For the purpose of forecasting meteorological time series, a number of established methodologies have been presented (analog techniques, neural networks, etc.). However, when dealing with multivariate time series that are spread spatially, the aforementioned approaches do not take into consideration all of the information that is accessible. In this research, we tackle the challenge of using Bayesian networks to provide accurate weather predictions. This research aims to offer the Bayesian network (BN) model as a means of estimating dustfall based on atmospheric ensemble forecasts. The following categories of data were collected and analyzed for this study: mean temperature, mean humidity, mean rainfall, mean wind speed, and maximum and lowest dustfall. Between the years 2021 and the beginning of 2022, the Iraqi agency for Meteorology and Seismology collected the data on the weather. The values available are month averages. The structure and parameters of the Bayesian network were learned with the help of these data. Through the use of dynamic Bayesian networks, inference in the Bayesian network was able to assist in anticipating the maximum and minimum dustfall that will occur in the following months. The model demonstrated an overall precision of 75% and 89.0% in predicting the maximum amount of dustfall, respectively.