Modelling the potential of Natural Flood Management practices in an upland peatland catchment with the distributed TOPMODEL

In order to reduce flood risk, there is a growing interest in more nature-based solutions, such as Natural Flood Management (NFM), rather than hard-engineering ones. To help build the evidence towards these solutions, we analysed the influence of different land cover change and land management in an upland peatland catchment in the Yorkshire Dales, UK. Using the distributed TOPMODEL from Gao et al., we first modelled the impact of revegetation, grazing and prescribed burning on the flood hydrograph. Then, tree planting as woodland and hedgerow, was implemented in the model and its influence on the flood peak was assessed. Revegetation of the whole catchment, from bare peat to \textit{Eriophorum} grasses, reduced the flood peak by 47% and delayed it by an hour and twenty minutes for an 18mm/hour rainfall. Taking a baseline \textit{Eriophorum} cover of the catchment, prescribed burning did not have a significant impact and grazing gave a 45-minute earlier and 31% higher flood peak. Woodland and hedgerow planting on grazed areas both had a positive influence on the flood peak. Planting hedgerows on every field edges in the catchment reduced the flood peak by 14% and delayed it by 15 minutes for the same above rainfall. Different scenarios of woodland planting were analysed: existing, floodplain, riparian and wider catchment woodlands. The most efficient scenarios were floodplain and riparian planting which gave a similar 3-hour late and 59% lower flood peak. The existing woodland scenario was the least efficient (12.5% reduction and 6-minute delay) and the wider catchment scenario was in between (40% reduction and 1-hour delay). Location of planting in the catchment was found to be very important and planting is near-stream zones is more efficient than in the headwaters or midslope of the catchment. Tree type and age at the catchment scale did not have a strong influence. Efficiency of tree planting on flood peak reduction and delay decreases with increasing rainfall intensity and storm duration. Floodplain woodland was found to be the most interesting scenario as flood risk was significantly reduced while only changing the land cover of a small portion of the catchment area (3%).