to capture data that can help control, manage, and mitigate this disaster. The emergence of UAVs has made the capture of this data more accessible, and at very high resolution. UAVs mounted with a camera or laser system can be an excellent choice for this. But looking at the cost of these systems, which data can be used for flood modeling in the Msimbazi river basin, located in the city of Dar es Salaam in Tanzania, which is prone to flooding. Hence the purpose of this study was to analyze and compare the quality of DTMs generated using UAV laser and UAV image data to demonstrate the application in flood modeling. Therefore to produce DTMs suitable for flood modeling, filtering of the point clouds into ground and non-ground is very important. So for this purpose, the PTIN algorithm, as implemented in Lasground, was used in this study. Six parameters that are crucial in the tuning of this algorithm are step, spike, bulge, standard deviation, offset, and sub. These parameters were tuned in both the LiDAR and DIM data and accuracy assessment done on both the point clouds and DTMs. Furthermore, an investigation was done to ascertain how tuning these parameters affect flood modeling in terms of extent, velocity, water surface elevation, and depth. The findings from the comparison of the LiDAR and DIM DTMs and point clouds indicate that as the parameters are being tuned, this heavily influences what gets added to the produced surface. This filtering consequently affects flooding. HEC-RAS modeling software was used for flood modeling. The analysis also showed that different landcover influence parameter settings when filtering. Also, the study showed that as the resolution of DTM reduces, flood extent increases because the DTM is more simplified at coarser resolutions, hence things like ramps, riverbanks, embankments may be removed. It was found that different parameter settings either increase or decrease the flooding effect. The analysis also showed that parameters that had the most significant influence when tuning the LiDAR data were bulge, step, and spike. And for DIM, the settings with the most significant impact were step, bulge, and offset. From the findings, it was observed that both LiDAR and DIM could be used for flood modeling. The only considerations should be landcover and terrain characteristics in the study area. Also, it is essential to know the limitations of both datasets, questions, like where does LiDAR work best and where does DIM work best should be answered. The most important thing is the removal of macro objects such as buildings, vegetation, bridges, and the preservation of ramps, riverbanks, embankments, dividers, etc., especially when the DTM is for flood modeling.
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:https://purl.utwente.nl/essays/85192
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page