This study evaluates the integration of the Systematic Approach to Tropical Cyclone Forecasting Aid (SAFA) into the Joint Typhoon Warning Center (JTWC) forecast process. A computer based tropical cyclone (TC) forecast expert system, SAFA, aids the TC forecaster in identifying the current synoptic situation, determining numerical weather prediction (NWP) model errors, and formulating the official TC forecasts. The TC forecaster develops the forecasts by following a process that includes a thorough interrogation of numerical model fields and development of a Non-Selective Consensus (NCON) TC track prediction. SAFA also uses a simple algorithm to compare available NWP model TC track forecasts in an effort to identify potentially erroneous NWP model predictions. If a positive identification of systematic model errors is made as result of a thorough model field review, the TC forecaster dismisses the erroneous model predictions and creates a Selective Numerical Consensus (SCON) versus a NCON to guide in the final production of the official forecast.

A homogeneous comparison between the TC forecast track errors of JTWC, NCON, and SCON is presented for the 2000 western North Pacific tropical cyclone season. JTWC forecast track errors (FTE) were the lowest since its formation in 1959, with 24, 48, and 72 h FTE of 81 nm, 142 nm, and 209 nm respectively. Mean errors for NCON were 78 nm at 24 h, 130 nm at 48 h and 200 nm at 72 h, which only represent improvements of 1%, 7%, and 2% with respect to the JTWC official forecast. Results of the study suggest that the use of NCON and the application of SAFA have added more value to the JTWC official forecast track than any other forecast aid in JTWC's history.