Search

Identifying and understanding the value of citizen science to improve flood modeling is of importance to flood risk management. However, there are few studies that explore the value of citizen science data, with most studies focusing on evaluating the accuracy of the data. This research articulates the added value of citizen science data in flood modeling studies. During flood events, citizen scientists measured river water levels at selected sites along a main reach of the Big Akaki River in Addis Ababa, Ethiopia. They also provided information to estimate water discharge of the ungauged tributaries. The data acquired was used to force a one-dimensional (1D) HECRAS flood model, and to evaluate the model's sensitivity to inputs and parameters. Varying the downstream boundary condition caused a significant difference in the simulated water level (up to 3.5 ​km upstream of the downstream boundary site). Correcting the Digital Elevation Model and consideration of river tributary flows in the model simulation resulted in an underestimation of the observed stage by 0.08 ​m. The sensitivity analysis also showed that results were more sensitive to the Manning roughness values of the channel than that of the floodplain. Finally, this study identifies future flood modeling data collection priorities (e.g. flow data for the tributary). The flood modeling of the study area would not have been realized without the citizen science data.

Introduction: Cities located in lower income countries are global flood risk hotspots. Assessment and management of these risks forms a key part of global climate adaptation efforts. City scale flood risk assessments necessitate flood hazard information, which is challenging to obtain in these localities because of data quality/scarcity issues, and the complex multi-source nature of urban flood dynamics. A growing array of global datasets provide an attractive means of closing these data gaps, but their suitability for this context remains relatively unknown.Methods: Here, we test the use of relevant global terrain, rainfall, and flood hazard data products in a flood hazard and exposure assessment framework covering Addis Ababa, Ethiopia. To conduct the tests, we first developed a city scale rain-on-grid hydrodynamic flood model based on local data and used the model results to identify buildings exposed to flooding. We then observed how the results of this flood exposure assessment changed when each of the global datasets are used in turn to drive the hydrodynamic model in place of its local counterpart.Results and discussion: Results are evaluated in terms of both the total number of exposed buildings, and the spatial distribution of exposure across Addis Ababa. Our results show that of the datasets tested, the FABDEM global terrain and the PXR global rainfall data products provide the most promise for use at the city scale in lower income countries.