| 英文摘要 |
Under the trend of climate change, it is essential to establish a heat warning system to protect public health in Taipei City. This pilot project aims to establish a monitoring system for Wet Bulb Globe Temperature (WBGT, a heat stress indicator), to evaluate suitable WBGT classification for a heat warning system, and to assess forecast models in order to lay a solid scientific foundation for a heat warning system in Taipei City. There are four major tasks, including establishing a heat-stress monitoring system; conducting a pilot demonstration for heat warming system; providing a training course on the principles, operation and maintenance of the instruments; and carrying out regular maintenance of the monitoring system and WBGT database. All tasks have been completed successfully on schedule; they are briefly described below. (1) Four movable meteorological stations, for monitoring and calculating WBGT in a real-time fashion, were purchased and set up in four administrative districts of Taipei City to establish a heat-stress monitoring system. Inter-comparison was conducted to evaluate the performance of these instruments. In addition, WBGT estimations based on observations at Taipei station of Taiwan Central Weather Bureau of 2002- 2016 were analyzed to provide suitable WBGT classification. Real-time WBGT values were obtained from the established heat-stress monitoring system in Taipei City from June to October, 2017. (2) These observations could be presented in website in a real-time basis. Moreover, Auto Regressive Integrated Moving Average (ARIMA) was used to assess the trend of WBGT and forecast WBGT based on observations from the previous months. The prediction-observation differences of ARIMA were compared with those based on two numerical models. (3) A training course on the principles, operation and maintenance of the instruments was held on November 8, 2017. (4) Regular maintenance of the monitoring system every month and WBGT database checking every day were carried out from June to October in 2017.
In summary, this pilot project successfully demonstrated to use WBGT monitoring and ARIMA for forecasting to establish a heat warning system in Taipei City. Several suggestions and key points are listed below. (1) The number of days with daily maximum WBGT in four stations exceeding 36.8 to 37.4 were around 38 to 50 days, and the number of days with daily average WBGT exceeding 36.8 to 37.4 were only one day. We proposed that threshold of WBGT in the heat warming system was either 36.8 or 37.4. Two-stage warning system operated in Hong Kong could be adopted for Taipei City. (2) The weather conditions of high WBGT days in June to October 2017 were analyzed; it was usually associated with high temperature, high solar radiation, and low humidity. (3) We also found that the daily WBGT values from 9 to 11 am were statistically significantly higher than those from noon to 2pm, which was different from the typical understanding. In other words, using WBGT as a heat stress indicator, considering four meteorological factors (temperature, humidity, wind speed, and solar radiation) affecting heat stress experienced by human beings, could warm people to stay away from the actual high heat-stress locations and time-periods. If only temperature was used as an indicator, the heat-warning system may not provide accurate information to keep people away from the actual danger. (4) The observations of WBGT from five meteorological stations in Taipei city were 2-6% higher than those obtained at the Taipei station of Taiwan Central Weather Bureau about. It showed that WBGT estimated based on observations from traditional meteorological stations in lawns would underestimate 2-6% of the actual heat stress of the general public. (5) The maximum WBGT at five observation stations in Taipei City were 36.5-41.3, while the minimum were around 15.3-16.8; the average WBGT values were 28.5-29.2 with the standard deviation of 3.60-4.38. Thus, it showed that WBGT in Taipei City had high spatial variability. At present, four stations already covered the east, south, central and west part of Taipei City but north. It was found that WBGT exceeding 36.8 were mostly occurred in Xinyi and Nangang districts and WBGT values at Nangang were obviously the highest for most days, which needs to be further explored. (6) In the future, the meteorological data from the campus observatory in the Taipei elementary schools could be used to enhance the spatial coverage of WBGT observations. (7) Using ARIMA to forecast WBGT for the next 24 hours with the observations in the past 1-2 months had R2 (0.69~0.72, the best scenario) close to the best performance of numerical model forecast plus manual adjustment (0.78~0.79). For Mean Biased Error (MBE), ARIMA overestimated WBGT (0.15~0.29) while numerical models underestimated WBGT (-1.34~-2.18). For Mean Absolute Gross Error, the result of ARIMA (1.14~1.28) was slightly better than those from numerical models (1.79~2.54). Therefore, in the future, ARIMA could be applied in WBGT prediction in the heat warning system of Taipei City.
|