The Statistical Properties of Temperatures and Their Long-term Trends in Taiwan

• Main Authors: Hsu, Shao-Ching, 徐紹青
• Other Authors: Weng, Shu-Ping
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/xse5hf

碩士 === 國立臺灣師範大學 === 地理學系 === 106 === This study uses the long-term surface temperature records of eight stations: Taipei, Taichung, Penghu, Tainan, Hengchun, Taitung, Hualien and Alishan, including daily average temperature, maximum temperature and minimum temperature data archived at Central Weather Bureau, to analyze the following five weather scenarios: rainy days, clear days, days before rain, days after rain, and normal day (whether or not there is rainfall). Their first 4 statistical moments, namely the mean, standard deviation, skewness and kurtosis are calculated and the related statistical characteristics during the base period (1961-1990) and the corresponding long-term (1897-2014) linear trend are explored. After analyzing the statistical moments of the base period for all kinds of combinations of the above situations, it is found that, although the mean and standard deviation represent the seasonal average state and its amplitude, skewness and kurtosis are calculated using the standard deviation of 3 or 4 times the power of the former signal can be amplified, and provide us the opportunity to examine the combination of various types of changes in the statistical characteristics in detail. For example, during the wintertime, there is a clear cooling effect at the windward stations due to the northeast monsoon. Although analyzing the mean can obtain its displacement change, analyzing the skewness is more able to show its tendency towards the low temperature statistical characteristics (i.e. right-sided probability density function distribution). Similarly, during the autumn season, the Tainan station, in the days after the rain, is observed to have the higher maximum temperature (i.e. the left-sided probability density function) and the positive kurtosis, suggesting that the scrubbing effect after the rainy day probably enhances the insolation, and highlights the autumn tiger phenomenon. Consistent with the previous studies, this study also finds that the effect of global warming on the surface air temperature is mainly reflected in increasing the night temperature. Analyzes of the background stations (Penghu, Hengchun and Alishan) found that the increased moisture content enhances the downward longwave radiation under a warming world, and the lower degree of the night temperatures can make the moisture saturated and the resulting increased cloudiness inhibits the upward long-wave radiation. Therefore, the nighttime temperature appears to have a higher degree of growing amplitude. The study also finds that aerosols and pollutants due to human activities may increase the nighttime warming instead of decreasing it. After analyzing the long-term trend of the minimum temperature at Tainan and Taipei stations, it is found that there is less winter cloud in Tainan, but the minimum temperature has the same warming trend as Taipei which is normally situated in an overcast environment in winters. The autumn analysis at Tainan station further finds that, on the first days after rain, the skewness decreased but the kurtosis increased. It thus suggests that there is a long-term phenomenon that the temperature of the autumn night falls above the average and becomes more concentrated. However, most of the other stations are not concentrated in high temperatures. Moreover, the comprehensive performance of temperature at most of the stations shows a turning point in the long-term trend, which generally occurs in the period when urbanization and industrial development were flourishing around the 1970s. Therefore, this study argues that the suspended particles and pollutants due to human activities have impacts on the long-term trend of regional climate. Under the global warming, the changes in the long-term trend at different seasons are not consistent. Based on the wintertime trend analysis for the nighttime temperature at Taipei station during the rainy and no-rain days, it is found that the nighttime temperature of the no-rain days is smaller (larger) than that of the rainy days before (after) the 1970s, suggesting the changing interactions between the rainfall processes and warming. In addition, the analysis of the long-term trend of the minimum temperature in September finds that the integrated values and trends of temperatures are closer to those in July and August as compared with those in June, suggesting the extension of the summer season. Based on the above analysis of the temperature, this study suggests that the seasonal climate in Taiwan should be divided into Spring season: February, March and April, Mei-Yu season: May-June, Summer season: July, August and September, Autumn season: October, November and Winter season: December and January. The above seasonal division based on the temperature analysis in this study is consistent with the timing of seasonal rainfall in Taiwan.