Wind tunnel experiments were conducted in Jiufeng Sandstorm Physics Laboratory of Beijing Forestry University. The experimental system is composed of three parts: the wind tunnel body, a three-dimensional measurement system, and wind speed measurement equipment. The wind tunnel is a standard, direct-current wind tunnel, with a total length of 24 m and seven sections ( Figure 1 ). An electric motor provides power to the fan blades, allowing airflow to enter the transition section. The stabilization section uses a honeycomb-shaped wire mesh to disperse large masses of air, making the airflow more uniform. When the airflow passes through the contracted section, it accelerates due to a reduction in the cross-section as it enters the experimental section. The length of the experimental section is 12 m, with sharp wedges (having a right-angled triangular pyramid shape, with a height of 20 cm and a length of 5 cm on the bottom inclined side) and rough elements (4 cm height × 5 cm width × 6 cm length) arranged at the front end to adjust the wind speed profile. The cross-sectional area is 0.6 m × 0.6 m ( Figure 1 ), and the boundary layer is 10 cm. After passing through the experimental section, the airflow traverses the diffusion section.

The wind speed in the wind tunnel is adjustable over the range of 3–42 m·s^-1, and the effective experimental cross-sectional wind speed fluctuation is less than 1.5%. The three-dimensional measurement system can move freely throughout the experimental section to obtain wind speed measurements at different positions and heights within the experimental section, effectively creating a three-dimensional measurement grid. The measurement system has a movement accuracy of 1 mm. The wind speed measurement equipment includes a hot film anemometer (IFA300, TSI, Shoreview, MN, USA) and a hot wire anemometer (KIMO, Montpon-Ménestérol France).

Populus popularis is a commonly used afforestation tree species in northern China, which has the characteristics of rapid growth and strong drought resistance. Salix psammophila is a common shrub species distributed in arid and semi-arid areas in China, and has the characteristics of drought resistance, barren resistance and saline-alkali resistance. This study takes P. popularis and S. psammophila as the research species and adopted a simulation tree model with a scaling ratio of 1:100. The model parameters are set according to the commonly used configuration models for planted forests ( Table 1 ). The tree height was set at 10 cm, the height-to-crown base was 3 cm, and the crown diameter was 3 cm. The shrub height was 2.5 cm and its crown diameter was 2 cm. The thickness of the boundary layer at the bottom of the experimental section was 25 cm, which exceeded the height of the model. The constrictivity of the forest belt model was less than 5%, which met the requirements of similar conditions for wind tunnel simulations. The configurations of the forest belts included single- and double-belt forests ( Figure 2 ). In the double-belt configuration, the two rows were arranged in a triangular pattern and numbered separately.

Wind speed measurements were conducted along the longitudinal sections and horizontall plane of the experimental section. The distribution of the measurement points is shown in Figure 3 . The layout range of longitudinal measurement points spanned from 30 cm in front of the tree line to 100 cm behind the tree line, with a spacing interval of 2 cm. The lowest measurement point was 1 cm, and the highest measurement point was 21 cm. The layout range of horizontal measurement points started from 50 cm behind the forest belt, with a spacing of 2 cm and a height of 2 cm. In combination with meteorological data in arid areas, we set the experimental wind speed as 8 m/s (The measured height is 2m). The minimum Reynolds number under the experimental conditions was 4.8 × 10⁵. At each measurement point, wind speed was measured with a KIMO hot wire anemometer for 10 s, and the average value was calculated as the wind speed at that measurement point.

The leeward wind profile of different forest belts is shown in Figure 4 . The presence of the forest belts significantly changed the structural characteristics of the wind speed profile. The configuration mode of mixed trees and shrubs in front of the forest belt provided the best wind prevention effect, followed by the double-belt trees, both of which were significantly better than the other configurations. Three times the height of the tree in front of the forest belt and the configuration modes of a single-belt tree forest and double-belt shrub forest had a reducing effect on wind speed within the height range of 5–10 cm. The wind speed difference between the other ranges and the control was small, while the single-belt shrub forest was consistent with the control wind speed at different height ranges, indicating that the shrub forest had no effect on wind speed here. Corresponding to the height of the tree in front of the forest belt, the wind speed of different configuration modes of the forest belt was lower than that at three times the tree height in the same forest belt, and the wind speed was lower than the control wind speed within the range of 0–10 cm. Except for the mixed forest of tree and shrub and the double-belt tree forest, the wind speed of the other configurations increased gradually above the control wind speed after exceeding a height of 15 cm. The wind speed profile at different positions behind the forest belt was similar. The wind speed of the forest belt with shrubs increased with the height; the wind speed decreased significantly within the range of shrub height. However, the tree forest belt showed a pattern in which the wind speed decreased at first and then increased due to the height-to-crown base. The protective effect of the double-belt tree forest was better than that of the single-belt forest. The windbreak effect of the mixed forest belt with tree and shrub was not as good as that of the two-row tree forest; however, this was compensated for by the protection of areas below the height of the branches. The wind speed of forest belts with different configurations exceeded the control wind speed within the tree height range.

The wind speed statistics along the longitudinal direction after flowing through different forest belts are shown in Table 2 . The effective protection height of the double-belt tree forest was the highest, exceeding the height of the tree itself, followed by that of the single-belt tree forest and the mixed tree and shrub forest belt. The effective protective height of the shrub belt was lower than that of the tree forest, but exceeded the height of the shrubs themselves. Within the effective protective height, the average wind speed behind the forest belt showed a pattern of two rows of forest belts being lower than that of the single forest belt, and the shrub forest being lower than that of the tree forest. The single-belt tree forest had the highest average wind speed, which was 1.29 times the lowest average wind speed of the double-belt shrub forest. Compared with the double-belt pure forest, the mixed forest of tree and shrub had a larger protective area than the shrub forest, and better protection capabilities than the tree forest, with the lowest average wind speed in the area behind the forest belt.

The interpolation simulation of the longitudinal wind speed flow field in different configuration modes is shown in Figure 5 . The airflow formed a distinct low wind-speed zone after passing through the forest belt, and the wind speed decreased significantly within the height range of the forest belt. The airflow above the forest belt experienced an acceleration effect, in which the wind speed exceeded that in front of the forest belt. There were multiple eddies in the wind-speed flow field of the pure tree forest belt, and the air disturbance was strong. Due to the presence of height-to-crown base, the near-surface wind speed was higher, and the protective effect at the height of the canopy layer was better. Within the effective protective height, the wind speed behind the single-belt tree forest was mainly within the range of 6.5–8.0 m/s, whereas the wind speed behind the double-belt tree forest ranged from 4.0 to 7.0 m/s. The air disturbance in forest belts with shrubs was not as severe as in tree forests, and the airflow was more stable after passing through the forest belts. The wind speed after traversing the single-belt shrub forest ranged from 4.5 to 7.0 m/s, whereas the wind speed after the double-belt shrub forest was 1.0 to 6.0 m/s. The mixed forest of tree and shrub had a high protective height; the wind speed near the surface was reduced significantly, thus combining the advantages of tree and shrub forest belts in which the wind speed behind the forest belt ranged from 2.5 to 7.0 m/s.

The horizontal wind speed flow field after interpolation simulation of different forest belts is shown in Figure 6 . The wind flow field of single-belt shrub and double-belt shrub forest had similar structural characteristics, both formed a clear low wind-speed zone and then the wind speed increased gradually, but the effect of wind speed reduction in double-belt shrub was more obvious. In addition, compared with the wind flow field at the higher position, the wind speed reduction effect near the surface was better, and the low wind speed area was dominated by the blue area. The structural characteristics of the wind flow field were similar at different heights when air flow passed through pure trees. After passing through the single-belt tree forest, the airflow formed a high wind speed zone due to the “narrow tube effect”. The contour line in the downwind zone was parallel to the wind direction, indicating that the airflow was relatively stable and the wind speed had not changed. The airflow that passed through the double-belt tree forest formed a high wind-speed zone within the range of the tree height. However, unlike the single-belt tree forest, the wind speed decreased gradually in the wind direction. Unlike shrub forests, the wind speed near the surface of trees was higher than that at high places due to the acceleration of “narrow tube effect” between trunks. After the airflow passed through the mixed forest of tree and shrub, a large area of low wind speed was formed behind the forest belt, and the near surface wind speed was significantly lower than the higher position. The reduction degree of wind speed and the area of low wind speed in the mixed forest of tree and shrub at different heights were better than those of other forest belts.

The wind protection efficiency of different forest belts was calculated, and the results are shown in Figure 7 . The double-belt shrub forest and the mixed tree and shrub forest had wind protection efficiency exceeding 30%, and the efficiency distribution was relatively uniform. The wind prevention efficiency of double-belt shrub ranged from 40% to 80%, while that of mixed forests of tree and shrub was mainly within the range of 30% and 70%. The wind prevention efficiencies of single-belt and double-belt forest were poor, with a minimum wind prevention efficiency of only 10%. Among them, the wind prevention efficiency of single-belt shrub ranged from 20% to 50%, single-belt tree forest from 10% to 50%. The wind prevention efficiency of double-belt tree forest from 20% to 40%, and the range of 30–40% accounted for 66% of the distribution.

The windbreak effect of a shelterbelt depends on the wind energy consumption of the forest canopy, and the effect is affected by tree height, crown width, and other factors (Zastrow, 2019; Cheng et al., 2020; Torita and Masaka, 2020). This study shows that the tree forest has played a good role in wind protection, and has a high protection height and strong airflow disturbance ability ( Figure 5 ). The wind speed after the tree forest belt gradually decreases within the experimental range and there is no trend of recovery ( Figure 6 ), indicating that the tree forest has a longer effective protection distance. Sai (Sai et al., 2021) reached the same conclusion that trees and shrubs have different effects on airflow, tree belts have a stratification effect on airflow, while shrub belts have a lifting effect on airflow. The shrub forest belts have better protective effects within effective protective heights ( Figures 5 , 6 ). The taller the vegetation, the longer the protection distance of the forest belt will be, consistent with Wu (Wu et al., 2013). High porosity reduces the protection effect, and low porosity can form strong eddy currents in the leeward side of the forest belt, shortening the protection distance (Dong et al., 2007), and we came to a similar conclusion. The presence of under-branch height in a tree forest leads to higher forest belt sparsity ( Table 1 ), while the short-fat” nature of shrubs leads to lower forest belt sparsity, which is also the main reason for the difference in wind protection effects between the two types of forest belts. It should be noted that in our study the model was made based on the characteristics of trees with leaves. In arid and semi-arid regions, the defoliation period of forest belts is usually a period of strong winds. A further increase in the sparsity of broad-leaved trees after defoliation leads to a continuous decrease in their protective effect. Therefore, attention should be paid to the supplement of shrub species in afforestation.

China has a large population, and the relationship between people and land is tense. Therefore, a large number of farmlands, grassland and other production land are distributed in arid and semi-arid areas. For these scenes, near-surface wind erosion protection is critical, such as preventing soil particles from being eroded by wind in degraded grassland, reducing wind speed near the surface in farmland to avoid lodging of crops. Shrub belts can not only reduce the wind speed near the surface, but also cover the ground with branches to form a protective layer, making them more suitable in arid areas. Shrubs can effectively increase the organic matter content of soil. Soil C content and C stocks were on average 148% and 117% greater in shrub-encroached grassland soil (Phesheya et al., 2019). In addition, shrubs can also intercept soil particles, form nebkhas, produce fertilizer island effect, and contribute to the improvement of soil physical and chemical properties and the change of microbial community structure (Wang et al., 2010; Cao et al., 2016). In summary, shrub belts should play an important role in afforestation of arid areas.