关键词: 最热月, 最冷月, 滑动平均法, 温度, 准确率

Abstract: The hottest and coldest month-long periods' temperatures are common indicators of Agro-climatical division. Commonly the hottest/coldest period is replaced by the full month of July/January directly, but this fixed full month is not an accurate reflection of the actual hottest /coldest period. The hottest/coldest period of the year changes in temperatures and starting to end dates every year. In this paper, the length of the month-long period was set to be 31 days, as the common hottest and coldest months, July and January, are 31 days long. By using the daily temperature data from 1951 to 2018, collected at Changde weather station in Hunan province, the temperature of the hottest/coldest 31-day period was calculated using moving average approaches.  The results showed that, (1)the hottest period spans from late June to early September, and the average hottest 31-day period was from mid-July to mid-August. Comparing the average hottest 31-day period and the full months of July and August with the actual highest 31-day period temperatures, the average temperature error was 0.5℃, 0.9℃, and 1.7℃ colder, respectively. (2) The coldest period spanned from early December to mid-March, and the average coldest 31-day period is from early January to early February. Compared to the actual coldest 31-day period, the temperatures of the average coldest 31-day period and the full months of January and February were 1.0℃, 1.1℃, and 2.9℃ warmer than the temperature of the actual month-long period, respectively. (3) With a temperature error within 1.0℃ considered to be acceptable, the average hottest 31-day period' s average temperature was 90% accurate in calculating the actual hottest 31-day period' s average temperature while using July' s temperature is only 61.2% accurate, which demonstrated that the average hottest 31-day period was more accurate than July. (4) With a temperature error within 2.0℃ and temperature accuracy above 80% considered to be acceptable, the effect during the average coldest 31-day was slightly better than January. In summary, the temperature error of the average 31-day hottest/coldest period was less than these of July/January, and the accuracy is the opposite. Therefore, it is recommended that when estimating the temperature of the hottest/coldest month-long period to not use the fixed full month of July/January but instead use the average hottest/coldest 31-day period while still taking into consideration in the adjustments from the actual temperature.

Key words: The hottest month-long period, The coldest month-long period, Moving average approach, Temperature, Accuracy