Influence of cloud-radiative feedback on tropical cyclone motion

Building on past work demonstrating the sensitivity of hurricane track to cloud microphysical assumptions (Fovell, Corbosiero and Kuo 2009; Fovell and Su 2007), idealized simulations have been used to study the influence of cloud processes in general on the motion of tropical cyclones. The simulations were made using the real-data version of WRF with a homogeneous environment without land and considered two versions of Seifert and Beheng's (2006) double-moment microphysical parameterization in addition to the commonly-employed single-moment schemes such as the Lin scheme, WSM3, WSM6, and Kessler. The simulations utilized 4 km horizontal grid spacing in 2500 km x 2500 km domains. The attached figure presents 72 h tracks for two experiments in which cloud-radiative feedback (CRF) is either included or excluded. For the CRF-on experiment (panel a), a fan of tracks are obtained. As an example, storm S2 (using one of the Seifert-Beheng variants) produces a northwest-moving storm while WSM3 (labeled W) yields a more northward-moving cyclone. Starting from the same point, this magnitude and direction discrepancy would result in a 150 km position difference in 48 h, roughly comparable to the average position error at that lead time in the Atlantic basin. Surprisingly, and importantly, the track variation nearly disappears in the CRF-off experiment (panel b). All of the model storms in that experiment translate northward after reaching maturity after about 36 h of integration. In the CRF-off experiment, condensate is not permitted to modulate longwave and shortwave radiation. In these simulations, there appears to be two principal processes competing to determine direction of storm motion: the beta effect, reflecting the advection of planetary vorticity by the hurricane's outer winds (as demonstrated by Fovell, Corbosiero and Kuo 2009), and latent heating owing to convection. We reveal that neglecting CRF profoundly alters the intensity and spatial asymmetry of diabatic heating in the cyclone, altering the relative balance between the beta drift and diabatic effects. This is demonstrated using a potential vorticity analysis motivated by Wu and Wang (2000, 2001). This work strongly motivates more basic research in cloud-radiative processes and microphysical parameterizations, and highlights the need to validate numerical model products with available observations. Supplemental figure: The figure shows 72 h tracks for the idealized WRF experiments, utilizing 6 microphysics schemes: L (Lin et al.), W (WSM3), W6 (WSM6), K (Kessler), and S1/S2 (two versions of Seifert-Beheng). For panel (a), cloud-radiative feedback (CRF) is included. For (b), CRF was deactivated. A section of the Gulf Coast is superposed for scale only; there is no land. Inset in (a) shows storm motion vectors for the last 24 h of the simulation. References: Fovell, Corbosiero and Kuo (2009): Journal of the Atmospheric Sciences. Fovell and Su (2007): Geophysical Research Letters. Seifert and Beheng (2006): Meteorology and Atmospheric Physics. Wu and Wang (2000): Monthly Weather Review. Wu and Wang (2001): Journal of the Atmospheric Sciences.