Analysis of the displacement error of the WRF–ARW model in predicting tropical cyclone tracks over the Philippines

The direct positional errors (DPEs) in tropical cyclone (TC) track predictions over the Philippine domain of the Weather Research Forecasting (WRF) model's Advance Research WRF (ARW) was studied. The 3 year dataset (i.e. 2012-2014) was categorized (i.e. tropical storm (TS), severe tropical storm (STS), typhoon (TY)) and the marginal distributions of the DPEs over the short-range (i.e. 84 h) prognosis of the model were analysed. The decreasing magnitudes and spread of DPEs at increasing TC intensity were determined. The median DPEs for the model initialization were 70 km (TS), 40 km (STS), 30 km (TY) which progressively increased by 50-90% (all subsets), 130-220% (TS and TY), 430% (TY) for the 1, 2 and 3 day forecasts, respectively. The variations in timing, frequency and length of significant DPE increments for successive lead time levels, as well as the daily significant DPE increments from the model initialization on the TC subsets were also studied. The formulations of the forecast cone of uncertainty (CONU) were likewise incorporated.