University of Twente Student Theses


Assessment of Population’s Exposure to Flood in A Leptospirosis Endemic Slum Area in Brazil Through the Integration of 2D and 3D Data

Dhakal, Surakshya (2021) Assessment of Population’s Exposure to Flood in A Leptospirosis Endemic Slum Area in Brazil Through the Integration of 2D and 3D Data.

[img] PDF
Abstract:Flood is a risk factor for the spread of leptospirosis disease, which is caused by Leptospira bacteria that spread through water. Slums that receive intense rainfalls, lie in floodplains, and lack storm water drainage systems are prone to flooding. Since leptospirosis is endemic in Salvador, Brazil, the slum communities like Alto do Cabrito are thought to be at a high risk of leptospirosis as the flood brings people in contact with the Leptospira bacteria. It is difficult to eradicate leptospirosis but exposure to flood water that spread the Leptospira bacteria in the slums can be reduced. To reduce the exposure, it is important to identify flood-prone areas and those in contact with the flood water. This study aimed to estimate building-level population in Alto do Cabrito, Salvador, identify flood-prone areas by simulating floods, and assess exposure of buildings and population to flood waters. Since leptospirosis is endemic in Salvador, an assumption that the exposure to flood is proportional to the exposure to Leptospira motivated the aim. To disaggregate the census-sector population of Alto do Cabrito to building-level population, a 3D dasymetric method that utilized building footprints derived from aerial images and building heights derived from digital terrain models was used. To simulate the flood water depths for 1-in-2-year, 1-in-5-year, and 1-in-10-year design storms, a 2D hydraulic modelling was conducted. To visualize flooding, 3D models of the water depths during and at the end of a 24-hr rainfall event were produced. To inform the wet periods that influence Leptospira bacterial growth and leptospirosis incidence, PERSIANN CDR, a satellite rainfall product, was analysed for seasonality of rain. To assess the exposure of buildings and people to flood, the number and percentage of buildings and population in close proximity to pooled water was estimated. To visualize the exposure, water depths juxtaposed with buildings and building-level population normalized by building heights were produced. Additionally, in a small subset of the population for which leptospirosis sero-survey data was available for Alto do Cabrito, an association between the sero-status and the exposure to pooled water was assessed. The results of the population disaggregation showed that the numerical measure used to summarize the building height, which is used to estimate the building volume and the population, as well as the rounding method to convert the estimated population to an integer approximation affected the precision of the disaggregated population. For Alto do Cabrito, the maximum of pixel values as the numerical measure of the building height with the population rounded to the nearest integer resulted in the 2010 building-level population with a low error. Consequently, this building-level population was used for exposure analysis in this study. According to the PERSIANN rainfall data, the rainfall varied annually. The months of April to June comprised the wet season but PERSIANN underestimated the rainfall. The flood simulations showed that high intensity rainfalls caused high rainfall-runoffs. At the end of the 24-hr rainfall events, most of the rainwater flowed away from the study area to the lowlands. The remaining water pooled along roads and around buildings. During the rainfall events, the water rose to the levels that inundated buildings of 6-7m and less in height. Since the water pooled across the study area, the use of proximity thresholds to assess the exposure showed that almost all of the study area was exposed to flood. The exposure to flood waters was likely to be more in small buildings housing a larger number of people. However, the small Odds Ratios (ORs) and the 95% confidence intervals (CIs) that contained the null value showed uncertainty in the association between leptospirosis and floods. In conclusion, the flood simulations aided to identify areas where water pooled after heavy 24-hr rainfall events. With the building-level population, a better estimate of people exposed to flood could be made. The 3D visualization of the flood was helpful to understand the exposure to pooled water during and after the heavy rainfalls. The association between leptospirosis and flood was uncertain. The uncertainty in the association is attributed to the study design, exposure misclassification, and various risk factors that were not considered in this study.
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