<p>In recent years, structural equation modeling (SEM) has been widely applied in fields such as education, psychology, and environmental science. However, most studies overlook the spatial dependencies within the data, and there is limited research on SEM for spatial data. Spatial random effects (SRE) models, which flexibly capture spatial variation through a series of spatial basis functions (e.g., multiresolution wavelet basis functions), have become a powerful tool for spatial data analysis. This study extends the traditional SEM by incorporating SRE, resulting in a spatial random effects structural equation model (SRE-SEM) for modeling complex environmental spatial data. The model is fitted using a hybrid variational Bayes algorithm, with some model parameters estimated via the standard mean-field variational Bayes approach. To address the intractable posterior and the dimensionality of latent variables, the Metropolis-Hastings algorithm is employed to sample from the exact conditional posterior distribution of the latent variables. Additionally, a fixed-form variational Bayes approach is used to estimate the matrix on which the spatial covariance matrix depends. Simulation studies and case analyses demonstrate that the proposed model effectively captures the structure of spatial data, while the introduced estimation algorithms significantly improve computational efficiency. This study provides a robust and efficient framework for spatial data analysis, offering a promising solution for modeling complex environmental and socio-economic phenomena.</p>