Abstract -- To improve the capability of typhoon forecast and to reduce the damages due to the destructive typhoons are the top ranked research topics of the Disaster Mitigation Research Program. The “Technical Development Studies on the Observations and Forecasts of Heavy Rains and Strong Winds Associating with Typhoons” is one of the integrated researches under the Program. This study is one of the individual studies under the integrated research. The objective of the study is to strengthen the capability of predicting the strong winds over Taiwan area when a typhoon is nearby. The tasks we completed in this year include: 1. Collected and built a complete database of the winds observed at 24 stations of Taiwan. 2. Implemented the forecasting methods and completed the wind speeds forecast verifications. 3. Provided a reference document for typhoon wind speed forecasts over those stations. From the analysis we obtained the climate mean of the distribution of wind speed over Taiwan when typhoons were nearby. From the forecast evaluations we also obtained the scores of each forecast method. Those scores are important information for selecting forecast guidance operational.
This study also found that the winds over Taiwan area when a typhoon is nearby are very complicated, and the distributions are linked to the topography. Stronger wind speeds at stations near the shore or in the surrounding small islands. The wind speeds are slowed down for stations further inland. For a station, usually the stronger wind occurred when the typhoon center moved into two particular areas. One area locates near the station. When a typhoon moved into the area, the storm’s stronger cyclonical circulation was blowing directly to or covered the station. The location of the second area depends on the location of the station.
The forecast evaluation scores reveal that the wind speed forecast based on the climatology average is capable of reflecting the trend of the wind speed change. The values of the forecast are not accurate enough for operational application. Both the Persistence Method and the Deviation Persistence Method improve the forecast of the lag time smaller than 6 hours. However, the improvement is not very useful as the lag time is too short for being real application. For meeting the requirement of the real application, a Bias Correction Method was developed and was shown that the method can produce useful predictions for the maximum wind speeds within the next 24-hour.