关键词:
烟台市
短时强降水
机器学习
相关性
时空分布
摘要:
本文利用山东省烟台市2019年1月~2023年12月降水实况资料,对烟台市的短时强降水发生频次的空间分布、日变化、月变化以及年变化特征进行分析,并尝试利用CatBoost机器学习算法对短时强降水进行预测,结论如下:烟台市短时强降水发生频次及最强降水主要集中在西北以及东南部沿海地区,其中西北部地区的短时强降水,从发生频次以及最大强降水来说均较大,在烟台中部、北部及东部短时强降水发生频次及强度均较小;烟台市短时强降水总频次日变化、月变化及年变化特征均较为显著,日变化方面,傍晚至凌晨短时强降水发生频次最多,凌晨及午后为发生频次第二多的时段,而日出前后短时强降水发生频次较低;其月变化波动幅度较大,呈先增大后减少的变化特征,短时强降水较为集中在6~8月;其年变化波动幅度也较大,也呈先增大后减少的变化趋势,在2022年达到最多为499次,之后逐渐减少,2019年最少为72次,短时强降水发生频次较多的时段集中在2021~2023年。用CatBoost模型对烟台市强降水进行初步预测,模型基本能够很好地捕捉烟台市短时强降水特征,在测试集上表现出较好的预测能力,这表明该模型具有很好的泛化能力。Based on the actual precipitation data of Yantai City, Shandong Province from January 2019 to December 2023, this paper analyzes the spatial distribution, daily variation, monthly variation and annual variation characteristics of the frequency of short-term heavy precipitation in Yantai City, and tries to use CatBoost machine learning algorithm to predict short-term heavy precipitation. The conclusions are as follows: The frequency of short-term heavy precipitation and the strongest precipitation in Yantai City are mainly concentrated in the northwest and southeast coastal areas. The short-term heavy precipitation in the northwest is larger in terms of frequency and maximum heavy precipitation. The frequency and intensity of short-term heavy precipitation in the central, northern and eastern parts of Yantai are small. The daily variation, monthly variation and annual variation characteristics of the total frequency of short-term heavy precipitation in Yantai City are significant. In terms of daily variation, the frequency of short-term heavy precipitation from the evening to the early morning is the most frequent, and the early morning and afternoon are the second most frequent periods, while the frequency of short-term heavy precipitation before and after sunrise is low. The fluctuation range of monthly variation is large, which increases first and then decreases. The short-term heavy precipitation is more concentrated in June-August. The fluctuation range of its annual change is