|
Cycle 9 (2020 Deadline)
Long-term impacts of land-use/land-cover dynamics on surface water quality in Botswana’s reservoirs using satellite data and artificial intelligence methods: Case study of the Botswana’s Limpopo River Basin (1984-2019)
PI: Yashon Ouma (yashon.ouma@mopipi.ub.bw), University of Botswana
U.S. Partner: Jiaguo Qi, Michigan State University
Project Dates: May 2021 - March 2024
Project Overview
The rising demand for water, food, and energy due to increasing population continues to create immense pressure on water resources. In particular, water quality around the globe is systematically degrading, primarily due to climate change and agricultural intensification associated with rapid population growth and urbanization. In-depth assessments of the inter-linkages between land–water resources that combine land-use and water quality and availability within the catchment supply chains such as the Limpopo River Basin (LRB) in southern Africa are still lacking. Semi-arid Botswana relies on the reservoirs within the LRB for water supply, which are particularly susceptible to the negative impacts of land-use and land-cover (LULC) activities and runoff because of their complex dynamics, relatively longer water residence times, and their role as an integrating sink for pollutants from their drainage basins.
This PEER project used data-driven artificial intelligence for quantitative determination of the relationships between LULC change, together with socioeconomic development indicators and climate change and their impacts on water quality and availability within the basin, both for 1984-2019 and to predict future scenarios (2020-2050). To advance data acquisition for LULC analysis and climate change, the study used optical Earth-observation and meteorological satellite data. To provide near real-time and cost-effective approach for continuous monitoring of reservoir water quality within the basin, the study sought to develop empirical models for water quality estimation and water quality index mapping using 35-years of in-situ water quality measurements and water spectral observations using drone-borne spectrometer and optical satellite imagery through regression modeling and geospatial methods.
Final Summary of Project Activities
In a joint field work campaign with the Water Utilities Corporation (Botswana), the PI and his team successfully carried out their planned water quality sampling for the Bokaa and Gaborone dams, including simultaneous drone imaging. The researchers downloaded data from Sentinel-2 MSI and Landsat-8 ETM+ satellite sensors and modeled their water quality predictions with data from the sensors. The field work also involved LULC classification ground-truth campaigns within the dam catchments.
The researchers developed machine-learning models for mapping and quantifying the spatial-temporal LULC change patterns in the Botswana LRB from 1984-2019, modeling water quality and quantity in the dam catchments under climate variability and socio-economic factors. They also developed a Land-Water Nexus (LWN) for the LRB, using climate factors, socioeconomic factors, and WEAP hydrological modeling software, establishing the interactions and relationships between land use and water demand and supply-indifferent regions. The resulting monitoring tools and models are freely available for replication and will be made available after all the publications of the results in peer-reviewed journals.
Five undergraduate students were part of the PEER project as research assistants/interns and acquired advanced skills in imaging using Earth Observation analytics (drones and satellites) and machine-learning algorithms. Three postgraduate students are continuing with different components of the research project, and the PI and other staff participated in training on advanced imaging using drones for resource mapping. Through the project funding for infrastructure upgrades (drone, RTK Base Station, computers, and computer accessories), the Hydroinformatics Engineering Research Group (HERG) now has a well-equipped laboratory. PEER support has also bolstered the research activities of the academic staff and the postgraduate students in the areas of water resources, climate change and land-use studies, and the application of GeoAI technologies, including Earth Observation analytics and machine-learning/AI.
As the key stakeholder on dam water resources monitoring and management, the project established a working Memorandum of Agreement with WUC (Botswana) on water quality sampling and testing for dams in Botswana. The collaboration extends to technology transfer for planning of water quality sampling protocols for monitoring dams, and as of May 2024 WUC was in the process of evaluating and adopting the use of drone technology for water quality monitoring. The PEER team also received a $200,000 grant from the Alliance for African Partnership for work on sensors and smart infrastructure in community health.
Publications
Ouma, Y.O., B. Nkwae, P. Odirile, D.B. Moalafhi, G. Anderson, B. Parida, and J. Qi. 2024. Land-Use Change Prediction in Dam Catchment Using Logistic Regression-CA, ANN-CA and Random Forest Regression and Implications for Sustainable Land–Water Nexus. Sustainability 16(4): 1699. https://doi.org/10.3390/su16041699
Ouma Y.O., A. Keitsile, B. Nkwae, P. Odirile, D. Moalafhi, and J. Qi. 2023. Urban land-use classification using machine learning classifiers: comparative evaluation and post-classification multi-feature fusion approach. European Journal of Remote Sensing 56(1): 2173659. https://doi.org/10.1080/22797254.2023.2173659
Ouma, Y.O., D.B. Moalafhi, G. Anderson, N. Boipuso, P. Odirile, B.P. Parida, and J. Qi. 2022. Dam Water Level Prediction Using Vector AutoRegression, Random Forest Regression and MLP-ANN Models Based on Land-Use and Climate Factors. Sustainability 14(22): 14934. https://doi.org/10.3390/su142214934
Ouma Y.O., M. Ditiro, G. Anderson, B. Nkwae, P. Odirile, B.P. Parida, N. Sebusang, T. Nkgau, and J. Qi. 2022. Predicting the variability of dam water levels with land-use and climatic factors using Random Forest and Vector AutoRegression models. Proceedings of SPIE Remote Sensing 2022: Remote Sensing for Agriculture, Ecosystems, and Hydrology XXIV, 122620J. September 6-9, 2022, Berlin, Germany. https://doi.org/10.1117/12.2635933
Ouma Y.O., B. Nkwae, D. Moalafhi, P. Odirile, B. Parida, G. Anderson, and J. Qi. 2022. Comparison of machine learning classifiers for multitemporal and multisensor mapping of urban LULC features. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B3-2022: 681-689. XXIV ISPRS Congress, June 6-11, 2022, Nice, France. https://doi.org/10.5194/isprs-archives-XLIII-B3-2022-681-2022
Ouma Y.O. et al. 2022. Land-Water (L-W) Nexus Project: Impacts of Land-Use & Climate Change on Water Quality and Quantity in Botswana's Limpopo River Basin (BLRB). United Nations/Ghana/PSIPW - 5th International Conference on the Use of Space Technology for Water Resources Management. Accra, Ghana, May 10-13, 2022. https://www.youtube.com/watch?v=YXjWnaxTXDg
Back to PEER Cycle 9 Grant Recipients
|
|
|
|
