ARS | Publication request: Understanding uncertainty in distributed flash flood forecasting for semiarid regions 1909

Submitted to: Water Resources Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: March 3, 2008
Publication Date: May 22, 2008
Citation: Yatheendradas, S., Wagener, T., Gupta, H., Unkrich, C.L., Goodrich, D.C., Schaffner, M., Stewart, A. 2008. Understanding uncertainty in distributed flash flood forecasting for semiarid regions. Water Resources Research, Vol. 44, W05S19, doi:10.1029/2007WR005940.

Interpretive Summary: One-third of the earth’s surface can currently be classified as arid or semi-arid. This fraction may increase in the future for example due to global warming effects. Many arid and semi-arid regions are particularly affected by flash floods, caused mainly by convective storm systems, and often resulting in significant damages to life and property. The short duration and the small geographic extent of these events make predicting the subsequent floods extremely difficult. To improve our predictive flood capability, we have developed a semi-arid specific model based on the well-established USDA-ARS event-based rainfall-runoff model KINEROS2. The model is driven by high-resolution, near real-time radar rainfall inputs. The ability of this modeling system to predict floods was investigated as well as how uncertainty in the rainfall data and the model affects these predictions. The model was found to make good predictions when calibrated but it was found that the predictive uncertainty of the model is dominated by the radar-rainfall depth estimates. This points out the need to use radar rainfall data and ground-based rain gauge data to improve predictions.

Technical Abstract: Semi-arid flash floods pose a significant danger for life and property in the US. One effective way to mitigate flood risk is by implementing a rainfall-runoff model in a real-time forecast and warning system. This study used a physically based, distributed semi-arid rainfall-runoff model driven by high resolution radar rainfall input. The predictive utility of the model and its dominant sources of uncertainty were investigated for several forecasted streamflow events within the Walnut Gulch watershed. Uncertainty sources considered were those in the rainfall estimates, in the model parameters, and in the initial conditions. A variance-based global sensitivity analysis technique indicated that the uncertainty in the modeled response was heavily dominated by biases in the radar rainfall depth estimates. Further, the study of sensitivity of response to model parameters indicated the need for improved representation of semi-arid hillslope hydrology in small basins, while pointing to specific influential, but poorly identified, hillslope and channel parameters towards which field investigations should be directed.