European Center for Medium Range Weather Forecasts (ECMWF) global analyses are used to develop an Eastern Pacific Intensity Change (EPIC) model . The model forecasts tropical cyclone intensity changes at 12 hour intervals, out to 72 hours. All named tropical cyclones during the years 1989 to 1995 are used to develop the model. Past statistical intensity models have primarily utilized climatology and persistence to measure tropical cyclone intensity changes. The research presented here uses climatology, persistence and synoptic predictors to develop an operational forecast tool for the Eastern North Pacific basin.
The synoptic variables examined in this study consist of shear, thermal and momentum variables. Absolute shear and the individual components of shear are investigated. Thermal variables include the difference in temperature at selected levels, equivalent potential temperature and a measure of potential intensity change derived from weekly sea surface temperatures. Planetary and relative eddy angular momentum flux, along with relative angular momentum, define the momentum variables used here. Multiple linear regression analysis is used to detect which variables best explain tropical cyclone intensity change in the Eastern North Pacific.
The EPIC model forecasts are compared to SHIFOR (climatology and persistence model) and official forecasts in the Eastern North Pacific basin for the 1996 and 1997 seasons. Mean forecast errors for the two year period show that the EPIC model overforecast intensity change at all time periods while SHIFOR and official forecasts overforecast intensity change in 1996 and underforecast in 1997. Mean absolute errors for the 1996 and 1997 period revealed that EPIC provides better intensity guidance than SHIFOR, Forecasts of intense tropical cyclones (Linda and Guillermo) indicate that EPIC provides the good predictions during the intensifying and weakening stages. This study suggests that the inclusion of synoptic predictors into a statistical hurricane intensity prediction scheme improves statistical intensity change forecasts