A Prototype Expert System to Forecast Typhoon Conditions at Cubi Point, Philippines

Abstract

A prototype expert system is designed to forecast the tropical cyclone related winds that may be used to set Conditions of Readiness (COR) at Cubi Point, Philippines. One set of rules modifies the storm position and strength forecasts to account for terrain interactions while crossing the Philippines. A second set estimates the local winds given the modified storm position and intensity. Tests using an independent storm set indicate the terrain-modified positions are comparable in accuracy to current Joint Typhoon Warning Center forecasts. However, the reduction of storm intensity due to terrain is underestimated and the westward translation of the storm is reduced too much. Finally, the conservative strategy of using worst-case wind gust estimates also contributes to an overprediction of the local winds and thus the COR. COR estimates are only 32% accurate, with a 95% capture rate for winds over 35 kt but a 70% false alarm rate due to overforecast of winds under 35 kt. More dynamic/statistical study appears to be required to refine the terrain- modification algorithm. Empirical rules form expert forecaster should be included in future systems. Theses.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1988
Accession Number
ADA201666

Entities

People

  • Bruce M. Hagaman

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Ground and Sea Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Artificial Intelligence
  • Computer Science
  • Computers
  • Cyclones
  • Databases
  • Detectors
  • Expert Systems
  • False Alarms
  • Meteorology
  • Prototypes
  • Research Facilities
  • Terrain
  • Tropical Cyclones
  • United States Naval Academy
  • Warning Systems

Fields of Study

  • Environmental science

Readers

  • Computational Modeling and Simulation
  • Geodesy
  • Ocean-Atmosphere Mesoscale Modeling, Data Assimilation, and Flux Boundary Layers