Unmasking the hidden hazard: Convection-permitting modelling reveals extreme tropical cyclone rainfall structures in Southeast Africa

Abstract

Tropical cyclones (TCs) in Southeast Africa pose severe flood risks, yet understanding these risks is hindered by a sparse observational network. While reanalysis products like ERA5 are standard tools for risk assessment, their coarse resolution often smooths out the intense convective features that drive flash flooding. This study challenges the sufficiency of current reanalysis data by evaluating km-scale convection-permitting simulations using the Conformal Cubic Atmospheric Model (CCAM).ERA5 data were dynamically downscaled from 2014 to 2023, identifying six high-impact TCs that affected Malawi, Madagascar, and Mozambique, including the record-breaking TCs Idai and Freddy. These simulations were compared against reanalysis, satellite products, and gauge observations.Results demonstrate that CCAM significantly corrects the underestimation bias found in ERA5 and satellite datasets. Crucially, the km-scale simulations reveal detailed structural features missed by coarser models, including complex inner-core structures and distinct asymmetric outer spiral bands. These structural details are not merely meteorological curiosities; they determine the spatial footprint of the hydrological hazard.It is concluded that relying on standard reanalysis products underestimates the true flood potential of TCs in the region. By resolving these fine-scale storm features, CCAM provides a more realistic baseline for understanding present-day flood risk and assessing future climate risk. This work highlights the critical need for convection-permitting approaches to support effective climate resilience in vulnerable communities.