Forecasts of North Pacific Maritime Cyclones with the Navy Operational Global Atmospheric Prediction System

Abstract

Seventy-two-hour forecasts of sea level cyclones from the Navy Operational Global Atmospheric Prediction System are examined. Cyclones that formed over the North Pacific region of maximum cyclogenesis frequency are included for study. The analysis is oriented to assist the forecaster in evaluating the numerical model guidance by emphasizing verification of operationally oriented factors (i.e., cyclogenesis, explosive deepening). Initially, systematic errors in forecast intensities and positions are identified. Maximum underforecasting errors (forecast central pressure higher than actual central pressure) occur over the central North Pacific region of climatological maximum cyclone deepening. Maximum overforecasting errors (forecast central pressure lower than the actual central pressure) occur over the region of climatological cyclone dissipation. Maximum position errors also occur over the central North Pacific region of climatological maximum deepening. These systematic error distributions indicate that there are diagnostic relationships between forecast performance, the cyclone track type, and whether the cyclone is deepening or filling at the forecast verification time. The forecast intensity and position errors are stratified based on the 72-h forecast intensity change, which is one possible measure of forecast accuracy that uses information known at the initial time of the forecast rather than the verifying time. Three classes of intensity change are identified as deepening, filling, and mixed deepening and filling. The systematic intensity errors mainly comprise instances when a 72-h deepening profile was not forecast and a deepening or mixed deepening-filling profile actually occurred. When the category of intensity change is correctly forecast, cyclones forecast to follow a western Pacific track tend to be overforecast, while those forecast to follow a central Pacific track tend to be underforecast. It is hypothesized that one reason for these differences may be due to the relative importance of adiabatic versus diabatic processes involved in the development of cyclones following each track type. Furthermore, central Pacific cyclones become more removed from available initializing data on the Asian continent. Position errors are more sensitive to the forecast track type rather than the forecast central-pressure profile. Model tendencies based on the forecast intensity change and track type are presented to aid the users of the numerical guidance recognize instances when the forecast performance may be exceptionally high or low.