Modeling urban extreme heat using spatially resolved radiative and morphological parameters

Cities affect local climate by profoundly modifying land use/land cover with associated changes in radiative and morphological properties of land surfaces. Accurately prescribing these boundary conditions is key for urban climate modeling. Fine-scale heterogeneity of urban surface parameters, and a corresponding dearth of intra-urban observations, have been challenges, but global high-resolution urban parameter datasets have recently emerged to address this gap. However, the effects of incorporating these spatially explicit properties on summertime urban climate simulations remain underexplored.

We use the WRF regional climate model to simulate recent extreme heat in 13 US cities at 1 km² resolution with two sets of geographic boundary conditions varying by local climate zone (LCZ) classification: one using default parameters (“default”), and one using the U-Surf satellite-derived, spatially continuous urban parameters (“params”). We show that “params” simulations correspond better with surface temperature observations in terms of both absolute values and intra-urban variation. In large part due to lower impervious fraction estimations, “params” simulations yielded lower air temperature than “default” for most US cities examined, especially in the suburbs and exurbs, leading to a smaller urban heat island. The smaller simulated area of transpiring vegetated land cover resulting from overestimating urban fraction somewhat offsets suburban moist heat overestimations from using default urban parameters.

Finally, we simulate two US cities using continuously varying (i.e. assigned for every grid point) morphological and radiative parameters, finding that this further enhances simulation accuracy. Decomposing the sources of increased accuracy shows that improved impervious fraction is the most important variable. Overall, results suggest that using high-resolution urban parameters enhances accuracy when modeling extreme heat events.