Integrated modelling of water security in data-sparse regions under uncertainty

Student thesis: Doctoral Thesis

Abstract

Freshwater scarcity and sustainability is one of the most complicated and difficult issues the world is currently facing, and it has been identified as a global concern. According to expert studies, 80% of the world’s population is projected to live in freshwater threats due to a plethora of factors viz., rapid population growth, urbanization, global climate change resulting from spatial and temporal changes in magnitude, frequencies and intensity of precipitation and temperature which leads to the transformation of the hydrologic cycle. Recent initiatives, including sustainable development goals, have been made to address these problems and offer solutions. However, the quantity and quality of freshwater systems and resources must be objectively and comprehensively understood and assessed at the scale of river basins to provide sufficient mitigation and resilience planning.Hydrologic modelling has been one the most suitable and efficient strategies for basin-scale assessment of freshwater dynamics to current and projected climate change and the focus has been on the application of traditional modelling framework which is tenable where data requirements are sufficient to couple hydrologic models with atmospheric data to account for climate change.The aforementioned strategy is a challenge in regions with inadequate ground-based observations necessary for climate and hydrologic modelling. The rarely available data in such regions may have repetitive gaps of missing data points with negative consequences including biased statistical representation of basin climatic features, ineffective model calibration and unreliable timing of peak flows which may amplify the uncertainties of the hydrologic dynamics leading to flawed depictions of watershed responses.Recently, integrated strategies are evolving that couple hydrologic models with climate data in water resource studies to account for uncertainties through the use of alternative data sources of many spatial climate data products from climate research centres to overcome the identified challenges.This research developed and applied a multi-criteria approach to examine the efficacy of gridded climate products using different performance metrics, a machine learning-based approach, Boruta random forest (BRF) to assess multiple GCM datasets required for hydro climatic studies and an integrated BRF-SWAT technique to define the relationship between the hydrologic variables and improve rainfall-runoff modelling in a data-sparse and climate sensitive watersheds.The developed model was applied to assess the projected green and blue water dynamics and sustainability in the Yobe-Komadugu basin of the greater Lake Chad, a watershed that is prone to extreme events (SPEI of flood and drought hazards). The results demonstrate that though the performance of the gridded data varies in space and time, multi-criteria assessment enhances the choice of a product with reduced uncertainty for climate modelling.The incorporation of the BRF approach in GCM evaluation indicates a consistent spatial and temporal representation of the climatological features with suitable mean correlation (R2 = 0.95), reduced mean annual precipitation bias of 0.69 mm/year and enhanced statistical trend and magnitude of the SPEI drought and flood hazards relative to identified and tested approaches from the literature.The integrated framework of the rainfall-runoff modelling strategy indicated that the hydrologic fluxes can be simulated fairly accurately with varying degrees of acceptability, irrespective of the watershed morphological properties, although there are significant trade-offs in model parameter sensitivity. The availability of satellite-based measurements of hydrologic fluxes and states, coupled with a machine learning feature selection and data refinement process has made integrated water balance modelling widely seen as a viable alternative for improving watershed hydrologic processes in data-sparse regions within acceptable uncertainty limits.Furthermore, the sub-watershed assessment of the projected changes in spatial and temporal green and blue water sustainability status has shown that the sub-basins will be ecologically fragile, and the identified freshwater geographic hotspots may be beyond restoration without adequate long-term river basin water resources plans. The modelling framework developed is, however, independent of the model and data type and can be applied to watersheds with similar modelling challenges.This study has provided a pathway or methods for managing and securing water resources information as a decision support tool to guarantee ongoing watershed monitoring and assessment of water security even in the face of increasingly unpredictable future circumstances in data-sparse watersheds that take into account uncertainty and chat a course for prospective risk assessment or the possibility and understanding that a certain effect brought on by climate-induced hazards would prevail in watershed freshwater sustainability.Therefore, it is essential to comprehend the constraints associated with forecasting changes in the water cycle to improve the climate and hydrologic modelling process, which is required to create effective strategies for adapting to climate change-related water resource hazards. Even in the face of severe uncertainty about the future, this will be essential in addressing concerns related to water security and management and promoting the climatic resilience of ecosystems and society.
Date of Award7 Dec 2023
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorDoug Bertram (Supervisor) & Chris White (Supervisor)

Cite this

'