Next generation flood hazard mapping for the African continent at hyper-resolution

Flood hazard and risk maps form the evidence base for decision-marking regarding issues such as land use planning, insurance and capital provision, emergency response and disaster preparedness. None of these essential activities could be planned properly without such data and this is recognised by high level policy such as the EU Floods Directive, the Sendai framework and the flood and water management act in the UK. However, across most of sub-Saharan Africa such data are absent posing a huge challenge to disaster risk managers. The high cost and expertise needed to create flood hazard maps is a barrier to their provision in many sub-Saharan countries meaning that innovation low cost solutions are needed if the provision of such maps and associated benefits for risk management are to become universal.
One solution is to use data from global flood models, which have emerged in the last five years, to fill the numerous gaps in coverage. These models make predictions everywhere based on techniques for hydrological prediction in ungauged basins combined with remotely sensed data sets on catchment topography and river size and location. Unfortunately, all global flood models have substantial limitations, such that, the data they produce are usually only considered accurate enough for high level national and transnational risk assessment. This hampers their ability to support a wide range of disaster risk management activities. A second generation of global flood models is therefore needed with sufficient predictive skill and quantification of uncertainty to discriminate risk levels at regional or even community scales. Only with such an advancements will it be possible to transform our understanding of risk and to identify risk hotspots where regional and community level risk reduction efforts would be best focused.
HYFLOOD will improve our understanding of the occurrence, location and intensity of flooding with unprecedented detail by building on an existing global flood model to develop regional to community scale flood hazard maps. We will do this by using the remotely sensed data record on flood occurrence for several satellites to disaggregate river reaches into those that we think go overbank more or less often. This information will be used to locally change the river channel characteristics that will then influence the simulated flood inundation extents, depth and duration for extreme events. By overlaying information on population and land use we will make improved estimates of who and what is exposed to flooding. We will trial our approach with end-users in the Democratic Republic of the Congo via an existing collaboration between the University of Bristol and the University of Kinshasa who host the Congo Basin Network for Research and Capacity Development in Water Resources.
The outcome of the project will be an improved flood hazard map for the African continent that for the first time can include local scale variability in river characteristics and a quantification of prediction uncertainty. This will be accompanied by the first estimate of river bathymetry at continental scale that can be used by other flood hazard and risk modelling groups. Therefore, HYFLOOD will improve our understanding of the hydrological and morphological factors that determine the occurrence, duration and impact of floods.

The beneficiaries of HYFLOOD include any organisation that uses risk information for Disaster Risk Management in Africa, and that will be a potential end-user of the new hazard and risk information that it will deliver. These include:

(1) Governmental and intergovernmental organizations responsible for land use, infrastructure design and management choices for disaster risk management.

For example, we have provided flood risk analytics data to the World Bank for country scale development and for climate change mitigation projects in Belize, Vietnam, Mozambique, Bangladesh, Albania and Fiji. These data are often used to identify hotspots of risk where further risk analysis is warranted. However, since existing global flood models all assume rivers overtop with the same frequency everywhere, there is usually little discrimination between floodplains or regions at risk of flooding in terms of hazard.

(2) International organisations and NGOs working in disaster risk management regionally or nationally.

For example, the World Bank's their web-based 'Think Hazard' tool (http://thinkhazard.org/) enables non-specialists in any country to consider the impacts of flood disasters on new development projects.

(3) Water resources managers in the Congo River Basin.

Through project collaboration with the University of Kinshasa we will exchange data and knowledge via the Congo Basin Network for Research and Capacity Development in Water Resources. This group will benefit from improved flood hazard information around ports, towns and agricultural areas. For example, 40,000 people in the south eastern Democratic Republic of Congo faced food shortages in 2016 after severe flooding had washed away crops. Flood hazard maps could help to make estimates of the cropped land lost to flooding after events and aid disaster response.

(4) Researchers at the University of Kinshasa.

We would also include university researchers from the University of Kinshasa given that in this region most PhD students enter industry directly after graduation, and thus training researchers and increasing research capacity is an effective pathways to deliver impacts outside academia.

Other beneficiaries are private companies who use/provide risk information, and that will both directly use risk information delivered by HYFLOOD or build on the HYFLOOD modelling method to expand their own analytical capacity. For example, our partner Fathom Global (flood analytics company) use similar modelling techniques and would utilise our new river bathymetry estimates for flood prediction for insurers for portfolio management and multinationals for assessing the risk to their operating locations worldwide.

Follow on impacts:

HYFLOOD will facilitate a number of potential follow on impacts. It is relatively straightforward to link inundation maps for particular return periods to outputs from existing forecasting systems. For example, we recently used our global flood model outputs in the USA to produce flood inundation now-casts and short-term forecasts for Hurricane Harvey using USGS gauge observation of return period flows and NOAA forecasts. Similar approaches would be feasible with forecasting systems such as the Global Flood Awareness System used in FATHUM, which is jointly developed by the European Commission and the European Centre for Medium-Range Weather Forecasts. This may in future provide a visually more effective way of communication forecasts in situations the forecasts are thought to have sufficient skill.

Finally, we are open to collaboration with other members of the SHEAR consortium who want to use our data to support/enhance their own objectives and outputs.