Integration of Satellite Observation and Dust Trajectory Modeling for Dust Transport and Dispersion Monitoring

Land use/cover change was distinguished as one of the most local environmental consequences and has an important role in increasing dust due to land degradation. The aim of study was to identify the local dust storm sources in Khuzestan province (southwestern Iran) due to land use/cover change and their dispersion and transport modeling from these sources by Lagrangian particle HYSPLIT model. In the rst part of study, Landsat images for 1984, 1992, 2002, and 2017 were used in order to produce the land use/cover maps and post classication comparison was applied for land use/cover change detection from 1984 to 2017. The results of change detection showed that wetland areas (Horolazim wetland and Shadegan international wetland) experienced sever changes and converted to probable dust storm centers. Subsequently, HYSPLIT model was performed in dust trajectory modeling from the detected dust storm centers for 365 days of 2016 in three elevation level. Aerosol optical depth (AOD) product of MODIS and air quality monitoring(AQM) data were used for HYSPLIT performance assessment. The ndings of modeling showed that storms originated from detected dust storm centers can be transferred to the transboundary of Khuzestan province, especially in hot season. Therefore, focusing on national programs could be as valuable as international cooperation for mitigation and controlling dust storms in southwestern Iran.


Introduction
Climate change phenomena result in more than just a change in the weather; a variation in climate can effect on some related aspects of reduced precipitation, drought occurrence followed by expansion of deserts, and drying of the earth (IPCC 2007). Dust is a key indicator of climate change (Shao et al. 2013) which could play an important role in past climate changes and may contribute to the future course of climate change (Harrison et al. 2001). Land use/cover change has been distinguished as one of the most local environmental consequences, but it is becoming a universal challenge (Foley et al 2005). It is beyond doubt that human activities such as land-use activities (Ning et (Watson et al., 1996) and they have an important role in increasing dust (Sokolik and Toon 1996). It is predicted that about half of dust particles are deposited near the source area (30%) and re-distributed on a local scale (20%) and the other half of them are expected to be transferred to farther distances (Kim et al., 2003). The dust cycle includes two key physical processes: (i) a wind lifting process due to raising up of dust particles from bare land surfaces (ii) a large scale transportation process with a high spatial coherence (Guarnieri et al. 2011;Kok et al. 2014).
Understanding the role of dust in the earth's climate system has led to more development of dust transport and distribution models from the 1980s. Air quality models are applied methods and algorithms in order to study and nd the relationship between source and receptor (Zannetti and Puckett 2004). The The Middle East is famous for its arid and semi-arid ecosystems with frequent and intense dust/sand storms (Furman 2003). Historical reports of dust storms for Iran indicated that during the past decade, dust/sand storms have increased in intensity especially in its southwest parts (Esmaili et al. 2006).
Experts believed that the dust in the southwest area of Iran originates from external and internal sources in a way that 5% of it has an internal origin and 95% of it has an external origin (Dargahian et al. 2017). Turkey, Iraq, and Syria have generated the largest amount of dust, and other countries such as Jordan, Saudi Arabia, and North African countries are also effective in the occurrence of this phenomenon International cooperation between Iraq, Syria, and Iran for future strategies in order to reduce dust storm with external origine and its impacts on region is costly and time consuming. Unfortunately, existing war and human con ict in Iraq and Syria have caused the dust storm and its impacts not to be the priority for their governments. Accordingly, focusing on national plans could be as valuable as international cooperation for mitigation and controlling dust storms in southwestern Iran, especially in Khuzestan province. According to the literature reviews, there is not any research that attempts to perform forward HYSPLIT Model for dust trajectory modeling from local dust sources in huzestan province. Therefore, in this research, besides the land cover/use change detection in a longer period than the previous researches for identifying the probable local dust sources due to land use/cover change, the dust forward transport and dispersion modeling was performed by the HYSPLIT model in seasonal formation (hot and cold seasons). Our objective is the recognition of dust trajectories that arise from local dust sources in order to control better the dust in Khuzestan and the other provinces in Iran.

Study area
Khuzestan province is located in southwestern Iran (between and E longitudes and and N latitudes) and covers an area of about 63.21 km 2 , which is 3.9% of the total area of Iran (Fig.1). The climate of Khuzestan is generally very hot and occasionally humid, particularly in the southern part, while its winter is relatively cold and dry, whereas the north of Khuzestan experiences cold weather. The annual mean of maximum and minimum temperatures are 50°C and 9°C, respectively. The annual mean of rainfall increases from south to north of Khuzestan gradually and the annual evaporation is 2000-4000 mm.
Also, some studies showed that northwestern and western winds are the prevailing winds in this area. The wind speed in summer and autumn was lower than spring and summer (  is considered to be one of the most precise classi ers since it is based on statistical characteristics. For more details please refer to Richards, 1986. The area was classi ed into ten main land cover classes including salt, urban, water body, forest, rangeland, rock, swamp, wetland, cropland, and aquaculture. The explanation of different land use/cover classes are expressed in table 3. Kappa coe cient is the most popular multi-variate technique for accuracy assessment (Congalton and Mead 1983; Stehman 1996). In this research after calculating the error matrix, the overall accuracy and Kappa coe cient were expressed as accuracy assessment.

Land use/cover change detection
In this study, the post-classi cation comparison method was applied for land use/cover change detection. It is the most popular method of land cover change detection, which compares independent thematic maps (Shalaby and Tateishi, 2007).

HYSPLIT model
In this research, the HYSPLIT model was used, which is a shared research program of the U.S. National Oceanic and Atmospheric Administration (NOAA) and Australia's Bureau of Meteorology (ABOM).
HYSPLIT is used in dust storm forward/backward trajectory simulations. The model has been widely used to distinguish the aerosol sources and sinking processes (Rana et al. 2009). Forward trajectory simulation is a perfect instrument for investigating transport pathways of air pollutants and representing  . 2), overall accuracy and Kappa coe cient were calculated using the error matrix (Table 4). Finally, land cover maps of the years 1984 and 2017 were compared by post-classi cation comparison.
One of the most considerable changes is the increase in cropland up to 12.38% (799918.02 ha) from 1984 to 2017 (Table 5) The degradation process of Shadegan international wetland goes more quickly. When the population in the surrounding of wetland increases, it lead to demand for economic activities and development. Most of the economic activities and the development plans around the wetland have adverse effects on wetland due to the lack of environmental assessment of development plans.
The results indicated a decrease in the area of Shadegan international wetland, especially in the northern part of the wetland called the freshwater wetland. Most of the threats in this area can be caused by increased access to the wetland (road construction). Also, water-fed wetlands through streams in this area, are entered into the wetland and strongly effect on the salt, sediment, and water quality of the wetland. The cumulative effects of the threats have a high risk on this part of the wetland. Due to the construction of several dams and the implementation of irrigation network development projects in recent years, the drought has led to changes in the vegetation cover of these wetlands, which was identi ed as a source of dust (Javadian et al. 2019). Rahimi Blouchi and Malekmohammadi (2013) showed that instead of the change in natural habitats, change in hydrological patterns due to dam construction and also water pollution has been identi ed as the most important risks of the Shadegan international wetland.

Transport and dispersion modeling
After the detection of dust storm sources based on land use/cover change in Khuzestan province, HYSPLIT forward trajectory model was performed to nd the possible paths of dust storm and their dispersion scale. This modeling was performed for the south of Horolazim (center 1) and north of Shadegan international wetland (center 2) in three elevation levels of 500, 1000, and 1500 meters for 365 days in 2016.
According to the HYSPLIT model results, dust particles in the hot period (spring and summer) of the year 2016, had tracks to the southwestern Khuzestan, southern Bushehr, Hormozgan, Persian Gulf, and even some days to Saudi Arabia (Fig. 5) According to Figure 6, the MODIS color composite image was similar to MODIS AOD product of July 19 th and 20 th 2016. Analysis of the visibility data of AQM stations indicated that there were 30 dust-storm days in July 2016 (visibility <1 km) in Khuzestan. Figure 7 shows that the visibility value during the studied days has decreased in Khuzestan, Ilam, and Lorestan's AQM stations. Based on the true color composite of MODIS sensor, dust storm in 17 July 2016 after passing through deserts of the north of Saudi Arabia and south of Iraq would extend to Khuzestan, Ilam, Hormozgan, and Bushehr located in the south and southwest of Iran. AOD product recorded a high value of particles in Khuzestan province (Fig   6). Khuzestan Province is the area that is heavily in uenced by deserts of the north of Saudi Arabia and south of Iraq. The images clearly show that about 61% area of Khuzestan province covered with desert ecosystems (having low density vegetation) and western provinces are engaged with dust storm, except some of the northern and eastern mountainous areas. Rajaee et al. (2020) and Borna et al. (2021) used MODIS colore composite image and AOD product for HYSPLIT modeling assessment.

Conclusion
The purpose of this study was the identi cation of the probable internal or local dust centers due to land use/cover change in Khuzestan province. In order to study the role of land use/cover change as one of the effective factors in the formation of the dust storm, land use/cover change detection was done during 1984-2017. The results represented that the southern regions of the Horolazim wetland and the northern area of Shadegan international wetland were as the local sources of dust storms. Reduced vegetation cover and soil disturbance can make more sediment available for emission in dust storms. It is worth mentioning that the sugarcane industry is playing a signi cant role in the Khuzestan economy and native and non-native people occupation and has direct or indirect negative environmental impacts on croplands and Shadegan international wetland. Unfortunately, Iranian authorities have not considered sustainable development for industrial growth such as sugarcane cultivation and industry in Khuzestan province. Unsustainable developments in this province has caused the formation of new dust storm centers due to land use/cover conversion that had negative environmental impacts on the human and natural areas inside and outside of Khuzestan.
The HYSPLIT model results showed that more than 90 percent of Khuzestan province area, southern of Lorestan province, Ilam, Fars and Bushehr provinces, southwest of Hormozgan province and parts of the Kohgiluyeh and Boyer Ahmad province will be subjected to the transport and dispersion of dust originated from these two centers specially in cold periods. The dust path trajectory modeling in warm period indicated the majority of dust path are leaded to south of Khuzestan and Persian Gulf and effected mariane ecosytme. Although the share of internal/local dust center is lower than external/regional hotspots, the severe land degradation especially cropland, wetlands, and rangelands conversion to deserts as important local dust storm centers and transfer of dust particles to other cities of Iran as internal management programs should be followed, seriously. Planting of endemic vegetation in the region, especially saline grasses or bushes in accordance with the ecological conditions of Khuzestan and avoiding water transfer projects can be useful for dust storm mitigation in the future.

Declarations
Funding (Not applicable) Con icts of interest/Competing interests (I certify that they have NO a liations with or involvement in any organization or entity with any nancial in the subject matter or materials discussed in this manuscript)          Visibility data of AQM stations in June 2016