dc.description.abstract | The current study assessed the ability of the WRF-ARW model to forecast tropical cyclones over
the Southwest Indian Ocean. The characteristics of tropical cyclones Jobo (2021), Kenneth
(2019), and Idai (2019) were simulated using WRF-ARW numerical model, and 6 hourly near
real-time Global Forecast System (GFS) data were used as initial and boundary
conditions. Kain-Fritsch and newly modified Tiedtke schemes were used to investigate the
impacts of cumulus parameterizations on the forecasted tropical cyclone's characteristics. The
microphysics sensitivity experiments used the complex and simple WRF-ARW single-moment
microphysics schemes (WRF Single Moment class 6 and WRF Single Moment 3). WRF model
forecasted well the movement of the tropical cyclones to the landfall location, the tracks were
closer to the observations with a lead time of up to 4 days. The average Direct Positional Error
(DPE) was found to be 58 ± 6, 86 ±17.2, 146±43.7 and 240 ±145 km at 24, 48, 72, and 96 hours
lead forecasts time, respectively. The values of Along Track Error (ATE) and Cross Track Error
(CTE) showed that the model had a southward and a slow bias near the land. WRF forecasted
central pressure (maximum wind speed) was higher (lower) than the observations. 24 hourly
accumulated precipitations from the model were compared against observations on the landfall
day. The result showed that the cumulus convection and microphysics of the model affect the
simulated tracks, intensity, and precipitation of tropical cyclones. Kain Fritsch schemes provided
the central pressure and wind evolution tendencies closer to the observations than the Modified
Tiedtke scheme. The combination of the WSM6 and KF schemes gave the best estimates of the
intensities of the cyclones closer to the observations compared to the MT scheme. The Tiedtke
scheme produced good forecasts of the tracks of Tropical Cyclone Idai. The cyclones modelled
using KF schemes were slower than in the observation causing the forecast to lag the observed
location. The study showed that the WRF model had a large slowdown bias closer to the land,
and the intensity during intensification and decay peaks were poorly simulated this requires
further investigations. Overall, the study concludes that the WRF-ARW model simulated well
the trajectory of tropical cyclones over the Southwest Indian Ocean for up to 4 days. The
characteristics of tropical cyclones forecasted with the WRF-ARW model were dependent on the
microphysics and cumulus convection parameterizations. Studies on other characteristics of
tropical cyclones should be carried out to find the best parameterizations to be used in
operational forecasting. | en_US |