Boundary-dependent urban impacts on timing, pattern, and magnitude of heavy rainfall in Istanbul

Abstract

An emphasis on whether the modeled response of rainfall to urban signature is consistent across different boundary forcings is lacking. Here, using the Weather Research and Forecasting (WRF) model, we compare urban impacts in the boundary-dependent simulations of a record-breaking rainfall event in Istanbul, designing urban and nourban land cover scenarios. The data used for the boundaries are ECMWF Reanalysis v5 (ERA5) and NCEP Final Analysis (FNL), and we perform simulations with variable boundary and parameterization configurations and use their ensemble mean in our analysis. Our results highlight that some aspects of the urban impact on rainfall produced by different boundary conditions resemble each other. First, hourly rainfall intensifies over and upwind part of the urban domain independent of the employed boundary forcing. Especially within the urban borders, the boost in rainfall amounts during the investigation period totals 6% and 21% for the simulations forced with ERA5 and FNL, respectively. In addition, our simulations exhibit urban signals not only in the rainfall amounts but also in their timing and concentration. This signal in timing is evident by the urban rainfall system that moves slightly ahead of its nourban counterpart at each analyzed hour. We, in particular, observe a spatially less concentrated rainfall pattern as the system positions over the urban area and point out a potential urbanization-topography interaction. We further support these findings with cross-section and integrated water vapor transport analyses. By contrast, the main dissimilarities between the simulations forced with different boundary data include rainfall changes downwind of the city center. This behavior is commensurate with the increase in the temperature and sensible-heat flux values that are not similar in strength in the simulations forced with ERA5 and FNL. These findings suggest that different boundary data can promote similar urban impacts only to some extent, underlining the importance of considering their regional performance before using them in modeling studies.