Supercells are the heavyweight champions of local severe storms: deep, long-lived thunderstorms whose updrafts rotate persistently, feeding them the power to generate giant hail, destructive straight-line winds, and most tornadoes. Because surface-based supercells draw their fuel from the lowest few hundred meters of the atmosphere, scientists have suspected for years that the altered surface conditions beneath them—heat, roughness, and moisture supplied by an urban canopy—could meaningfully reshape their evolution. Case studies over the US Great Plains and long-term simulations over Moscow have previously hinted at storm intensification over cities, but those studies either examined single events or relied on heavily simplified urban representations. The new work, led by Francesco De Martin of the University of Bologna and the University of Oklahoma together with Christopher Rozoff, Andrea Zonato, Stefano Alessandrini, and Silvana Di Sabatino, takes a fundamentally more rigorous approach.

The team ran ensembles of idealized numerical simulations with the Weather Research and Forecasting model at one-kilometer horizontal grid spacing, pairing the model with state-of-the-art urban parameterizations: the Building Effect Parametrization and the Building Energy Model, which together represent the effects of roofs, streets, walls, canyon shadowing, radiation trapping, and anthropogenic heat on the atmosphere. The setup was deliberately clean. A flat domain covered in croplands hosted a single circular city, and a supercell was triggered upstream using a thermal bubble so that the storm, after two hours of development, would collide with the urban area near its peak intensity. The environment was tuned to a classic severe weather profile: roughly 2740 joules per kilogram of surface-based convective available potential energy, strong deep-layer shear of about 32 meters per second, and moderate storm-relative helicity in the lowest three kilometers.

What makes the study statistically powerful is its scale. Twenty-five numerical experiments were conducted, each with eleven ensemble members generated using a stochastic kinetic energy backscatter scheme that perturbs the flow in physically realistic ways. This allowed the researchers to separate genuine urban signals from the chaotic variability that always plagues convective simulations. The morphological parameters varied were city size, with diameters up to 30 kilometers; urban fraction, the proportion of the surface covered by built materials; building height, spanning 10 to 60 meters; and building density, ranging from dispersed to highly compact configurations. Parameter ranges were anchored in real morphological data from European and American cities frequently hit by severe storms, using the GLOBUS global building-height dataset, with Milan, Italy serving as the reference city.

The first major finding is a statistically significant weakening of the supercell as city size increases. Updraft helicity—a diagnostic combining vertical velocity and vorticity that tracks storm rotation and intensity—declined consistently in the ensemble means as the city grew wider, with the suppression most pronounced over and downstream of large cities with high urban fractions. In the most extreme configuration, a 30-kilometer-wide, fully built-up city, the suppression of the incoming storm’s updraft helicity swath was nearly complete. In smaller cities the effect was weaker and sometimes not statistically significant, with a few ensemble members even showing transient intensification, a reminder that storm-city interactions remain sensitive to the fine details of the flow.

The culprit behind the weakening is not the buildings themselves but the moisture budget of the city. Urban surfaces, lacking the vegetation of surrounding croplands, evapotranspire far less water, producing an urban dry island. In the simulations the dry island reduced near-surface specific humidity by up to about 2 grams per kilogram and relative humidity by roughly 13 percent, values consistent with observations from Moscow and Beijing. That drying carved out a wake of sharply reduced convective available potential energy extending up to 80 kilometers downwind, with CAPE deficits reaching 1000 joules per kilogram—about twice the reduction reported in a previous Dallas–Fort Worth case study. Because surface-based supercells feed on low-level air, this depleted reservoir of instability starved the storm. The authors note this is the first time an urban-induced CAPE reduction has been directly linked to an urban-induced weakening of severe storms, and the discrepancy with earlier idealized studies that found intensification appears to hinge on the fact that those studies omitted the dry island entirely.

But the city has a second face. The same urban surfaces also trap heat, generating an urban heat island of about 1.5 kelvin in the simulations. Through a quantitative decomposition of pressure perturbations, the researchers showed that this heat island thermodynamically induces a pressure minimum of up to about one hectopascal downstream of the city, driven overwhelmingly by buoyancy rather than by the mechanical blocking of airflow by buildings. That pressure minimum draws in converging surface winds, and along the convergence line, vertical motions intensify, locally reaching one meter per second at the 950-hectopascal level. In the largest and most fully urbanized experiment, this mechanism proved potent enough to do something remarkable: while the original supercell was dissipated by the dry island, new convective cells fired along the convergence corridor and, in 8 of the 11 ensemble members, organized into a brand-new right-moving supercell downwind of the city. Notably, this corridor also coincided with enhanced storm-relative helicity on its northern flank, providing the rotation the new storm needed.

Perhaps the most surprising result concerns skyscrapers. Contrary to the long-standing building-barrier hypothesis, which holds that tall structures mechanically force air upward and can even split storms around them, building height had little impact on supercell dynamics. In fact, taller buildings slightly reduced the urban heat island—deeper street canopies are shaded and retain moisture longer—and consequently weakened the downstream pressure minimum and convergence. Even with 60-meter-tall buildings, comparable to dense districts of New York or Chicago, the vertical velocities generated mechanically by the buildings were negligible compared with those produced by the heat island’s pressure signature. Building density fared little better, emerging as the least influential parameter in the entire study. Together, these findings suggest that in strongly synoptically forced environments, the thermodynamic personality of a city, not its architecture, dictates how it treats passing supercells.

The simulations also carry a hopeful message for urban design. An additional experiment in which every roof of the idealized city was covered with vegetation dramatically muted the city’s effect on the approaching storm, because green roofs restore evapotranspiration and shrink the dry island. For emergency managers and forecasters, the implications are equally practical: the same city that weakens one supercell can seed another downwind, meaning warnings cannot simply assume a storm will decay as it crosses a major metropolitan area. The results align with radar-based climatologies showing that storms approaching US Great Plains cities and Berlin are more frequently weakened than intensified downwind, and they may explain past case studies in which urban-triggered storms appeared to intensify—they were likely new convection born along the urban convergence line rather than a revived original storm.

The authors are careful to note the limits of their idealized framework. A single environmental profile was tested, aerosol effects were not represented, nocturnal events—when the heat island is strongest but storms tend to be elevated—remain unexplored, and the parameterizations, while sophisticated, do not explicitly resolve individual buildings. Large-eddy simulations at finer resolutions and studies across a wider range of atmospheric environments are logical next steps. Yet the core conceptual model stands: a city is simultaneously a moisture thief and a heat engine, capable of starving one supercell while igniting another. For the millions of people living where supercells and skylines intersect, that dual identity is now written into the physics.

Subject of Research: Urban land use effects on supercell thunderstorm intensity and initiation

Article Title: The dual effect of urban areas on supercell storms

Article References: De Martin, F., Rozoff, C., Zonato, A., Alessandrini, S., & Di Sabatino, S. (2026). The dual effect of urban areas on supercell storms. Weather and Climate Dynamics, 7(4), 2015-2042. https://doi.org/10.5194/wcd-7-2015-2026

Image Credits: AI Generated

DOI: 10.5194/wcd-7-2015-2026

Keywords: supercell, urban heat island, urban dry island, convective storms, numerical simulations, WRF model, city size, building height, convection initiation, CAPE, storm-relative helicity, urban meteorology