摘要
参考作物蒸散发(ET_0)是估算作物需水量的基础指标,对区域农田用水灌溉、管理和调配具有重要指导意义。针对ET_0估算模拟精度不高的问题,基于最小二乘支持向量机(LSSVM)模型,利用由耦合模拟退火算法与单纯形法组合得到的CSA-SIM算法,进行模型参数的自适应调节,以此提出CSA-SIM-LSSVM模拟模型。以华北平原边界上4个典型站点1970年1月至2019年11月的每日气象资料为基础,以Penman-Monteith公式计算所得ET_0为标准,结合必要的影响因子筛选和时间序列降噪等技术和农时知识,提出模型进行日尺度和月尺度的ET_0估算模拟。结果表明:在控制变量的基础上,所提出的CSA-SIM-LSSVM模型整体表现最好,4个站点平均的日ET_0模拟的RMSE为0.045mm/d,月ET_0模拟对于双峰段预测较好。相较而言,CSA-SIM-LSSVM模型可作为一种较为可信的ET_0估算途径。
Reference crop evapotranspiration(ET 0)is the basic index to estimate crop water requirements,which is of great significance for irrigation,management and allocation of regional farmland water.In view of the low simulation accuracy of ET 0 estimation,based on the least squares support vector machine(LSSVM)model,a CSA-SIM algorithm is obtained by coupling simulated annealing and simplex method.The CSA-SIM-LSSVM simulation model is proposed by adaptive adjustment of model parameters.Based on the daily meteorological data of four typical stations on the boundary of North China Plain from January 1970 to November 2019,the ET 0 value calculated by Penman-Monteith formula was taken as the standard.Combined with the necessary influence factor screening and time series noise reduction technology and agricultural time knowledge,the model is proposed to simulate the daily and monthly ET 0 estimation.The results are as follows.Based on the control variables,the proposed CSA-SIM-LSSVM model performs best overall.The average RMSE of daily ET 0 simulation of the four stations is 0.045 mm/d,and the monthly ET 0 simulation is better for bimodal prediction.In contrast,CSA-SIM-LSSVM model can be used as a more reliable ET 0 estimation approach.
作者
王文川
赵钊
张磊
WANG Wenchuan;ZHAO Zhao;ZHANG Lei(College of Water Resources,North China University of Water Resources and Electric Power,Zhengzhou 450046,China;School of Water Conservancy,North China University of Water Resources and Electric Power,Zhengzhou 450046,China)
出处
《华北水利水电大学学报（自然科学版）》
北大核心
2021年第2期76-83,共8页
Journal of North China University of Water Resources and Electric Power：Natural Science Edition
基金
国家自然科学基金项目(51909092,51509088)
河南省科技攻关项目(202102310259,202102310588)
河南省高等学校重点科研项目(21A570003)。
关键词
参考作物蒸散发
耦合模拟退火
单纯形
最小二乘支持向量机
reference crop evapotranspiration
coupled simulated annealing
simplex
least squares support vector machine
