A systematic approach to tropical cyclone (TC) track forecasting in the western North Pacific is being developed for use by the Joint Typhoon Warning Center (JTWC) to improve the quantitative accuracy and qualitative utility of official TC track forecasts. The four key components required by the approach are: (i) a meteorological knowledge base of conceptual models that provide dynamically-based explanations for observed TC motion; (ii) a scenario-specific forecast model traits knowledge base documenting how the objective TC track forecast models available to the forecaster tend to perform in various situations; (iii) a procedural framework for employing the knowledge bases to methodically formulate the best possible official track forecast from the available objective guidance; and (iv) a vehicle to enable the forecaster to overcome the information management challenges to accomplishing
(iii) under the time, manpower, and experience constraints inherent in operational forecasting at JTWC. To realize requirement (iv), a Systematic Approach Expert System (SAES) application is being developed to run in conjunction with the Automated Tropical Cyclone Forecasting (ATCF) system currently operated by JTWC on Hewlett-Packard UNIX workstations. This talk provides a progress report on the development of the SAES.
The SAES is being designed to help the forecaster accurately accomplish three key tasks: (i) classification of the meteorological scenario that explains the observed motion of the TC at the current warning time; (ii) classification of the meteorological scenario that explains the track forecast of each numerical TC forecast model for which the forecaster has access to the necessary forecast fields; and an (iii) selection of one of three possible qualitative accuracy assignments to the meteorological scenario and associated track forecast by each numerical model using the model traits knowledge base. The three model forecast accuracy assessments may be: usable without significant adjustment; usable but requiring significant adjustment due to expected model biases; or unusable. The above accuracy assignments will then enable the forecaster to formulate an official track forecast via a "selective" consensus approach that heavily and equally weighs the model tracks deemed to be usable, gives less weight to the biased tracks, and ignores the tracks deemed to be unusable.
The SAES is being designed to be highly proactive in helping the forecaster to accomplish the above tasks, and includes such feature as simultaneous animations of analyzed and predicted fields from two models at any level selected by the forecaster, and objective analysis of the spatial and temporal progression of numerical model track forecast positions to alert the forecaster to potential changes in model forecast accuracy