Thesis (MSc)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Floods have been identified as one of the major natural hazards occurring in South Africa. A disaster risk assessment forms the first phase in planning for effective disaster risk management through identifying and assessing all hazards that occur within a geographical area, as required by the Disaster Management Act (Act No. 57 of 2002). The National Water Act (Act No. 36 of 1998) requires that flood lines be determined for areas where high risk dams exist and where new town developments occur. However, very few flood hazard maps exist in South Africa for rural areas. The data required for flood modelling analysis is very limited, particularly in rural areas. This study investigated whether flood hazard maps can be created using the existing data sources. A literature review of flood modelling methodologies, data requirements and flood hazard mapping was carried out and an assessment of all available flood-related data sources in South Africa was made. The most appropriate data sources were identified and used to assess an evaluation site. Through combining GIS and hydraulic modelling, results were obtained that indicate the likely extent, frequency and depth of predicted flood events. The results indicate that hydraulic modelling can be performed using the existing data sources but that not enough data is available for calibrating and validating the model. The limitations of the available data are discussed and recommendations for the collection of better data are provided. / AFRIKAANSE OPSOMMING: Vloede is van die vernaamste natuurlike gevare wat in Suid-Afrika voorkom. 'n Ramprisiko-analise is die eerste stap in die proses van suksesvolle ramprisiko-beplanning deur middel van die identifisering en analise van alle gevare wat voorkom in 'n geografiese gebied, soos vereis deur die Rampbestuurwet (Wet 57 van 2002). Die Nasionale Waterwet (Wet 36 van 1998) bepaal dat vloedlyne slegs vir gebiede waar hoë-risiko damme voorkom en vir nuwe uitbreidingsplanne in dorpe vasgestel moet word. Egter is die data wat vir vloedmodelleringsanalises benodig word baie skaars in Suid-Afrikaanse landelike gebiede. Hierdie studie het ondersoek of vloedgevaar-kartering met die beskikbare data moontlik is. 'n Literatuurstudie oor vloedmodelleringsmetodologieë, data-vereistes en vloedgevaarkartering is voltooi en alle beskikbare vloed-verwante data in Suid-Afrika is geëvalueer. Geskikte data-bronne is gekies en gebruik om 'n toetsgebied te assesseer. Deur GIS en hidrouliese modellering te kombineer, is die omvang, waarskynlikheid en diepte van die voorspelde vloedgebeurtenisse gemodelleer. Die studie het bevind dat, alhoewel vloedgevaarkartering met die beskikbare data moontlik is, daar nie genoeg data beskikbaar is om die model te kalibreer en te valideer nie. Tekortkominge van die bestaande data word bespreek en aanbevelings oor die verbetering van die bestaande data vir toepassings in vloedgevaarkartering word gemaak.

This interdisciplinary doctoral dissertation addresses land subsidence in different and diverse study cases in the world, employing advanced techniques and methodologies to measure their magnitude and comprehensively explore its causes, and implications. Investigating areas such as the San Luis Potosi metropolitan area, Alaşehir-Sarıgöl sub-basin (ASSB) in Türkiye, and the Alto Guadalentín Valley in Spain, the research unveils critical insights into the complex dynamics of subsidence phenomena. Utilizing advanced remote sensing techniques like Persistent Scatterer Interferometry (PSI) and Coherent Pixels Technique (CPT), the study assesses subsidence rates and correlates them with factors such as trace faults, groundwater extraction, and soft soil thickness. Validation methodologies were developed and proposed to the scientific community on the first stage, integrating Global Navigation Satellite System (GNSS) benchmarks, enhance the reliability of Differential Synthetic Aperture Radar Interferometry (DInSAR) measurements, ensuring a robust foundation for subsequent analyses. The research aims to contribute to the understanding of land subsidence and contribute to create a decision-support framework to mitigate the phenomenon while addressing specific research objectives within each identified topic of inquiry. The research topic 1 includes the “DInSAR for monitoring land subsidence in overexploited aquifers”. In the San Luis Potosi metropolitan area (Mexico), the application of CPT technique reveals intriguing correlations between trace faults, land subsidence, and groundwater extraction. Specifically, areas in the municipality of Soledad de Graciano Sánchez exhibit subsidence values ranging between -1.5 and -3.5 cm/year, while in San Luis Potosi, values range from -1.8 to -4.2 cm/year. The validation of CPT results against five Global Navigation Satellite System (GNSS) benchmarks establishes a robust correlation of 0.986, underlining the reliability of InSAR-derived deformations. Additionally, in regions like the Alaşehir-Sarıgöl sub-basin (Türkiye), where water stress is heightened due to intensive agricultural irrigation, the study explores the roles of tectonic activity and groundwater withdrawal in land subsidence. Utilizing the P-SBAS algorithm, 98 Sentinel-1 SAR images in ascending orbits and 123 in descending orbits were analysed, covering the period from 2016 to 2020. Independent Component Analysis was applied to distinguish long-term displacements from seasonal variations in the DInSAR time series data. Displacement rates of up to -6.40 cm/year were identified, thus, the proposed P-SBAS algorithm facilitates the monitoring of displacement, revealing direct correlations between DInSAR displacement and critical factors like aquitard layer compaction. These findings contribute valuable insights into the dynamic interactions shaping overexploited aquifers. The research topic 2, developing parallelly to topic 1, consists of the “Validation of DInSAR data applied to land subsidence areas”. Addressing the imperative for validation methodologies in subsidence assessments, a systematic approach introduces statistical analyses and classification schemes. This methodology is designed to validate and refine DInSAR data, enhancing the reliability of subsidence assessments. By normalizing Root Mean Square Error (RMSE) parameters with the range and average of in-situ deformation values and employing the squared Pearson correlation coefficient (R²), a classification scheme is established. This scheme facilitates the acceptance/rejection of DInSAR data for further analyses through the application of automatic analysis supported by a Matlab © code, ensuring a more accurate representation of land subsidence phenomena. The research topic 3 covers the exploitation of DInSAR data for assessing flooding potential and determining characteristic parameters of aquifer systems. The first one is “Impact of land subsidence on flood patterns”. The study in the Alto Guadalentín Valley, a region experiencing extreme flash floods jointly with high-magnitude land subsidence, integrates flood event models, Differential interferometric SAR (DInSAR) techniques, and 2D hydraulic flow models. Through Synthetic Aperture Radar (SAR) satellite images and DInSAR, land subsidence's magnitude and spatial distribution are quantified. The results demonstrate significant changes in water surface elevation between the two 1992 and 2016 temporal scenarios, leading to a 2.04 km² increase in areas with water depths exceeding 0.7 m. These outcomes, incorporated into a flood risk map and economic flood risk assessment, underscore the pivotal role of land subsidence in determining inundation risk and its socio-economical implications. The research offers a valuable framework for enhancing flood modelling by considering the intricate dynamics of land subsidence. The second application of DInSAR data is about the “Automatic calculation of skeletal storage coefficients in aquifer systems”. In response to the need for automating data analysis for specific storage coefficients in aquifer systems, a MATLAB© application is introduced. This application streamlines the correlation between piezometric levels and ground deformation, significantly reducing analysis time and mitigating potential human interpretation errors. The developed application integrates temporal groundwater level series from observation wells and ground deformation data measured by in-situ or remote sensing techniques (e.g., DInSAR). Through the automatic construction of stress-strain curves, the application contributes to the estimation of skeletal storage coefficients, offering a valuable tool for evaluating aquifer system behaviours. This comprehensive research, guided by the complexities of these three distinct research topics, yields detailed insights and methodological advancements. By integrating diverse datasets and employing advanced techniques, this dissertation offers a multidimensional understanding of land subsidence dynamics and provides a robust foundation for sustainable groundwater management globally. / This research is funded by the PRIMA Programme supported by the European Union (Grant agreement 1924), project RESERVOIR.