Multiple factors influence the incidence of aeolian dust, including dust sources, driving forces, soil and land surface conditions, and human activities. This study investigated these influencing factors and their interactions and considered the uncertainty among these factors to effectively predict the incidence of aeolian dust. In this study, we focused on dust incidence in the downstream area of the Dajia River, Taiwan, and collected various types of data (i.e., weather and air quality monitoring data, high-resolution land data assimilation system, HRLDAS, data, and satellite images) to construct Bayesian belief network models for predicting the PM_(10) concentration. The PM_(10) concentration of 125 μg⁄m^3 was set as the threshold, and was discretized into four levels, in the two proposed Bayesian belief network models, respectively. The cross-validation and testing of the models revealed an overall prediction accuracy of >97% and >86% for river dust incidence and PM_(10) concentration, respectively. The results suggest that the models can produce accurate and credible predictions of the river dust incidence and PM_(10) concentration.