基于粒子群算法与最小二乘支持向量机的ET0模拟

以月最高气温、月最低气温、月平均气温、平均风速、日照时数以及相对湿度6个气象因子的不同组合作为输入数据,以FAO Penman-Monteith公式计算结果作为标准值,构建基于粒子群优化算法与最小二乘支持向量机的ET_0预测模型(PSO-LSSVM)。选取新疆额尔齐斯河流域哈巴河气象站1986—2013年的气象数据进行模型训练与预测,并与其他常用ET_0计算公式进行对比研究。结果表明,PSO-LSSVM模型能够很好地反映ET_0同各气象因子之间的非线性关系,其中气温条件是影响ET_0模拟精度最重要的因素,同时随着气象因子输入的减少PSO-LSSVM模型模拟精度有所下降;当分别基于辐射条件、温度条件计算时,PSO-LSSVM模型模拟结果较Priestley-Taylor公式、Hargreaves-Samani公式计算结果要优。基于多因子量化指标的ET_0预测模型实现了精度和实用性的统一,可为缺资料地区ET_0研究预报提供科学参考。

Simulation of ET0 based on particle swarm optimization and least squares support vector machine

Different combinations of meteorological factors, including monthly maximum temperature, monthly minimum temperature, monthly average temperature, average wind speed, sunshine duration, and relative humidity were used as the input data, the results calculated by the FAO Penman-Monteith equation were used as the calibration values, and a PSO-LSSVM model based on the least squares support vector machine(LSSVM)and particle swarm optimization(PSO)was established for prediction of ET₀. Meteorological data from the Habahe Meteorological Station in the Irtysh River Basin over the period from 1986 to 2013 were used to train and test the model, and the results calculated by the PSO-LSSVM model were compared with those calculated by other commonly used ET₀ calculation formulas. The results show that the PSO-LSSVM model can reflect the non-linear relationships between ET₀ and the meteorological factors well, and that temperature is the most important factor that influences the accuracy of simulation. However, as the number of meteorological factors decreases, the accuracy of simulation will decrease. When the calculation is based on radiation and temperature conditions, the PSO-LSSVM model has higher accuracy than the Priestley-Taylor and Hargreaves-Samani equations. The PSO-LSSVM model, with multi-factor quantitative indicators, is both precise and practical, providing scientific references for ET₀ study in areas that lack data.