Impact of precipitation forecast uncertainties and initial soil moisture conditions on a probabilistic flood forecasting chain

- We analyzed the behavior of a operational probabilistic flood forecast chain by using synthetic experiments.
- The chain uses quantitative expert precipitation forecast and considers both single-site and multi-catchment approach.
- We investigated the propagation of the errors on precipitation forecast to the streamflow forecast.
- We analyzed the effects of different initial soil moisture conditions.
- Finally we studied the impact of the ensemble size on the final results.

SummaryOne of the main difficulties that flood forecasters are faced with is evaluating how errors and uncertainties in forecasted precipitation propagate into streamflow forecast. These errors, must be combined with the effects of different initial soil moisture conditions that generally have a significant impact on the final results of a flood forecast. This is further complicated by the fact that a probabilistic approach is needed, especially when small and medium size basins are considered (the variability of the streamflow scenarios is in fact strongly influenced by the aforementioned factors). Moreover, the ensemble size is a degree of freedom when a precipitation downscaling algorithm is part of the forecast chain. In fact, a change of ensemble size could lead to different final results once the other inputs and parameters are fixed. In this work, a series of synthetic experiments have been designed and implemented to test an operational probabilistic flood forecast system in order to augment the knowledge of how streamflow forecasts can be affected by errors and uncertainties associated with the three aforementioned elements: forecasted rainfall, soil moisture initial conditions, and ensemble size.