
研究興趣
目前研究主題在機器學習於環境微型感測器結合應用,利用微型感測器之大數據結合機器學習演算法,進行污染熱點分析、污染趨勢判斷及感測器數據現地修正。此外,建立熱危害評估模式,評估都市與社區之民眾受熱浪之影響及預警分析,是另外一個主要研究主題。
代表著作
研究成果介紹
Wet-bulb globe temperature (WBGT) is a well-known heat-stress indicator, since it is associated with heat-related health impacts. The WBGT index is calculated as a weighted average of the dry temperature (Ta), natural wet-bulb temperature (Tw), and globe temperature (Tg), measured directly from heat stress monitors such as the QuesTempTM monitors (TSI Incorporated, Minnesota, USA). In my previous study, WBGT at ground level was found to be on average 0.2–2.9 °C higher than at higher levels (above a height of 10–15 m). However, these monitoring instruments for Tg and Tw are usually not weather-resistant, and can be difficult to maintain. An alternative method was applied to calculate the WBGT index based on numerical models. A WBGT model, which originated from Liljegren et al. (2008), is used to estimate Tg and Tw based on several theoretical algorithms using the inputs of the dry-bulb temperature, relative humidity, wind speed, and solar radiation. I developed a Visual basic package to make the iterative calculation of these theoretical algorithms much easier, and then rewrote the codes in a Python package which can be embedded on a website for an expanding application base. However, the impact of the solar zenith angle was miscalculated in the original equations, which led to extreme overestimation of Tw in the early morning and in the late afternoon at sunset. Therefore, I corrected this in the original codes, and adjusted Tg be in concordance with ISO 7243. This modified WBGT program has been validated according to the observations of the WBGT instrument, and was delivered to the Central Weather Bureau of Taiwan for computational estimation of the heat stress index, directly in relation to public health.