Impact of Flood Inundation on Local Weather Patterns

The impact of flood inundation on local weather pattens are investigated in this study, for the case of the flood event that impacted the Canadian regions of Ottawa, Gatineau, Montreal and surroundings during the spring of 2017 using high-resolution (4 km) regional climate model simulations, with and without flood inundation regions. In the absence of an interactive inundation parameterization in the regional climate model, flood inundated regions/grid cells are prescribed based on flood extent polygons derived from Radarsat-2 satellite imagery in the simulation that includes inundation. The flood event for most of the regions of interest was caused by rain-on-snow events, with the heavy rainfall in May being associated with a mid-latitude cyclone.The control simulation, without inundation, driven by ERA5, accurately depicts the overall regional patterns of precipitation observed compared to Daymet, but exhibits large overestimation locally, i.e., in the vicinity of the flooded regions, during the 24 hours following the peak rainfall period. However, simulations with flood inundation reveal discernible variations in precipitation over the flood inundation regions, when compared to the control simulation, reducing the precipitation biases locally. This is brought by the temperature modulations and thereby temperature gradients in the simulation with inundation leading to circulation changes locally. Sensitivity experiments reveal robust results, i.e., improved precipitation with inundation representation, despite the small changes in precipitation associated with circulation changes for small changes in flood inundation fractions. The results suggest the need to incorporate dynamic flood inundation module in climate model simulations, especially at high resolutions, to improve the realism of simulations, which in addition to facilitating process understanding can also aid in adaptation planning of inundated regions.