ABSTRACT In order to understand and adequately manage hydrological stress, it is necessary to simulate groundwater levels accurately. In this research, gene expression programming (GEP) and M5 model tree (M5) are used to simulate monthly groundwater levels. The models are combined with wavelet transform to produce two hybrid models: wavelet gene expression programming (WGEP) and wavelet M5 model tree (WM5). For the simulation, groundwater level, temperature and precipitation values from three observation wells and one meteorological station, located in Iran, are used. The results indicate that the hybrid models, WGEP and WM5, lead to a better performance than the simple models, GEP and M5. The performance of the two hybrid models is similar. It is also observed that selecting a suitable time lag for inputs plays an important role in the accuracy of the simple models. The selection of a suitable decomposition level strongly affects the accuracy of hybrid models.

中文翻译：

---

基于基因表达编程和M5模型树结合小波变换的地下水位模拟

摘要 为了理解和充分管理水文压力，有必要准确模拟地下水位。在这项研究中，基因表达编程（GEP）和 M5 模型树（M5）被用来模拟每月的地下水位。这些模型与小波变换相结合，产生两种混合模型：小波基因表达编程（WGEP）和小波 M5 模型树（WM5）。在模拟中，使用了位于伊朗的三个观测井和一个气象站的地下水位、温度和降水值。结果表明，混合模型 WGEP 和 WM5 的性能优于简单模型 GEP 和 M5。两种混合模型的性能相似。还观察到，为输入选择合适的时间延迟对简单模型的准确性起着重要作用。选择合适的分解级别会强烈影响混合模型的准确性。