This study addresses this imperative by proposing an intelligent hybrid algorithm, integrating Group Method of Data Handling (GMDH) with Harmony Search (HS), to offer a novel, scalable, and cost-effective framework. The objective is to significantly enhance dust storm prediction accuracy by effectively capturing intricate environmental interactions, thereby critically advancing the state-of-the-art in environmental hazard forecasting. This study aims to model the frequency index of dust storm days (FDSD) using integrated Group Method of Data Handling - Harmony Search (GMDH-HS) intelligent algorithm at eight synoptic stations in Kurdistan Province, Iran over a 50-year statistical period (1971–2020). The performance of two intelligent models, Group Method of Data Handling (GMDH) and Group Method of Data Handling - Harmony Search (GMDH-HS), was evaluated and compared. The results, based on goodness-of-fit criteria during the training and testing phases, indicated a marginal but statistically insignificant difference in accuracy between the two models. The Hybrid model of Data Handling - Harmony Search (GMDH-HS) model exhibited superior performance, achieving lower error values of 0.113, 1.231 percent respectively for NRMSE and MAPE metrics. The selection of a suitable model should be guided by the specific characteristics of the project, the dataset, and existing constraints. By leveraging the strengths of both components—GMDH’s ability to automatically select the optimal network structure and HS’s efficiency in fine-tuning the model parameters—this hybrid framework offers a state-of-the-art, scalable, and computationally efficient predictive tool, thereby critically advancing the accuracy and efficiency of environmental hazard forecasting models.