Performance of MPAS-A and WRF in predicting and simulating western North Pacific tropical cyclone tracks and intensities

Abstract

Performances of the Model for Prediction Across Scales-Atmosphere (MPAS-A) in predicting and the Weather Research and Forecasting (WRF) model in simulating western North Pacific (WNP) tropical cyclone (TC) tracks and intensities have been compared. Parallel simulations of the same historical storms that made landfall over southern China, namely, TCs Hope (1979), Gordon (1989), Koryn (1993), Imbudo (2003), Dujuan (2003), Molave (2009), Hato (2017) and Mangkhut (2018), were carried out using WRF and MPAS-A, with initial conditions (and, for WRF, lateral boundary conditions as well) taken from ERA-interim. For MPAS-A, the model was integrated using a standard 60-to-3-km variable-resolution global grid mesh and also on 160-to-2-km grids customized to cover the TC tracks with the highest resolution mesh. The WRF model was integrated using a 15-km/3-km nested domain. No TC bogus scheme was applied when initializing the MPAS-A and WRF simulations. It was found that while TC tracks were reasonably captured by the two models configured variously, the storm intensities were underestimated in general. Given MPAS-A runs were initial value predictions whereas WRF runs were dynamically downscaled from ERA-interim, the finding that MPAS-A has comparable (or slightly better) performance as (than) WRF is noteworthy. To further examine the sensitivity of the MPAS-A TC forecasts to the initial data, additional experiments were carried out for TCs Molave and Hope using ERA5 reanalysis as initial conditions. The ERA5 initialized runs showed significant (slight) improvement in intensity (track) evolution, suggesting that the underestimated TC intensity is likely related to inferior representation of storms in the ERA-interim initial fields. Furthermore, additional runs using another customized 60-to-2-km mesh showed a reasonable improvement in capturing the TC tracks, suggesting that the track forecast accuracy of MPAS-A in TC can be sensitive to the grid resolution in the coarsest part of the variable-resolution mesh used.

Minor impacts of major volcanic eruptions on hurricanes in dynamically-downscaled last millennium simulations

“…There is agreement in the location of TC tracks for both comparisons, however the downscaled ERA-I underestimates TC intensity. This underestimation is likely related to the downscaling resolution of 30 km and the representation of storms in ERA-I (Lui et al 2021). There is also a notable lack of TC tracks in the Gulf of Mexico for the downscaled ERA-I compared to the IBTrACS (Fig.…”

Key ingredients in regional climate modelling for improving the representation of typhoon tracks and intensities

“…This suggests that the TS setting significantly improves the simulation results. A previous study shows an MSLP of around 950 hPa at the peak time, resulting in a bias exceeding 20 hPa when compared to the best track data from HKO (Lui et al, 2021).…”

“…Compared with CMA, the average bias is around 0-0.6 • , except for two experiments of Hagupit (TH_CuOFF_(U V )_ST, W6_CuOFF_(PT + U V )_ST), inheriting a bias of 0.6-0.8 • . Lui et al (2021) show the WRF performance of Typhoon Hato in mean track bias within 1.5 • compared with Hong Kong Observatory's (HKO) best track. Overall, the simulated tracks of simulations nudging horizontal wind above 500 hPa, which may reasonably capture the larger-scale circulation patterns, are close to the best tracks.…”

Numerical Weather Prediction at 200 m Local Resolution Based on an Unstructured Grid Global Model

“…Its dynamical core solves a set of fully compressible non‐hydrostatic equations using the finite‐volume method, discretized horizontally on a C‐grid staggered unstructured spherical centroidal Voronoi tessellation (SCVT) (Ringler et al., 2010; Thuburn et al., 2009) and vertically in a geometric‐height hybrid terrain‐following coordinate (Klemp, 2011). MPAS‐A has been used for applications such as medium‐range convection‐permitting ensemble forecasts (Schwartz, 2019), study of spatial and temporal wind speed variability during open cellular convection (Imberger et al., 2021), diurnal variation of Mei‐yu rainfall over East China (Xu et al., 2021), investigation of western North Pacific tropical cyclone tracks and intensities (Lui et al., 2021), etc. Modifying the source code of MPAS‐A, we extend the model to cover usages where one would conventionally use WRF with domain nesting for high‐resolution simulations around a region of interest.…”