Characterizing Sand and Dust Storms (SDS) Intensity in China Based on Meteorological Data

Sand and dust storms (SDS) are global phenomena that significantly impact the socio-economy, human health, and the environment. The characterization of SDS intensity is a fundamental aspect of SDS issues and studies. In this study, a sand and dust storms index ( SDSI ) is developed to characterize SDS intensity by addressing the potential impacts of sand and dust storms on sensitive elements. Compared with other indices, SDSI includes four SDS-related components: SDS frequency, SDS visibility, SDS duration, and SDS wind speed. Using SDSI , this study characterizes the SDS intensity in the Three-North Forest Shelterbelt Program (TNFSP) region of China. The SDSI results show that high values of SDSI are mostly concentrated in southern Xinjiang, western and central Inner Mongolia, western and central Gansu, and northern Ningxia. By analyzing the SDSI components, over half of the stations experienced sand and dust storms no more than once per year on average. Most of the SDS events reduced horizontal visibility to less than 500 m, one-third of SDS events last more than two hours, and the wind speed of over half of the SDS events varied between 10–17 m/s. In comparison with SDS frequency, SDSI performs better in reflecting the spatial and temporal variation of SDS events. Therefore, instead of SDS frequency, SDSI can be applied to studies relevant to SDS intensity. Finally, five major SDS transportation routes were identified based on the surface prevailing wind direction, SDSI , and the existing literature. The SDS routes, combined with SDSI , could help governments and policy-makers cooperate on a regional level to combat SDS events more effectively.


Introduction
Sand and dust events are atmospheric processes that result in a reduction in visibility when strong and turbulent winds blow over desert or aridisol surfaces [1]. As a common phenomenon in arid and semi-arid areas, they have a variety of regional names and categories. For example, in the Middle East, sand and dust events are classified as one of the following three types: shamal, frontal, and convective [2]. Shamal events are the sand and dust storms coming from the north that are observed in summer and winter [3]. Frontal events mix the dust in the air and transport it great distances. Frontal

Data
In this study, SDS data were extracted from the Chinese Sand and Dust Events Sequence and Supporting Dataset (1954Dataset ( -2007 (http://data.cma.cn/), which was collected by the China Meteorological Administration (CMA) from 798 meteorological observation stations. The dataset provides relevant information about sand and dust events, including the geographical location of each observation station (longitude and latitude), wind direction, maximum 10-minute average wind speed (hereinafter refer to as wind speed), start time, end time, and visibility of each sand and dust event. The SDS dataset was obtained by selecting records with a visibility of less than 1 km, including sand and dust storms, strong sand and dust storms, and severe sand and dust storms. Before 1979, the visibility of sand and dust events at meteorological observation stations was recorded on a scale between 0 and 9, which is not consistent with the criterion afterward. Therefore, this study only incorporated SDS records from 1980 onward.

Data
In this study, SDS data were extracted from the Chinese Sand and Dust Events Sequence and Supporting Dataset (1954Dataset ( -2007 (http://data.cma.cn/), which was collected by the China Meteorological Administration (CMA) from 798 meteorological observation stations. The dataset provides relevant information about sand and dust events, including the geographical location of each observation station (longitude and latitude), wind direction, maximum 10-min average wind speed (hereinafter refer to as wind speed), start time, end time, and visibility of each sand and dust event. The SDS dataset was obtained by selecting records with a visibility of less than 1 km, including sand and dust storms, strong sand and dust storms, and severe sand and dust storms. Before 1979, the visibility of sand and dust events at meteorological observation stations was recorded on a scale between 0 and 9, which is not consistent with the criterion afterward. Therefore, this study only incorporated SDS records from 1980 onward.

Sand and Dust Storms Index (SDSI)
The sand and dust storms index (SDSI) is defined as the amount of dust received by the object per unit area during an SDS event. The SDSI was developed to address the potential impacts of SDS on sensitive elements, such as population, infrastructure, agriculture, and livestock. In order to calculate the SDSI, we provide a simplified illustrative diagram in Figure 2. The SDSI is calculated as follows: where m is the total mass of dust during an SDS event, S is the cross-sectional area of the object perpendicular to the direction of dust movement, ρ is the concentration of dust mass, V is the volume of the dust mass, l is the distance of dust movement, v is the velocity of dust movement, and t is the duration of the SDS event.

Sand and Dust Storms Index (SDSI)
The sand and dust storms index (SDSI) is defined as the amount of dust received by the object per unit area during an SDS event. The SDSI was developed to address the potential impacts of SDS on sensitive elements, such as population, infrastructure, agriculture, and livestock. In order to calculate the SDSI, we provide a simplified illustrative diagram in Figure 2. The SDSI is calculated as follows: where m is the total mass of dust during an SDS event, S is the cross-sectional area of the object perpendicular to the direction of dust movement, ρ is the concentration of dust mass, V is the volume of the dust mass, l is the distance of dust movement, v is the velocity of dust movement, and t is the duration of the SDS event. For the Chinese SDS sequence and supporting dataset, no information was recorded about the concentration of dust mass (ρ). However, ρ is closely correlated to the visibility of SDS events [42][43][44][45]. In this project, we replaced ρ by visibility using the equation below [46]: where ρ is the dust concentration in ug m -3 , and Dv is visibility in km. In addition, wind speed in the Chinese SDS sequence and supporting dataset is the maximum 10-minute average wind speed; thus, we modified SDSI as SDSImax: where SDSImax is the maximum possible SDSI during an SDS event, and vmax is the maximum 10minute average wind speed. Thus, the final SDSImax at the station level is defined as: where ASDSImax,j is the final SDSImax at station j, n is the total number of SDS events recorded at station j, and Tj is the valid years at station j (although the dataset ranged from 1980 to 2007, not all of the stations had full records for the 28 years). Through this method, SDS frequency was also incorporated into SDSI, where SDS frequency is defined as the number of SDS records in this study. For the Chinese SDS sequence and supporting dataset, no information was recorded about the concentration of dust mass (ρ). However, ρ is closely correlated to the visibility of SDS events [42][43][44][45]. In this project, we replaced ρ by visibility using the equation below [46]: where ρ is the dust concentration in ug m -3 , and D v is visibility in km. In addition, wind speed in the Chinese SDS sequence and supporting dataset is the maximum 10-min average wind speed; thus, we modified SDSI as SDSI max : where SDSI max is the maximum possible SDSI during an SDS event, and v max is the maximum 10-min average wind speed. Thus, the final SDSI max at the station level is defined as: where ASDSI max,j is the final SDSI max at station j, n is the total number of SDS events recorded at station j, and T j is the valid years at station j (although the dataset ranged from 1980 to 2007, not all of the stations had full records for the 28 years). Through this method, SDS frequency was also incorporated into SDSI, where SDS frequency is defined as the number of SDS records in this study.

Characteristics of ASDSI max and Its Components
From Figure 3, high values of ASDSI max were concentrated in southern Xinjiang, western and central inner Mongolia, Western and central Gansu, and north Ningxia. Particularly, Minfeng station, which is located on the southern edge of the Taklimakan Desert in Xinjiang, experienced the most significant SDS intensity from 1980 to 2007. The ASDSI max value of this station exceeded 146. The second highest ASDSI max value was recorded at the Guaizihu station of inner Mongolia, in the north of the Badain Jaran Desert. In addition, most of the stations with high values of ASDSI max were distributed around Gobi or deserts, such as the Taklimakan Desert, Badain Jaran Desert, Tengger Desert, and Mu Us Desert. These deserts have been recognized as important sources of sand and dust storms in China, and even East Asia [24][25][26]47]. SDS intensity was usually not severe at stations in northern Xinjiang, southern Shaanxi, Shanxi, Hebei, Tianjin, Beijing, and the three northeast provinces. The ASDSI max values at these stations were mostly below 25.

Characteristics of ASDSImax and Its Components
From Figure 3, high values of ASDSImax were concentrated in southern Xinjiang, western and central inner Mongolia, Western and central Gansu, and north Ningxia. Particularly, Minfeng station, which is located on the southern edge of the Taklimakan Desert in Xinjiang, experienced the most significant SDS intensity from 1980 to 2007. The ASDSImax value of this station exceeded 146. The second highest ASDSImax value was recorded at the Guaizihu station of inner Mongolia, in the north of the Badain Jaran Desert. In addition, most of the stations with high values of ASDSImax were distributed around Gobi or deserts, such as the Taklimakan Desert, Badain Jaran Desert, Tengger Desert, and Mu Us Desert. These deserts have been recognized as important sources of sand and dust storms in China, and even East Asia [24][25][26]47]. SDS intensity was usually not severe at stations in northern Xinjiang, southern Shaanxi, Shanxi, Hebei, Tianjin, Beijing, and the three northeast provinces. The ASDSImax values at these stations were mostly below 25. Since ASDSImax is developed based on SDS frequency, SDS visibility, SDS duration, and SDS wind speed, the characteristics of ASDSImax at different stations are determined by these components.
From Figure 4a, over half of the stations had an SDS frequency of less than 28, which means that these stations experienced sand and dust storms no more than once per year on average from 1980 to 2007. Seventeen stations had more than 280 records of sand and dust storms, of which most were located in Xinjiang and inner Mongolia. The highest SDS frequency was observed at Minfeng station, which also had the highest ASDSImax value. Among all of the SDS records, almost all of the SDS events reduced horizontal visibility to less than 500 m (Figure 4b), which means that these SDS events would probably be classified as strong or severe sand and dust storms. About 45% of the SDS events finished within one hour, whereas one-third of the SDS events lasted more than two hours ( Figure 4c). The longest SDS event, which lasted more than 24 hours, was recorded at the Sonid Left Banner station in inner Mongolia in March 2002. In terms of spatial distribution, over 90% of the SDS events that were longer than two hours occurred in Xinjiang, inner Mongolia, Gansu, Ningxia, and Qinghai. Generally, the wind speed of the SDS events was not less than 5 m/s, and over half of the events had wind speeds varying between 10-17 m/s ( Figure 4d). However, some stations, especially in inner Mongolia, Gansu, and Xinjiang, always experienced strong winds (>20 m/s) during the SDS events. The Alataw Pass station in Xinjiang recorded a wind speed that was greater than 40 m/s in April 1984.
Generally, SDS events occurred less than 10 times per year on average in the TNFSP region. However, most of the events led to a considerable reduction in visibility to even less than 50 m, and Since ASDSI max is developed based on SDS frequency, SDS visibility, SDS duration, and SDS wind speed, the characteristics of ASDSI max at different stations are determined by these components.
From Figure 4a, over half of the stations had an SDS frequency of less than 28, which means that these stations experienced sand and dust storms no more than once per year on average from 1980 to 2007. Seventeen stations had more than 280 records of sand and dust storms, of which most were located in Xinjiang and inner Mongolia. The highest SDS frequency was observed at Minfeng station, which also had the highest ASDSI max value. Among all of the SDS records, almost all of the SDS events reduced horizontal visibility to less than 500 m (Figure 4b), which means that these SDS events would probably be classified as strong or severe sand and dust storms. About 45% of the SDS events finished within one hour, whereas one-third of the SDS events lasted more than two hours (Figure 4c). The longest SDS event, which lasted more than 24 hours, was recorded at the Sonid Left Banner station in inner Mongolia in March 2002. In terms of spatial distribution, over 90% of the SDS events that were longer than two hours occurred in Xinjiang, inner Mongolia, Gansu, Ningxia, and Qinghai. Generally, the wind speed of the SDS events was not less than 5 m/s, and over half of the events had wind speeds varying between 10-17 m/s (Figure 4d). However, some stations, especially in inner Mongolia, Gansu, and Xinjiang, always experienced strong winds (>20 m/s) during the SDS events. The Alataw Pass station in Xinjiang recorded a wind speed that was greater than 40 m/s in April 1984.
lasted longer than half an hour. In addition, an SDS event in the TNFSP region was usually accompanied by strong winds with wind speeds no less than 10 m/s. With strong winds and long duration, SDS events resulted in significant social and economic losses.

Comparison of SDSI and SDS Frequency
SDSI combines SDS frequency, visibility, duration, and wind speed, so the index can theoretically reflect SDS intensity. In this section, we compare the spatial and temporal variations between SDSI and SDS frequency, since SDS frequency has been most commonly used in the existing literature. Figure 5 shows the spatial comparison of ASDSImax and SDS frequency (FREQ) as a percent of the total. As with ASDSImax, FREQ is calculated as the number of records of sand and dust storms in the valid years: where n is the total number of records of sand and dust storms at station j, and Tj is the valid years at station j. Generally, the ASDSImax and FREQ at each station have similar spatial distribution in terms of the percent of the total ( Figure 5). The highest value of both ASDSImax and FREQ were calculated at Minfeng station, but the value of FREQ was lower than ASDSImax in terms of the percent of the total. Actually, the sand and dust storms that occurred at this station were not only the most frequent, but also always lasted longer with lower visibility. Therefore, SDS intensity could be underestimated if we used FREQ to reflect the intensity at Minfeng station. For Kelpin station at the northern edge of Generally, SDS events occurred less than 10 times per year on average in the TNFSP region. However, most of the events led to a considerable reduction in visibility to even less than 50 m, and lasted longer than half an hour. In addition, an SDS event in the TNFSP region was usually accompanied by strong winds with wind speeds no less than 10 m/s. With strong winds and long duration, SDS events resulted in significant social and economic losses.

Comparison of SDSI and SDS Frequency
SDSI combines SDS frequency, visibility, duration, and wind speed, so the index can theoretically reflect SDS intensity. In this section, we compare the spatial and temporal variations between SDSI and SDS frequency, since SDS frequency has been most commonly used in the existing literature. Figure 5 shows the spatial comparison of ASDSI max and SDS frequency (FREQ) as a percent of the total. As with ASDSI max , FREQ is calculated as the number of records of sand and dust storms in the valid years: where n is the total number of records of sand and dust storms at station j, and T j is the valid years at station j.
Sustainability 2018, 10, x FOR PEER REVIEW 6 of 13 lasted longer than half an hour. In addition, an SDS event in the TNFSP region was usually accompanied by strong winds with wind speeds no less than 10 m/s. With strong winds and long duration, SDS events resulted in significant social and economic losses.

Comparison of SDSI and SDS Frequency
SDSI combines SDS frequency, visibility, duration, and wind speed, so the index can theoretically reflect SDS intensity. In this section, we compare the spatial and temporal variations between SDSI and SDS frequency, since SDS frequency has been most commonly used in the existing literature. Figure 5 shows the spatial comparison of ASDSImax and SDS frequency (FREQ) as a percent of the total. As with ASDSImax, FREQ is calculated as the number of records of sand and dust storms in the valid years: where n is the total number of records of sand and dust storms at station j, and Tj is the valid years at station j. Generally, the ASDSImax and FREQ at each station have similar spatial distribution in terms of the percent of the total ( Figure 5). The highest value of both ASDSImax and FREQ were calculated at Minfeng station, but the value of FREQ was lower than ASDSImax in terms of the percent of the total. Actually, the sand and dust storms that occurred at this station were not only the most frequent, but also always lasted longer with lower visibility. Therefore, SDS intensity could be underestimated if we used FREQ to reflect the intensity at Minfeng station. For Kelpin station at the northern edge of Generally, the ASDSI max and FREQ at each station have similar spatial distribution in terms of the percent of the total ( Figure 5). The highest value of both ASDSI max and FREQ were calculated at Minfeng station, but the value of FREQ was lower than ASDSI max in terms of the percent of the total. Actually, the sand and dust storms that occurred at this station were not only the most frequent, Sustainability 2018, 10, 2372 7 of 13 but also always lasted longer with lower visibility. Therefore, SDS intensity could be underestimated if we used FREQ to reflect the intensity at Minfeng station. For Kelpin station at the northern edge of the Taklimakan Desert, FREQ was higher than ASDSI max as a percent of the total. Over two-thirds of the SDS events ended within half an hour at this station. At some stations in Qinghai, such as Mangnai station and Gangcha station, ASDSI max was also lower than FREQ. SDS events occurring at these stations usually had shorter durations or higher visibility. Compared to ASDSI max , FREQ also underestimated SDS intensity in inner Mongolia. Inner Mongolia is not just a source of sand and dust storms in China, as the region is also significantly affected by SDS events [41,47,48].
Additionally, the annual trends in total SDSI max and total FREQ for all of the stations are shown in Figure 6. Both annual SDSI max and annual FREQ decreased with fluctuations from 1980 to 2007, except in 2003 and 2006. Annual SDSI max ranged from approximately 720 to 4360, whereas annual FREQ varied between 266-1373. However, annual SDSI max provides a sharper contrast for reflecting the interannual change. Especially in 1983, 1989, 1997, 2001, and 2006, the peaks in annual SDSI max were higher, whereas the troughs in annual SDSI max were lower compared with those of annual FREQ. In fact, the sand and dust storms in the peak years were not only more frequent, but also stronger and caused much larger socioeconomic losses [49]. Therefore, SDSI is more sensitive than SDS frequency at reflecting the temporal variation in SDS events. the Taklimakan Desert, FREQ was higher than ASDSImax as a percent of the total. Over two-thirds of the SDS events ended within half an hour at this station. At some stations in Qinghai, such as Mangnai station and Gangcha station, ASDSImax was also lower than FREQ. SDS events occurring at these stations usually had shorter durations or higher visibility. Compared to ASDSImax, FREQ also underestimated SDS intensity in inner Mongolia. Inner Mongolia is not just a source of sand and dust storms in China, as the region is also significantly affected by SDS events [41,47,48]. Additionally, the annual trends in total SDSImax and total FREQ for all of the stations are shown in Figure 6 1983, 1989, 1997, 2001, and 2006, the peaks in annual SDSImax were higher, whereas the troughs in annual SDSImax were lower compared with those of annual FREQ. In fact, the sand and dust storms in the peak years were not only more frequent, but also stronger and caused much larger socioeconomic losses [49]. Therefore, SDSI is more sensitive than SDS frequency at reflecting the temporal variation in SDS events. Further case studies and field validations are required to test the performance of SDSI in reflecting SDS intensity, but SDSI appears to be suitable for accurately characterizing sand and dust storms by combining SDS frequency, SDS visibility, SDS duration, and SDS wind speed [38,39,50]. Instead of SDS frequency, SDSI can be applied to studies that are relevant to SDS intensity. For example, the United Nations Convention to Combat Desertification (UNCCD) is developing global methodologies for SDS risk assessment and SDS vulnerability mapping for governments, policy makers, and communities for the prevention and mitigation of SDS impacts. A more accurate reflection of SDS intensity is an important factor for both methodologies; SDSI could improve the results of SDS risk assessment and SDS vulnerability mapping.

Major Transportation Routes of Sand and Dust Storms in the TNFSP Region
In order to trace the transportation routes of SDS events in the TNFSP region, the prevailing surface wind directions were drawn based on SDS records from each meteorological station. Stations with ASDSImax <1 and numbers of prevailing wind <10 were excluded to remove the interference of low-intensity SDS events and highlight the trend in SDS movements (Figure 7a).
Based on wind direction and the existing literature, the major transportation routes of sand and dust storms are illustrated (Figure 7b). Sand and dust storms caused by cold airflows from northern Asia (usually Siberian and Mongolian cyclones) [47,51] always sweep over Mongolia and inner Mongolia, and further extend toward the southeast and northeast provinces, such as Ningxia, Further case studies and field validations are required to test the performance of SDSI in reflecting SDS intensity, but SDSI appears to be suitable for accurately characterizing sand and dust storms by combining SDS frequency, SDS visibility, SDS duration, and SDS wind speed [38,39,50]. Instead of SDS frequency, SDSI can be applied to studies that are relevant to SDS intensity. For example, the United Nations Convention to Combat Desertification (UNCCD) is developing global methodologies for SDS risk assessment and SDS vulnerability mapping for governments, policy makers, and communities for the prevention and mitigation of SDS impacts. A more accurate reflection of SDS intensity is an important factor for both methodologies; SDSI could improve the results of SDS risk assessment and SDS vulnerability mapping.

Major Transportation Routes of Sand and Dust Storms in the TNFSP Region
In order to trace the transportation routes of SDS events in the TNFSP region, the prevailing surface wind directions were drawn based on SDS records from each meteorological station. Stations with ASDSI max <1 and numbers of prevailing wind <10 were excluded to remove the interference of low-intensity SDS events and highlight the trend in SDS movements (Figure 7a). Based on wind direction and the existing literature, the major transportation routes of sand and dust storms are illustrated (Figure 7b). Sand and dust storms caused by cold airflows from northern Asia (usually Siberian and Mongolian cyclones) [47,51] always sweep over Mongolia and inner Mongolia, and further extend toward the southeast and northeast provinces, such as Ningxia, Shaanxi, Shanxi, and Hebei. The airflows are occasionally blocked by the Beishan Mountains when passing across western inner Mongolia, so the sand and dust storms either turn right to northern Xinjiang or left into the Hexi Corridor of Gansu. The SDS routes mentioned above indicate one of the most significant SDS sources in the TNFSP region and East Asia: the Mongolian Plateau SDS source region, including deserts and Gobi deserts on the Mongolian Plateau and its southern extensions: the Ordos Plateau and Alxa Plateau [25,[52][53][54].   Two pathways of prevailing winds are responsible for SDS events in this region [55][56][57]. Westerly winds cross the Pamir Plateau and arrive in the Tarim Basin through Kashgar. Airflows in this direction usually do not move out of the Tarim Basin due to the blockage created by the Tianshan Mountains. Easterly winds flowing into the Tarim Basin from the east open area also play an important role in generating SDS activities in the Taklimakan Desert. Therefore, most of the SDS events occurring in the Taklimakan Desert would deposit into the desert, especially in the Hotan, Yutian, and Minfeng regions, where the convergence of the two prevailing winds lead to high-intensity sand and dust storms [47,58]. In addition to the ground surface SDS routes, the Taklimakan Desert is also regarded as a source of long-distance SDS events in the remote North Pacific Ocean, where dust particles are raised by strong upward winds that climb over the mountains [48].
In the spring and summer, westerly winds from Central Asia dominate Northern Xinjiang, and airflows usually move into this region through the Alataw Pass and the south edge of the Altai Mountains. Compared to southern Xinjiang, northern Xinjiang is not a major SDS source in the TNFSP region. However, airflows in this region usually spread into most parts of the TNFSP region, such as Qinghai, Gansu, and inner Mongolia, causing SDS events [57,59,60]. In this study, two SDS routes in Gansu and Qinghai were extracted due to the geographical settings. The Hexi Corridor in Gansu is located between the Beishan Mountains and Qilian Mountains, and borders Kumtag Desert to the west, Badain Jaran Desert to the north, and Tengger Desert to the east. Besides the local SDS source, the Hexi Corridor experiences SDS transported from the west (Gobi and Kumtag Desert) and north (Gobi, Badain Jaran Desert and Tengger Desert) [61][62][63]. In Qinghai, the Qaidam Basin is surrounded by the Qilian and Altun Mountains to the north and the Kunlun Mountains to the south. Therefore, the downwind areas of Qaidam Desert experience SDS events when westerly winds from Xinjiang flow into Qinghai [64].
Whether the Three-North Forest Shelterbelt Program is performing well in combating desertification and SDS events is still debatable, but countermeasures should be adopted in this region [65,66]. The map of the major transportation routes of sand and dust storms, combined with the results of SDSI, could help the government and policy makers more accurately and effectively address SDS issues. For example, Figure 5 indicates that SDS intensity is higher in inner Mongolia. Considering the major transportation routes in this region, more ecological projects should be implemented to restore rangelands and control the movement of sand and dust, especially in the upwind area [67,68]. The SDS routes also imply that SDS events originate from both inside and outside China [54,69]. Therefore, regional cooperation should be addressed to combat sand and dust storms.

Conclusions
Single parameters of sand and dust storms (SDS) cannot accurately reflect SDS characteristics. In this study, we developed a new index (SDSI) combining four parameters: SDS frequency, SDS visibility, SDS duration, and SDS wind speed. The distribution of ASDSI max at each station showed that southern Xinjiang, western and central Inner Mongolia, western and central Gansu, and north Ningxia usually experience intense sand and dust storms. Over half of the stations in the TNFSP region experienced sand and dust storms less than once per year on average, and the visibility of most SDS events was reduced to less than 500 m. In addition, one-third of the SDS events lasted more than two hours, and 45% finished within one hour. The wind speed during the SDS events usually ranged from 10 m/s to 17 m/s. SDS intensity at Minfeng station was underestimated if SDS frequency instead of SDSI was used. SDS frequency also exaggerated the SDS intensity at stations such as Kelpin, Mangnai, and Ganghca. Compared with SDS frequency, SDSI is more sensitive to the temporal variation in SDS events, since it provides a sharper contrast for the interannual change. In general, SDSI characterized sand and dust storms more accurately, and this index can be applied to other SDS-related issues and studies.
Surface prevailing wind direction with an ASDSI max of less than 1 and a frequency of prevailing wind of less than 10 were used to extract major SDS transportation routes. Five transportation routes were identified in this study. The first originates from Mongolia and inner Mongolia, and usually sweeps over the entire TNFSP region. The second route circulates inside Tarim Basin in southern Xinjiang. Although the SDS route originating in Northern Xinjiang does not cause severe local SDS events frequently, it is a significant SDS pathway when moving forward to inner Mongolia, Gansu, and Qinghai. Another two SDS routes in Gansu and Qinghai were also identified due to their unique geographical settings. The SDSI map and SDS routes can help governments and policy makers enable regional cooperation and combat SDS events more effectively.
Author Contributions: H.C. and J.L. conceived and designed the methodology for this study; H.C., W.Z. and C.F. wrote the manuscript.