A detailed analysis of the performance of the U.S. Navy Nested Tropical Cyclone Model (NTCM) for western North Pacific tropical cyclones is made based on five storm-related factors: latitude, longitude, intensity, 12-h intensity change and size (radius of 15 m s−1 (30 kt) winds). The error measures used to assess the accuracy of the NTCM forecasts include: mean and median forecast errors, the systematic errors in the zonal and meridional directions, and the cross-track and along-track components relative to a climatology-persistence (CLIPER) track. These different measures provide insights into the different characteristics of the NTCM forecasts. Although the mean forecast errors are widely reported, they provide no indication of directionality. The zonal and meridional systematic errors provide additional information, but are difficult to interpret since both eastward and westward moving storms are included. Referencing the cross-track and along-track components to a standard forecast technique (CLIPER) provides directions information on the NTCM forecast errors that will be useful to the forecaster.
The analyses based on the latitude and longitude stratifications suggest that the NTCM predictions are most accurate for low-latitude storms and those in the western region of the western North Pacific. However, the model does not perform very well for storms north of 17° or east of 129°E. The NTCM provides better guidance when the observed intensity is close to that of the bogus vortex inserted in the NTCM. For cyclones with a radius of 15 m s−1 winds ≥ 389 km, the NTCM forecasts do not have good skill relative to those from the CLIPER scheme. Intensity changes in the past 12 h do not appear to affect to affect significantly the performance of the NTCM. Most of the results from these analyses may be attributed to a slow bias in the NTCM. Other potential sources of error include the fixed intensity and size of the bogus vortex and the domain size of the nested grid.
The study is intended as a prototype for evaluating an objective track forecast aid based on storm-related factors. Results from such an evaluation can be used not only by the forecasters but also in future modifications of the forecast aid. The results of this study indicate strongly a need to improve specification of the initial conditions in the NTCM, and especially to introduce a bogus storm that is more representative of the storm.