The sand and dust weather pertains to a meteorological occurrence characterized by vigorous winds that elevate substantial quantities of dust and sand particles from the Earth’s surface into the atmosphere, contingent upon specific geographical and underlying surface conditions, causing the air to become turbid and reducing horizontal visibility [1]. It is a disastrous weather event that commonly occurs in semi-arid, arid, and desert regions. Sand and dust weather events are characterized by their sudden onset, short duration, and wide-ranging impact [2]. The repercussions of dust weather are multifaceted; it detrimentally impacts air quality [3, 4] alters regional climates [5], disrupts marine ecosystems [6, 7], poses risks to human health [8], and hinders economic growth [9, 10]. Moreover, its effects can be delayed yet persistent with long-term consequences [9, 11].
In recent times, the association between levels of particulate matter (PM) with sand and dust events has attracted considerable interest from numerous scholars [4, 12]. Research indicates that during such events, levels of PM10 and PM2.5 can rise to various extents. Consequently, PM concentration serves as a crucial indicator, reflecting not only the intensity of dust events but also their significant contribution to atmospheric pollution [13]. Presently, employing ground-based observation stations for the long-term continuous monitoring of PM10 has proven effective in tracking dust weather phenomena. This method is extensively utilized across various countries [14, 15], regions [16, 17] and cities [13, 18] for predicting and monitoring dust conditions [19].
The occurrence of dust weather is influenced by the interplay of surface conditions and meteorological factors, presenting a complex and intertwined relationship [20, 21]. Vegetation coverage, surface roughness, and soil terrain play a substantial role in influencing the emission of sand particles by modulating the mechanisms involved in their upliftment [22, 23]. Meteorological factors represent and characterize large-scale atmospheric pressure, wind patterns, and climatic shifts that can either exacerbate or mitigate the emergence of sand and dust events [24]. These factors exert diverse effects on the frequency of sand and dust occurrences. For example, while wind speed is a critical determinant in these events’ manifestation, while surface conditions have a significant impact on sand emissions, it is important to note that they are not the sole influencing factor [25]. Temperature, precipitation, relative humidity, evaporation, and various other elements can indirectly affect the release of dust by modifying the underlying surface [26]. Simultaneously temperature affects sand and dust activity both directly and indirectly through alterations in soil surface properties, atmospheric circulation patterns ,and vegetation density [27, 28]. Furthermore, the relationship between individual meteorological factor and dust varies. For example, precipitation usually suppresses the occurrence of dust, but a certain amount of precipitation sometimes actually helps to lift the dust. Consequently, the interplay between sand, dust weather, and meteorological variables is intricate. Quantifying the threshold of meteorological factors that affect sand and dust weather, and revealing the analysis of the mechanism of sand and dust occurrence and development, can lay a certain foundation for the study of sand-lifting mechanisms in different regions. This, in turn, enables more effective localized prevention measures against these natural hazards.
To further mitigate the losses caused by sand and dust weather events, accurate forecasting of such incidents is crucial. Currently, a variety of methods exist for predicting sand and dust conditions. Machine learning, in particular, has seen extensive use in meteorological observation and monitoring, mumerical forecasting, identification and warning of hazardous weather, and processing of meteorological satellite data, leading to significant advancements [29]. Commonly employed tools include artificial neural networks [30], deep neural networks, and random forests [31, 32]. Research indicates that the random forest approach excels in fitting non-linear patterns when simulating and predicting meteorological elements as well as air pollution levels [33], which explains its growing use in modeling pollutant concentrations. Research has found that the integration of a vertical gradient between neighboring pressure fields, utilizing the random forest model, enhances the accuracy of PM10 prediction. The random forest regression model not only provides timely and effective forecasts of PM10 and PM2.5 concentrations across China but also demonstrates a strong predictive capacity for shifts in particle concentration levels and peak occurrence dates. Clearly demonstrating its high predictive accuracy with air quality indices and pollutant concentrations alike; the random forest method boasts impressive generalizability and robustness in performance.
The majority of the North China region falls within the mid-latitude zones characterized by arid and semi-arid climates. This area is peppered with numerous Gobi deserts and sandy expanses, including notable formations such as Mu Us Sandy Land, Otingdag Sandy Land, and Horqin Sandy Land [34]. Vegetation coverage here is sparse, a condition that becomes particularly evident in spring as the soil warms and thaws, the exposed surface, characterized by its fine sandy soil, serves as a significant material reservoir for the generation of sand and dust, contributing to their occurrence. Moreover, unsustainable agricultural practices and living habits have exacerbated land desertification in this region, leading to more frequent sand and dust storms [3]. North China stands out as one of China’s most densely populated regions with vigorous industrial activities and economic development. Notably, there has been an uptick in the frequency of sand and dust weather phenomena during springtime over recent years [19]. Since March 2023 alone, there have been eight significant episodes of such weather conditions across China—the highest count for this period in over ten years. The sandstorm event spanning March 19 to March 23 was particularly severe—the most intense with the broadest impact recorded since the start of 2023—impacting twenty provinces (regions/cities) across an area exceeding 4.85 million square kilometers.
Current research on sandstorms in North China tends to focus on isolated incidents; comprehensive studies that delineate patterns or characteristics reflecting temporal changes under climate change are comparatively rare. To mitigate environmental damage and economic losses due to these storms, it is crucial to analyze their spatial-temporal variations along with their underlying formation mechanisms; additionally tracking their trajectories can offer valuable insights for predicting future occurrences—providing scientific support for regional pollution control strategies.
In summary, this article’s primary research objectives include: (1) Identify sand and dust weather events that have occurred in the North China region over the preceding nine-year period (2015–2023) using actual measurement data from national air quality monitoring stations and remote sensing satellite imagery, while analyzing the frequency of these events across various temporal scales; (2) Unraveling the intricate relationship between meteorological factors and the incidence of sand-dust occurrences, as well as investigating the underlying mechanisms that trigger these weather phenomena; (3) Selecting representative sand-dust weather episodes to employ the HYSPLIT model for pinpointing dust origins and tracing their transport routes; (4) Leveraging the established link between meteorological conditions and dust incidents to identify key atmospheric variables influencing dust event frequencies. These variables are then used to forecast PM10 concentrations in major North China cities via a random forest approach, with subsequent validation of its predictive accuracy. This study endeavors to enhance our understanding of the dynamics of sand-dust weather, thereby providing a theoretical and scientific basis for strategies concerning dust management in environmentally fragile areas, urban ecological planning, disaster prevention, and mitigation.