Abstract Predicting stable groundwater seepage using known data is an important task in groundwater dynamics. This is crucial for ensuring the rational utilization of groundwater resources and effectively controlling groundwater pollution. However, traditional finite difference methods and analytical solutions have significant limitations in dealing with complex groundwater flow problems, such as poor applicability and low computational efficiency under irregular boundary conditions. In recent years, although methods based on radial basis function (RBF) have improved in accuracy, they still face the problem of insufficient flexibility in high-dimensional space. This article proposes a new method for predicting stable groundwater seepage, using the multiple quadratic curve (MQ) numerical method to address this challenge. We first use RBF interpolation to decompose the groundwater flow field, and then solve the corresponding linear equation system to obtain a numerical solution for stable groundwater seepage. This method abandons the traditional construction of grids and elements, provides greater flexibility in high-dimensional space, and adopts a node based approximation method to ensure computational accuracy. The experimental results show that this method performs well in simulating groundwater flow and changes, with relative errors as low as a specific value. Compared with higher precision radial basis function methods, this new method improves the accuracy by two orders of magnitude, providing a new solution for accurate prediction of groundwater flow.