Evaluation of cumulus schemes of HWRF model in forecasting tropical cyclone characteristics, Gonu tropical cyclone case study

Sensitivity of numerical models in the prediction of Tropical Cyclone (TC) characteristics has been considered in numerous research studies. In this research, application of five cumulus schemes of HWRF (Hurricane Weather Research and Forecasting) model, including KF, SAS, BMJ, TiedTKE and SASAS has been examined during Tropical Cyclone Gonu (TCG) from 4 to 7 June 2007. The simulations have been conducted using three nests with 27, 9 and 3 km resolutions. To this aim, the performance of schemes in predicting TCG intensity using minimum surface pressure and maximum 10-m wind speed are analyzed. Following, their effect on forecasting the radius of maximum wind is evaluated. The parameters of lower-level divergence, upper-level convergence, potential temperature, potential vorticity, Convective Available Potential Energy (CAPE), wind vector (both horizontal and vertical components), wind shear, precipitation and radar reflectivity have been analyzed. The results of the simulations have been compared with the analysis data, IMD and TRMM observational data and routine atmospheric parameter measured at the Chabahar station. The comparison was done in different time of TCG lifetime. To examine the performance of HWRF cumulus schemes for track and intensity of the TCG, the whole life cycle of TCG was considered. To test the efficiency of HWRF cumulus schemes in predicting some dynamical and thermodynamical parameters, the time of maximum intensity of TCG (18 UTC on 4 June 2007) was focused on. To evaluate the functionality of HWRF cumulus schemes in the coastal area, the outputs were discussed in the last two days of the TCG life cycle. Results showed that based on the used configuration, none of the five cumulus schemes predicted the TCG reaching the southern coast of Iran. Moreover, neither the pressure decrease nore the maximum wind speed were predicted accurately at the time of maximum intensity of TCG. Until TCG intensity was more that category-3, neither minimum surface pressure trend and nor the maximum wind speed trend have been forecasted well. However, for the less intense conditions, two schemes of TiedTKE and SAS produced the nearest values. The performance of all five cumulus schemes, similarly predicted the radius of the maximum wind, except TiedTKE scheme that predicted the super cyclone 6 hours earlier. The analysed and simulated of the vertical cross sections of potential temperature and horizontal wind were similar, respectively. The simulated values of the vertical component of wind were considerably larger than those from the analysis data and were also closer to the TCG center. The maximum values of simulated CAPE were off the Oman coast compared to the analysis values. Only the simulations using SASAS cumulus schemes showed the strongest potential vorticity near the surface. The simulated updrafts and downdrafts were larger than those from the analysis data. The simulated values of the major updrafts and downdrafts were closer to the center of the TCG, comparing to those from the analysis data. The upper-level divergence patterns were seen in both simulations using all 5 cumulus schemes and also in the analysis data, while the lower-level convergences were not captured neither in the simulations nor in the analysis data. The maximum value of the simulated accumulated precipitation using all 5 cumulus schemes were 80 mm in a 6 hour interval, however, the observational value from the TRMM was 25 mm/h. The predicted radar reflectivity from the simulations were similar and the simulated maximum values were the same, but the expansions of the simulated maximum values were different. All cumulus schemes predicted the wind shear values less than the analytical values. At Chabahar station, the observational values of the 10-m wind speed, sea level pressure, and temperature have been compared to the simulated values using all 5 cumulus schemes, in the period of 6-7 Jun 2007. The statistical parameters of correlation, standard deviation and root mean square were used to identify the best cumulus scheme. The least error prediction was obtained using KF cumulus schemes to predict the 10-m wind, the TiedTKE cumulus scheme to simulate sea level pressure the observed, and SASAS cumulus schemes to produce temperature.