High-Resolution Downscaling of Urban Land Surface Temperature via Machine Learning

Green roofs can help cool hot cities, and the Weather Research and Forecasting model lets us test how much. We upgraded its multi-layer green-roof scheme, adding plant shading, realistic heat flow in wet soil, evaporation from soil and wet leaves, and water uptake from deeper soil. Tested on a green roof in London, Canada, the new scheme cut errors in heat entering the roof by about three-quarters and captured bursts of evaporation after rain, improving how we study green-roof cooling.

Green roofs can help cool cities, but models must represent them accurately to quantify this potential. Yet few studies evaluate green roof models against real data. We evaluated two versions of EcoRoof, the green roof module in EnergyPlus, and a green roof option for WRF (Weather and Research Forecasting model) using measurements from London, Ontario. EcoRoof generally matched observed heat fluxes, surface temperature, and soil moisture, while WRF overestimated heat and underestimated cooling.