Using novel data sources and data driven methods to improve hydrological monitoring and forecasting for water and food security in developing regions

Extreme hydrological events such as floods and droughts account for the majority of impacts of all natural hazards, in terms of human and economic costs. Populations in developing countries are particularly vulnerable and may be increasingly at risk because of climate change. Reducing these impacts requires a range of mitigation measures to be put in place, including early warning of hazards using hydrological monitoring and forecasts. To date, our ability to monitor and forecast conditions has generally been modest at best, because of lack of on the ground data, especially in developing regions, and the low skill of climate and hydrological models to predict future conditions. Current forecasts are also often focused on metrics with limited utility for decision-making, such as seasonal average rainfall. Therefore, the uptake of forecasts by stakeholders has been low, despite their potential benefits. In developing countries, this may also be hampered by lack of capacity to develop and use such information. Despite this, the skill of models is improving incrementally as we understand climate and hydrological systems better, and new methods for collecting and integrating data are developed that have the potential to transform how we develop and constrain forecasts.

This project aims to leverage from the wealth of new satellite remote sensing and new data sources, such as low cost sensors, citizen science, and social media to improve the skill of monitoring and forecasting, and to explore data driven methods such as data assimilation and machine learning to improve integration of data and skill of forecasts. This work will be focused on the following objectives:

• Analyse the information value of remote sensing and novel data sources for hydrological monitoring and forecasting
• Develop and apply assimilation methods for integrating these data sources into hydrological forecast models
• Compare these with data-driven machine learning methods based on traditional and novel data sources.
• Apply to real-world situations (e.g. short-term flood risk or seasonal drought) in collaboration with stakeholders

The PhD project brings together supervisors with expertise in hydrological sciences including monitoring, forecasting and data assimilation (Sheffield) and computer based methods for data gathering, integration and machine learning (Ramchurn). The project will be supervised in the context of several ongoing broader research projects that are aimed at different aspects of water and food security in developing regions. These include the GCRF GROW project BRECcIA “Building research capacity for sustainable water and food security in drylands of sub-Saharan Africa” which is working with 11 institutions in the UK and Africa. The student will have opportunities to work closely with other students and researchers on these projects, including via fieldwork and stakeholder engagement.

The successful candidate will receive both Faculty training, and specific training in hydrological modelling and forecasting, sensor based technologies, and data assimilation/machine learning methods.

The ECaS research group focusses on climate change impacts and adaptation, sustainability science, and global environmental monitoring including innovative use of Earth observation data, including Earth system science. We have a world-leading reputation for research on climate change impacts and adaptation strategies, with lead authorships in the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment report.

Candidates must have or expect to gain a first or strong upper second class degree, in an appropriate discipline, not necessarily Geography, with good computational skills. For the latest information see www.southampton.ac.uk/geography/postgraduate/research_degrees/studentships.page?

The PhD project will commence September 2019.