A grid sensitivity study of the Weather Research and Forecasting (WRF) model for simulating surface winds over the small island state of Fiji

A grid sensitivity study of the Weather Research and Forecasting (WRF) model was conducted for simulating the surface winds and the diurnal cycle over the small island developing state (SIDS) of Fiji. Two different sets of grid resolutions: 20 km – 4 km – 1.33 km and 15 km – 5 km – 1 km are used with the two-way nested approach and 1deg gridded input data from the National Centers for Environmental Prediction - (Final) Operational Global Analysis (NCEP-FNL) as initial and boundary conditions. Simulations are performed for an austral summer (January 2017) and austral winter (July 2017) month using the dynamical-downscaling approach and the tropical suite of the physics parameterization scheme to simulate the surface winds and the diurnal cycle of wind speed for both grid set-ups. The results revealed that the WRF model is able to capture the surface winds and the diurnal cycle of wind speed more accurately for the higher grid resolution of 1 km x 1 km in comparison with the 1.33 km x 1.33 km, indicating that the topographical representation is better in the higher grid resolution. For January (July) the bias reduced from 0.85 m/s (1.04 m/s) to 0.50 m/s (0.83 m/s), which is a reduction of 20.78 % (15.44 %). The Root Mean Square Error (RMSE) reduced from 2.08 m/s (2.55 m/s) to 1.96 m/s (2.50 m/s). The Standard Deviation Error (STDE) is 1.65 m/s (1.92 m/s), which is almost the same for both grids. The diurnal cycle correlation also improved from 0.69 (0.60) to 0.81 (0.75).