Large-Eddy Simulation of an Idealized Tropical Cyclone

Abstract

A dvances in computing have allowed presentday numerical weather-prediction models to forecast and/or simulate tropical cyclones (TCs) with high resolution (horizontal grid spacing ~1 km). While these numerical models of TCs can capture many of their important structural features (eyewall, rainbands, etc.), the effects of small-scale [~0 (100 m)], three-dimensional turbulence must still be parameterized. No objective basis has yet been found for such parameterizations, yet the maximum wind speeds in axisymmetric numerical TC models have been shown to be highly sensitive to the radial turbulent transfer and diffusion at these scales. As there is little observational guidance on the nature of radial turbulent diffusion in a TC, the present study was conceived to indicate the impact of these effects through computation of the small-scale scale turbulence (i.e., a large-eddy simulation, or LES) using a numerical weather prediction model applied to an idealized tropical cyclone.The numerical experiments reported on herein were carried out with the Advanced Research Weather research and forecasting (ARW) model, version 2.2, using six nested grids centered on the TC. As the ARW model is nonhydrostatic, it can and has been used as an LES model, 1 and therefore one can expect the numerical solutions to capture turbulent eddies, 1 Examples using ARW as an LES model to simulate the idealized convective planetary boundary layer and the sea breeze are listed in the bibliography. given sufficient resolution. Here we show that using a horizontal grid size < 100 m in a TC simulation produces three-dimensional turbulent eddies, which have the effect of decreasing the maximum sustained (1-min average) wind and the maximum predicted azimuthally averaged wind. TURBULENCE EFFECTS INTROPICAL-CYCLONE MODELS. Existing tropical-cyclone models need to simulate circulations reaching outward to a distance of ~0 (1,000 km) from the TC center. Modern-day computers have allowed real-time numerical forecasts and sensitivity studies of the TC with horizontal grid spacing as small as 1 km. We note that TC models fall into the general category of mesoscale model wherein, according to a 2004 Journal of the Atmospheric Sciences essay by J. Wyngaard, the grid scale A is much greater than the length scale I of threedimensional turbulent eddies that may be O (100 m) and smaller, and consequently, all three-dimensional turbulence effects must be parameterized. The parameterization of horizontal diffusion in most mesoscale models is such that the horizontal diffusion coefficient is proportional to A so as to arrest frontal collapse; in the typical case, the front is long and straight and the horizontal diffusion serves its purpose without significantly changing the forecast wind intensity. However, in a TC, the eyewall, which is a type of front (as discussed by K. Emanuel in a 1997 Journal of the Atmospheric Sciences article), is circular, and diffusion exerts great control over the radius of the eyewall (i.e., the radius to which fluid parcels will con...

How Vertical Wind Shear Affects Tropical Cyclone Intensity Change: An Overview

“…Despite the confident pronouncements that resulted from their TC composite analysis, Knaff et al (2004) made the concession that their symmetric analysis could not determine how the warm-core erosion occurs, partially due to the coarse (50km) resolution of the AMSU. Even Rotunno et al (2009) expressed the need for still higher resolution and computing resources, yet lamented the computational cost of the then-current experiments. One of the ways forward appears to be the proper parameterization, or explicit resolution of the turbulent eddies in the near core region of TCs, especially as they relate to the transport of entropy into and out of the TC eye (Frank and Ritchie, 2001;Bryan and Rotunno, 2009;Rotunno et al, 2009).…”

“…Even Rotunno et al (2009) expressed the need for still higher resolution and computing resources, yet lamented the computational cost of the then-current experiments. One of the ways forward appears to be the proper parameterization, or explicit resolution of the turbulent eddies in the near core region of TCs, especially as they relate to the transport of entropy into and out of the TC eye (Frank and Ritchie, 2001;Bryan and Rotunno, 2009;Rotunno et al, 2009). The impact of this turbulence is large, as, when parameterized, it looks to consistently affect both the radius of the eyewall and the TC's maximum tangential velocity ).…”

“…Considering the typical mesoscale model has a resolution of 1-3km and much of the critical venting occurs through turbulent eddies on the order 100m the importance of the models' ability to resolve or approximate turbulence is essential. Building on this fact, Rotunno et al (2009) published a study which examined the importance of turbulent transfer on TC modeling using a large-eddy simulation of a tropical cyclone using the Advanced Research WRF. Employing six nested grids with a horizontal resolution of 62m, they analyzed the impact of the resolution of turbulent eddies on maximum TC wind speed.…”

Towards Global Large Eddy Simulation: Super-Parameterization Revisited

“…For the latter, small-scale features of convective updrafts affect such aspects of cloud microphysics as cloud droplet activation and diffusional growth and the initiation and development of an ice field (e.g., as in Bryan et al 2003;Bryan and Morrison 2012). LES simulations of a tropical cyclone document similar resolution requirements for the realistic cyclone strength simulations (Rotunno et al 2009).…”

Vacillation Cycles in WRF Simulations of Hurricane Katrina

“…The non-hydrostatic WRF model The Weather Research and Forecast model (WRF) is an atmospheric simulation system, which has been previously applied to a wide range of meteorological problems, including tropical cyclone intensification (Rotunno et al 2009, Li and Pu 2008, Nolan et al2009, Fierro et al 2009). The Advanced Research WRF (ARW) dynamic core, chosen here, uses the non-hydrostatic perturbation form of the Euler equations written in flux-form (Skamarock 2008).…”