In recent decades, due to the population growth and low precipitation, the overexploitation of ground water resources has become an important issue. To ensure a sustainable scheme for these resources, understanding the behavior of the aquifers is a key step. This study takes a numerical modeling approach to investigate the behavior of an unconfined aquifer in an arid area located in the east of Iran. A novel hybrid model is proposed that couples the numerical modeling to a data assimilation model to remove the uncertainty in the hydrodynamic parameters of the aquifer including the hydraulic conductivity coefficients and specific yields. The uncertainty that exists in these parameters results in unreliability of the head values acquired from the models. Meshless local Petrov-Galerkin (MLPG) is used as the numerical model, and particle filter (PF) is our data assimilation model. These models are implemented in the MATLAB software. We have calibrated and validated our PF-MLPG model by the observation head data from the piezometers. The RMSE in head values for our model and other commonly used numerical models in the literature including the finite difference method and MPLG are calculated as 0.166, 1.197 and 0.757 m, respectively. This fact shows the necessity of using this method in each aquifer.

近几十年来，由于人口增长和降水量少，地下水资源的过度开采已成为一个重要问题。为了确保这些资源的可持续计划，了解含水层的行为是关键步骤。本研究采用数值模拟方法来研究位于伊朗东部干旱地区的无承压含水层的行为。提出了一种新的混合模型，将数值建模与数据同化模型相结合，以消除含水层水动力参数的不确定性，包括水力传导系数和比产量。这些参数中存在的不确定性导致从模型中获取的水头值不可靠。采用无网格局部 Petrov-Galerkin (MLPG) 作为数值模型，粒子滤波器（PF）是我们的数据同化模型。这些模型在 MATLAB 软件中实现。我们已经通过压力计的观测头数据校准和验证了我们的 PF-MLPG 模型。我们的模型和文献中其他常用数值模型（包括有限差分法和 MPLG）的水头值 RMSE 分别计算为 0.166、1.197 和 0.757 m。这一事实表明在每个含水层中使用这种方法的必要性。分别。这一事实表明在每个含水层中使用这种方法的必要性。分别。这一事实表明在每个含水层中使用这种方法的必要性。