泰安市蒸发量变化趋势分析与基于神经网络的预测  

Analysis of Evaporation Trend Changes in Taian City and Prediction Based on Neural Networks

作　　者：于小鸽[1] 王世超 李岩 钱丽丽 YU Xiao-ge;WANG Shi-chao;LI Yan;QIAN Li-li(Shandong University of Science and Technology,College of Resources,Taian 271000,China;Shandong University of Science and Technology,College of Earth Science and Engineering,Qingdao 266500,China;Tai'an Hydrological Center,Tai'an 271000,China)

机构地区：[1]山东科技大学资源学院,泰安271000 [2]山东科技大学地球科学与工程学院,青岛266500 [3]泰安市水文中心,泰安271000

出　　处：《科学技术与工程》2024年第10期3984-3996,共13页Science Technology and Engineering

基　　金：国家自然科学基金(42002282)。

摘　　要：蒸发量是水文特征里的一个重要指标,为科学准确地分析及预测泰安市蒸发量的特点和走势,利用泰安市黄前水库、东周水库、大汶口和戴村坝4个代表性水文观测站1985—2021年的调查数据,通过Mann-Kendall检验法、滑动t检验法检测其突变特征后,使用R/S分析法预测未来蒸发量变化趋势。使用泰安站2005—2022年蒸发量日值观测数据,通过Neural-Prophet算法耦合Optuna算法建模进行蒸发量的预测,并与其他预测模型的评价指标做出比较。结果表明:泰安市年及各季的蒸发量都呈现出明显的减少趋势,且在今后的一段时期内,大部分区域都将延续这样的发展态势。模型给出的预测数据准确率很高,符合要求,可以利用到日常生产及科研指导中,为蒸发量的预测提供了一种新途径。Evaporation is considered an essential indicator within hydrological characteristics.To scientifically and accurately analyze and predict the characteristics and trends of evaporation in Taian City,data from four representative hydrological observation stations in Tai'an City—Huangqian Reservoir,Dongzhou Reservoir,Dawenkou,and Daicun Dam—from 1985 to 2021 were utilized.The abrupt change characteristics were analyzed through the Mann-Kendall test and sliding t-test,while the future evaporation trend was forecasted using the R/S analysis method.Daily evaporation observation data from the“Tai'an”station from 2005 to 2022 were employed,and a model coupling the NeuralProphet algorithm with the Optuna algorithm was developed for evaporation prediction.This models predictive performance has been compared against other forecasting modelsevaluation metrics.The findings indicate that the annual and seasonal evaporation rates in Tai'an City demonstrate a clear decreasing trend.For the foreseeable future,this trend is expected to continue in most areas.The forecasted data provided by the model exhibit high accuracy and meet the set standards,proving valuable for daily operations and scientific research guidance.This study offers a novel approach to predicting evaporation.

关 键 词：MK突变检验 滑动t检验 R/S分析法 NeuralProphet算法 Optuna算法 

分 类 号：P332.2[天文地球—水文科学]