Grain Size Characteristics of Sediments Found in Typical Landscapes in the Playa of Ebinur Lake, Arid Central Asia

: A playa usually refers to a salt desert landscape mainly composed of loose and ﬁne lacustrine sediments. Severe wind erosion on a playa causes the playa to become a source of dust and salt dust and poses a threat to vast areas downwind. Currently, little is known about the impact of wind erosion on the particle size distribution of sediments in different landscapes in the playa. In the present study, six dominant different landscapes in a natural state with the same sedimentary environment in the playa of Ebinur Lake were selected to provide insights into the different characteristics of particle size distribution under the effect of long-term wind erosion. The results reveal that the grain-size composition clearly differed among different landscapes. All samples had a common dominant size group consisting of very ﬁne sand and sand. The very ﬁne sand and sand content of Haloxylon ammodendron desert zone (LS5) was the lowest, while the clay and silt content was the highest at both depths among the six landscapes. The lowest clay and silt fraction and highest sand fraction appeared in the herbal desert zone (LS3) at both depths. Almost all of the sediment samples were of a bimodal distribution mode, with signiﬁcant differences. The cumulative curve showed a similar S-shape, while the probability cumulative curve showed an inverted S-shape with three subpopulations of granularity characteristics. The smallest mean particle diameter appeared in LS5. The majority of the sediments were moderately to poorly sorted. The mean particle size of the sediments from the six landscapes was signiﬁcantly different ( p < 0.05), while no signiﬁcant difference was observed among the other three parameters. Generally, it can be inferred that LS5 can reduce wind speed effectively, probably due to the smaller leaves and dense branches of Haloxylon ammodendron, which results in a high level of coverage. The results of the present study will have some implications for the grain size characteristics for changes in intensity in regional wind erosion environment and will also have some basis for wind erosion prevention and control in the playa of Ebinur Lake.


Introduction
A playa usually refers to a discharging intracontinental basin with a negative water balance, remaining dry for 75% of the year, and is often associated with evaporites in arid and semi-arid land [1,2]. Specifically, playas are a salt desert landscape that came into being after the existing water balance was broken by climate change and human activity; they are mainly composed of loose and fine lacustrine sediments with a high salt content which accumulate due to the evaporation of highly mineralized groundwater in tail-end lake basins. Playas are mainly located in arid central Asia, the western United States, North Africa, Australia, and other arid and semi-arid places around the world [3][4][5].
Arid regions receive a lot of attention, since they compose 41% of the world's surface area and are occupied by 2 billion people [6]. Most arid land, especially the arid central Ebinur Lake is located in the western part of Junggar basin, northwest China (Figure 1a(a)). A large area of the salt desert landscape formed in the dry lake bed under the influence of climate change and human activities. Strong airflows from the Alashankou (gate of Junggar Basin) determines the development of wind erosion in this region [46], and dust (salt) storms caused by wind erosion can increase soil salinization in downwind areas, accelerate snow and ice melt in mountain areas, and affect vegetation growth by affecting normal photosynthesis [47,48] around the region. Currently, little is known about the impact of wind erosion on the particle size distribution of sediments in different landscapes within the playa of Ebinur Lake in arid central Asia. In the present study, six dominant different landscapes in a natural state with the same sedimentary environment were selected to provide insights into the different characteristics of the particle size distribution under the effect of wind erosion at a small regional scale, and the results will explain the difference and changes in particle size composition in different landscapes, provide a basis for understanding the quantitative relationship between vegetation and aeolian processes, and provide a basis for the selection of biological and physical measures to control wind erosion in the playa in arid zones around the world.

Regional Setting and Wind Erosion
Ebinur Lake basin is surrounded by mountains in the south, west and north (Figure 1a), and lies in the southwest border region of the Junggar basin in arid central Asia with an area of 50,321 km 2 [8,49]. Ebinur Lake is the lowest water collection point of the basin (82 • 35 -83 • 10 E, 44 • 54 -45 • 09 N) and is the tail-end lake of the Boertala River and Jing River, at present. This region has a typical temperate continental climate characterized by rare precipitation, strong evaporation and perennial winds. The number of days per year with wind speeds greater than 8 m/s is up to 164 days, and the maximum instantaneous wind speed can peak at 55 m/s [50].
Since the second half of the 20th century, Ebinur Lake has undergone a rapid decline in area due to the coupled impact of regional climate fluctuations and human activities, leading to the formation of an area of about 1000 km 2 of modern salt desert landscape ( Figure 1a). Severe wind erosion on the surface sediments rich in sulphates and chlorides occur in spring and autumn, causing the playa of Ebinur Lake to be a source of dust and salt dust and thus posing a threat to farming and animal husbandry in vast areas downwind, as well as a serious threat to the ecological security of oases along the north-facing slope of the Tianshan Mountains.

Samples and Analytical Methods
The present study selected a flat terrain zone with different vegetation types and the same hydrodynamic conditions to ensure no significant differences in sedimentary environments. Sediments of 0-30 cm and 30-60 cm from six typical different vegetation types (Table 1) were sampled using the Eijkelkamp stainless steel soil sampler in the historical playa within the southeastern region of Ebinur Lake (Figure 1b, about 1500 m length, 1000 m width). In total, one hundred and eight samples weighing 500 g each were collected from the chosen zone. The geographical position, status of vegetation and characteristics of sediments around the sample sites were also recorded ( Table 1). Wind erosion in Ebinur Lake playa mainly occurs from April to July every year, and therefore the sampling time was selected to be in June, after severe wind erosion. Sediment samples were air dried, crushed and passed through a 2 mm mesh, and the plant roots, sand and other gravel-sized materials were removed by hand. Before each measurement, chemical pretreatment was performed to remove the organic matter and to provide discrete particles for further processing. The particle size distribution of samples was analyzed by using the Malvern Mastersizer laser grain size analyzer 2000, with a measurement range from 0.02 µm to 2000 µm, and this resulted in a better than 2% accuracy and a better than 2% variation in terms of reproducibility. The salt content was also measured after pretreatment by means of the residue-drying method. Firstly, 1:5 soil:water suspensions were prepared by weighing 50 g of soil in a 500 mL triangular flask, adding 250 mL of deionized water, and shaking for 3 min on an end-over-end shaker. The soil:water extracts were used for further analysis after suction filtration. Then, we took a certain amount of soil:water extract and placed it in a 100 mL porcelain evaporating dish, evaporated the water from the soil:water extracts in a water bath, oxidized the organic matter with hydrogen peroxide, dried it in an oven at 105-110 • C and weighed it to obtain the qualities of the dried residue, namely the salt content.
Particle size analysis is mainly used to study the structural characteristics of sediments (particle size and particle size distribution). In this paper, the values of particle size distribution were converted to the Φ (phi) unit, and the Folk and Ward [51] formula was used to obtain particle-size parameters of sediments from different landscapes. Duncan's multiple testing methods were used for multiple comparisons of size parameters from different landscapes, for which p < 0.05 indicated that the difference was significant.

Grain Size Composition of Sediments from Six Different Landscapes
The classification of the grain size composition intuitively expresses the fraction of different grain size groups. According to the grain size classification criteria for sediment of Friedman and Sanders [52], the grain size composition of the sediments from six different landscapes is divided into 11 levels ( Table 2). The grain size composition clearly differed among different landscapes. All samples had a common dominant size group that consisted of very fine sand and sand, accounting for around seventy percent, with a size mainly ranging from 63 µm to 500 µm. The grain size composition of the top layer (0-30 cm) of sediment from the six landscapes had distinct and significant differences, except for the very fine sand, coarse sand and very coarse sand; however, that of the lower depth (30-60 cm) showed no significant differences, except for medium sand. Generally speaking, the very fine sand, medium sand and sand content of LS5 was the lowest at both depths (55.57% and 57.97%), while the clay and silt content was the highest at both depths (43.92% and 41.15%) among the six landscapes, indicating that wind erosion presumably imposed a lighter impact on LS5 because of the vegetation coverage and morphological characteristics of vegetation. LS1 and LS6 demonstrated little difference in the sand and clay content in the top layer, but the difference in the lower layer was obvious. The lowest clay and silt content appeared in LS3 in both depths, which was not very different from LS2 and LS4. This trend also applied to sand of LS3, LS2 and LS4 with intermodulation. It must be mentioned that the clay and silt proportion from LS6 was significantly lower than that of LS5 at both depths; this is what we must and will pay more attention to in the Discussion section.
The frequency, cumulative and log-probability cumulative curves of the grain size distribution sediments from different landscapes present a lot of useful information on particle size components and their transport patterns. Creep, saltation and suspension subpopulation can be observed from the curves concretely and easily [53].
Almost all of the sediment samples were of a bimodal distribution mode with significant differences ( Figure 2). From LS1 to LS6 (Figure 2a), the first crest particle size was around 20 µm, with this crest being highest in LS5; the second crest particle sizes ranged from 100 µm to 170 µm, and the percentage of grain size composition ranged from 5% to 7%, with the lowest in LS5 being around 100 µm, and the highest in LS4 being around 150 µm in grain diameter. The crest particle sizes at lower depths ranged from 110 µm to 190 µm, and the percentage of grain size composition ranged from 6% to 7.5%, with the same trend in the upper layer ( Figure 2c). LS1 and LS6 showed the same trend in both depths, as did LS2 and LS3. The percentage of grain size composition in LS6 showed no significant differences compared to the other landscapes. These observations indicated the strengthening of wind activity among different landscapes. This characteristic is also in accordance with the observations of grain size composition mentioned previously.
The cumulative curve of all samples showed a similar S-shape ( Figure 2). The cumulative curve of LS5 almost transformed into a linear shape because of the similar contents of the different size groups in the top layer (Figure 2b), which showed a significant difference from other landscapes. All the sediments from different landscapes at the lower depths have a similar shaped cumulative curve that shared the same gradient with a small creep population (Figure 2d). These different characteristics of sediment subpopulations from different landscapes at different depths can be clearly seen from the log-probability cumulative curves ( Figure 3). For all the samples, there were three subpopulations, namely creep, saltation and suspension, which were truncated at almost the same point, about 3.50 Φ (creep group) or 10.60 Φ (suspension group). The creep and saltation subpopulation accounted for the majority, while the suspension subpopulation only occupies a small proportion. This indicated that sediments were transported by the same saltation processes with different sorting parameters. It can be seen more clearly that the saltation subpopulation (3.50-10.60 Φ) in LS5 was the highest and those in LS2 and LS3 were the lowest among the six landscapes, while the creep (2.00-3.50 Φ) and suspension (10.60-11.50 Φ) subpopulations, which are not as susceptible to wind erosion, were the lowest for both Land 2021, 10, 1132 6 of 15 depths at present, which was the same as in the previous analysis. There were no significant differences in the content of the three subpopulations of sediments among the other five landscapes, especially LS6.
icant differences ( Figure 2). From LS1 to LS6 (Figure 2a), the first crest particle size was around 20 μm, with this crest being highest in LS5; the second crest particle sizes ranged from 100 μm to 170 μm, and the percentage of grain size composition ranged from 5% to 7%, with the lowest in LS5 being around 100 μm, and the highest in LS4 being around 150 μm in grain diameter. The crest particle sizes at lower depths ranged from 110 μm to 190 μm, and the percentage of grain size composition ranged from 6% to 7.5%, with the same trend in the upper layer ( Figure 2c). LS1 and LS6 showed the same trend in both depths, as did LS2 and LS3. The percentage of grain size composition in LS6 showed no significant differences compared to the other landscapes. These observations indicated the strengthening of wind activity among different landscapes. This characteristic is also in accordance with the observations of grain size composition mentioned previously. The cumulative curve of all samples showed a similar S-shape ( Figure 2). The cumulative curve of LS5 almost transformed into a linear shape because of the similar contents of the different size groups in the top layer (Figure 2b), which showed a significant difference from other landscapes. All the sediments from different landscapes at the lower depths have a similar shaped cumulative curve that shared the same gradient with a small creep population (Figure 2d). These different characteristics of sediment subpopulations from different landscapes at different depths can be clearly seen from the log-probability cumulative curves ( Figure 3). For all the samples, there were three subpopulations, namely creep, saltation and suspension, which were truncated at almost the same point, about 3.50 Φ (creep group) or 10.60 Φ (suspension group). The creep and saltation subpopulation accounted for the majority, while the suspension subpopulation only occupies a small proportion. This indicated that sediments were transported by the same saltation processes with different sorting parameters. It can be seen more clearly that the saltation subpopulation (3.50-10.60 Φ) in LS5 was the highest and those in LS2 and LS3 were the lowest among the six landscapes, while the creep (2.00-3.50 Φ) and suspension (10.60-11.50 Φ) subpopulations, which are not as susceptible to wind erosion, were the lowest for both depths at present, which was the same as in the previous analysis. There were no significant differences in the content of the three subpopulations of sediments among the other five landscapes, especially LS6.

Grain Size Parameters of Sediments from Six Different Landscapes
The mean particle size of the sediments from six landscapes showed significant differences (p < 0.05), while no significant differences were observed among the other parameters (standard deviation, skewness and kurtosis) (Figure 4). The largest mean particle diameter (4.36 Φ) appeared in LS5 at both depths, indicating that the particle size distribution mainly consisted of very coarse silt and clay (31-63 μm); compared with other landscapes, this makes this area very susceptible to wind erosion, and the same results  Table 2. Grain size composition of sediments collected from six landscapes (%, different letters in the same column indicate significant differences in percentage of grain size composition between different landscapes (* p < 0.05, ** p < 0.01); the significance decreases in the order of a > b > c.).

Grain Size Parameters of Sediments from Six Different Landscapes
The mean particle size of the sediments from six landscapes showed significant differences (p < 0.05), while no significant differences were observed among the other parameters (standard deviation, skewness and kurtosis) (Figure 4). The largest mean particle diameter (4.36 Φ) appeared in LS5 at both depths, indicating that the particle size distribution mainly consisted of very coarse silt and clay (31-63 µm); compared with other landscapes, this makes this area very susceptible to wind erosion, and the same results were shown in the previous section. LS3 demonstrated the smallest mean particle size in both layers, with a significant difference from LS5 (2.90 Φ and 2.83 Φ for top and lower layers, respectively). The mean particle size, standard deviation, skewness and kurtosis of LS6 show no significant differences with sediments from other landscapes. In general, the range of the mean particle size of the surface layer is larger than that of the lower layer. The mean particle sizes in the surface layer in LS1 and LS2 are larger than those in the lower layers at a depth of 30-60 cm, and those of the surface layer in LS3, LS4, LS5, LS6 are smaller than those in the layer at a depth of 30-60 cm. This indicates that LS1 and LS2 contain more clay and silt than the lower layer, while the clay and silt content in LS3, LS4 and LS5 have reduced, probably due in part to the more serious wind erosion.  The spatial distribution of the standard deviation (σI) of the sediments collected from six landscapes showed that the majority of the sediments were moderately to poorly sorted, with a value between 0.8 and 2.4 ( Figure 5). The variation of the standard deviation (σI) of the sediments showed that sorting generally decreased from LS1 to LS6. This sorting was generally due to the decrease in the grain size of top sediments. Sands collected from the upper depth are moderately to poorly sorted; however, sands of the lower depth were better to moderately sorted. Very fine sand and very coarse silt were mostly poorly to very poorly sorted at both depths. Overall, LS2 and LS3 were moderately sorted, and LS5 was mostly in the poorly sorted category. This correlation showed that finer sediments are characterized by poor sorting. The spatial distribution of the standard deviation (σ I ) of the sediments collected from six landscapes showed that the majority of the sediments were moderately to poorly sorted, with a value between 0.8 and 2.4 ( Figure 5). The variation of the standard deviation (σ I ) of the sediments showed that sorting generally decreased from LS1 to LS6. This sorting was generally due to the decrease in the grain size of top sediments. Sands collected from the upper depth are moderately to poorly sorted; however, sands of the lower depth were better to moderately sorted. Very fine sand and very coarse silt were mostly poorly to very poorly sorted at both depths. Overall, LS2 and LS3 were moderately sorted, and LS5 was mostly in the poorly sorted category. This correlation showed that finer sediments are characterized by poor sorting.
(σI) of the sediments showed that sorting generally decreased from LS1 to LS6. This sorting was generally due to the decrease in the grain size of top sediments. Sands collected from the upper depth are moderately to poorly sorted; however, sands of the lower depth were better to moderately sorted. Very fine sand and very coarse silt were mostly poorly to very poorly sorted at both depths. Overall, LS2 and LS3 were moderately sorted, and LS5 was mostly in the poorly sorted category. This correlation showed that finer sediments are characterized by poor sorting. The distribution of the skewness of the superficial sediments showed that LS5 and LS6 were generally dominated by very positively skewed very fine sand and very coarse silt (0.35 < SK I < 0.55), with the exception of LS3, which was dominated by positively skewed sand and very fine sand at the depth of 0-30 cm. This same pattern was observed at the lower depth. The variation of the skewness of sediments showed no obvious trends from LS1 to LS6. The correlation between the mean grain size and the skewness showed that the majority of sands were positive and very positively skewed. However, very fine sand and very coarse silt were very positively skewed ( Figure 6). The distribution of the skewness of the superficial sediments showed that LS5 and LS6 were generally dominated by very positively skewed very fine sand and very coarse silt (0.35 < SKI < 0.55), with the exception of LS3, which was dominated by positively skewed sand and very fine sand at the depth of 0-30 cm. This same pattern was observed at the lower depth. The variation of the skewness of sediments showed no obvious trends from LS1 to LS6. The correlation between the mean grain size and the skewness showed that the majority of sands were positive and very positively skewed. However, very fine sand and very coarse silt were very positively skewed ( Figure 6). Correlations between the standard deviation (σI) and the skewness (SKI) of the sediments collected from the different landscapes are shown in Figure 7. Sediments from different landscapes were, in the majority, moderately to poorly sorted and positively and very positively skewed (Figure 7). This is the case for sands for most sediments from different landscapes in the playa of Ebinur Lake. Correlations between the standard deviation (σ I ) and the skewness (SK I ) of the sediments collected from the different landscapes are shown in Figure 7. Sediments from different landscapes were, in the majority, moderately to poorly sorted and positively and very positively skewed (Figure 7). This is the case for sands for most sediments from different landscapes in the playa of Ebinur Lake. Correlations between the standard deviation (σI) and the skewness (SKI) of the sediments collected from the different landscapes are shown in Figure 7. Sediments from different landscapes were, in the majority, moderately to poorly sorted and positively and very positively skewed (Figure 7). This is the case for sands for most sediments from different landscapes in the playa of Ebinur Lake.

Discussion on the Grain Size Characteristics of Sediments Found in Typical Landscapes and Its Implications
Grain size distribution is a very important issue for aeolian research [54]. Examining grain size characteristics is one basic method used to determine the development of the aeolian landforms, as well as the transportation and sorting of aeolian particles [55]. The grain size composition includes creep, saltation, and suspension in aeolian environment formation and development. The grain size study of different landscapes at a small scale, usually found in arid and semi-arid land, has been carried out rarely worldwide. For a set of given conditions, in order to ensure the consistency of the sediment deposition environment, in this paper, we selected a flat terrain zone with different vegetation types and the same hydrodynamic conditions as an example to illustrate the regional long-term wind erosion environment and the response of different landscapes to the wind erosion environment using the calculated grain size characteristics.
Plant morphology, spatial structures, and vegetation density are all key factors controlling the interaction between the surface and vegetation, which affects sediment transport and which will then impact the grain size distribution [56,57]. The study area we selected is located on the historical playa of the Ebinur Lake, and the hydrodynamic environment within the area is considered the same within such a small regional scale. Thus, the prominent factors that will impose impacts on the grain size composition will be the vegetation types and the intensity of evaporation affected by vegetation types. The different coverage caused by the individual morphological characteristics of plants will protect the sediments in different landscapes from wind erosion by reducing the wind speed, and so the grain size composition and mean particle size of sediments clearly differed among different landscapes, especially for the upper layer, for which all grain size grading shows a significant difference (Table 2 and Figure 4). Overall, the Haloxylon ammodendron desert zone (LS5), a typical desert shrub, has the highest clay and silt content and the lowest content of very fine sand and sand at both depths among the six landscapes, consistent with the results found by Sharifi et al. [58] in the playa of Urmia Lake in Western Asia as well as with the results found by King et al. [59] at the Salton Sea in the USA; this can probably be attributed to the longer leaves and dense branches of Haloxylon ammodendron, which can reduce wind speed effectively. Thus, LS5 has the smallest mean particle diameter (4.36 Φ) and the highest salt content, with significant differences (p < 0.01) compared to the other five landscapes in the top layer ( Table 2 and Figure 8). The frequency, cumulative and log-probability cumulative curves of the grain size distribution of the sediments from different landscapes also highlighted the difference of LS5 from other landscapes. only had a certain degree of wind erosion, which caused the content of fine fraction and salt content to be lower than that seen in LS5. The grain size composition, mean grain size, standard deviation, skewness, and kurtosis of the Phragmites australis desert zone (LS4) show no significant difference compared to LS1 because of similar morphological characteristics which can protect the sediments from these landscapes from being significantly transported in any form (creep, saltation, suspension).   The results of the present paper will provide some implications regarding the grain size characteristics for the change in the regional wind erosion intensity environment and a basis for wind erosion prevention and control in the playas of arid land. During strong wind erosion periods in arid land, landscapes without vegetation in the playa may have stronger resistance to wind erosion because of the emergence of a salt crust, rather than landscapes with low vegetation coverage and relatively small evaporation. It can be seen that the control of wind erosion within the playa in arid areas cannot be solved only through an increase in vegetation coverage due to the fact that different vegetation types have different protective effects on sediments; selecting more appropriate plant types according to the suitable soil/sediment conditions is also necessary. An optimal vegetation density and planting pattern is also an important factor that needs to be considered. The shrubs desert zone (LS2) and the herbal desert zone (LS3) are the highest at both depths regarding the sand fraction and the lowest at both depths regarding the clay and silt fraction; this may be because, in reality, the low vegetation coverage of LS2 and LS3 (about <6%) and the relatively low level of evaporation lead to the formation of loose surface sediments, with a small amount of salt accumulating near to the surface (Figure 8). Wind erosion is more serious in the clay and silt fraction, which is prone to erosion, which causes corresponding content reduction and leads sediment surfaces to become much rougher over a long period of wind erosion [60].
However, bare land (LS6) will accumulate a large amount of salt close to the surface because of the strong evaporation in this region ( Figure 8). When evaporation is strong and the groundwater level drops, the salt will condense and bond with clay and silt, forming a hard salt-silt-clay combination with bacteria, namely a biological soil crust, and the critical threshold wind speed becomes higher, which improves the resistance to wind erosion [61][62][63]. This can be confirmed by the fact that the salt content of the sediment has a significant positive correlation with the mean particle size (Table 3). Biological soil crusts in dry lake beds can be used to stabilize dust and sand from dried-up lake beds when faced with wind erosion [61,64]. During the period of freezing and thawing, the salt crust only had a certain degree of wind erosion, which caused the content of fine fraction and salt content to be lower than that seen in LS5. The grain size composition, mean grain size, standard deviation, skewness, and kurtosis of the Phragmites australis desert zone (LS4) show no significant difference compared to LS1 because of similar morphological characteristics which can protect the sediments from these landscapes from being significantly transported in any form (creep, saltation, suspension).
The results of the present paper will provide some implications regarding the grain size characteristics for the change in the regional wind erosion intensity environment and a basis for wind erosion prevention and control in the playas of arid land. During strong wind erosion periods in arid land, landscapes without vegetation in the playa may have stronger resistance to wind erosion because of the emergence of a salt crust, rather than landscapes with low vegetation coverage and relatively small evaporation. It can be seen that the control of wind erosion within the playa in arid areas cannot be solved only through an increase in vegetation coverage due to the fact that different vegetation types have different protective effects on sediments; selecting more appropriate plant types according to the suitable soil/sediment conditions is also necessary. An optimal vegetation density and planting pattern is also an important factor that needs to be considered.
In the present study, we presented a grain size study on the variation of the particle size distribution of the soil/sediment from different landscapes, which will help us to understand the anti-erosion effect of different landscapes and the change trend of particle size distribution. The quantitative interactions between the vegetation (plant height, porosity, canopy width, leaf area, etc.) and aeolian transport (dust emission) in drylands, especially in the playas in arid lands, still requires further research to make up the knowledge gap through the coupling of models and field monitoring.

Conclusions
In the present study, six dominant different landscapes in a natural state with the same sedimentary environment in the playa of Ebinur Lake in arid central Asia were selected to provide insights into the different characteristics of the particle size distribution of sediments from different landscapes in the playa under the effects of wind erosion. The results and conclusions can be summarized as follows.
The different coverage caused by the individual morphological characteristics of plants protected the sediments in different landscapes from wind erosion by reducing the wind speed, and so the grain size composition clearly differed among different landscapes. All samples had a common dominant size group consisting of very fine sand and sand, accounting for approximately seventy percent of the sample, with sizes mainly range from 63 µm to 500 µm. The sand fraction of LS5 was the lowest at both depths, while the clay and silt content was the highest at both depths among the six landscapes. The lowest clay and silt content appeared in LS3 at both depths due to wind erosion causing the surface to become much rougher.
Almost all of the samples were of a bimodal distribution with significant differences. The percentage of grain size composition ranged from 5% to 7%, with the lowest value, in LS5, being around 100 µm in grain diameter. The cumulative curve showed a similar S-shape, while the probability cumulative curve shows an inverted S-shape with three subpopulations according to granularity characteristics. The saltation subpopulation in LS5 was the highest among the six landscapes, while those in LS2 and LS3 were the lowest among the six landscapes after long-term wind erosion.
The mean particle size of the sediments from the six landscapes showed significant differences (p < 0.05), while no significant differences were observed among the other three parameters. The smallest mean particle diameter appeared in LS5 with a value of 4.36 Φ. The standard deviation (σ I ) of the majority of the sediments was moderately to poorly sorted, with a value between 0.8 and 2.4 among the six landscapes. LS2 and LS3 were moderately sorted, and LS5 was in the poorly sorted category, which showed that finer sediments are characterized by poor sorting. LS5 and LS6 were generally dominated by very positively skewed very fine sand and very coarse silt of the superficial sediments.
The Haloxylon ammodendron desert zone (LS5) had the strongest resistance to wind erosion, which can be attributed to the longer leaves and dense branches of this type of vegetation, which can reduce wind speed effectively. The shrubs desert zone (LS2) and the herbal desert zone (LS3) showed the weakest resistance to wind erosion due to the low vegetation coverage and relatively low level of evaporation. However, the wind erosion resistance of bare land (LS6) was stronger than expected because of the emergence of a salt crust. The present results provide an insight into the anti-erosion effects of different landscapes. It must be pointed out that the interaction between the vegetation and aeolian