Deciding on fitness‐for‐purpose ‐ of models and of natural flood management

This article has taken a distributed modelling approach to examine the potential for hillslope storage bunds as a natural flood management (NFM) strategy to mitigate the effects of downstream flooding if deployed extensively in the 209 km(2) River Kent catchment in Cumbria (UK). To do so, a novel invalidation approach has been taken to calibrate an ensemble of Dynamic Topmodel parameter sets considered as fit-for-purpose for the prediction of flood peaks. Only 67 realizations out of 100 000 survived evaluation at the peaks of ranked flood hydrographs including the notable 2005, 2009 and 2015 events, together with a process constraint on the minimum area contributing saturation overland flow in the largest storm of record. These 67 events were used as the baseline (before intervention results) to simulate the effect of introducing 1549 storage bunds to temporarily retain saturation overland flows in optimal hillslope locations. The analysis is extended to simulating the reduction of inundation and damage estimates by running the ensemble of Dyanmic Topmodel simulations as inputs to the distributed hydraulic model JFlow for the 2015 Storm Desmond event. Even given such a large number of interventions the effects on the flood peaks and damages in this case were relatively small. The results of the study emphasize the large amounts of storage required in any NFM scheme, with individual features large enough that they are not too quickly overwhelmed, in order to have a significant impact on flood hydrographs and consequent flood damages.