The impacts of freeze–thaw cycles on saturated hydraulic conductivity and microstructure of saline–alkali soils

Study on the microscopic structure of saline–alkali soil can reveal the change of its permeability more deeply. In this paper, the relationship between permeability and microstructure of saline–alkali soil with different dry densities and water content in the floodplain of southwestern Shandong Province was studied through freeze–thaw cycles. A comprehensive analysis of soil samples was conducted using particle-size distribution, X-ray diffraction, freeze–thaw cycles test, saturated hydraulic conductivity test and mercury intrusion porosimetry. The poor microstructure of soil is the main factor that leads to the category of micro-permeable soil. The porosity of the local soil was only 6.19–11.51%, and ultra-micropores (< 0.05 μm) and micropores (0.05–2 μm) dominated the pore size distribution. Soil saturated water conductivity was closely related to its microscopic pore size distribution. As the F–T cycles progressed, soil permeability became stronger, with the reason the pore size distribution curve began to shift to the small pores (2–10 μm) and mesopores (10–20 μm), and this effect was the most severe when the freeze–thaw cycle was 15 times. High water content could promote the effects of freeze–thaw cycles on soil permeability and pore size distribution, while the increase of dry density could inhibit these effects. The results of this study provide a theoretical basis for the remediation of saline–alkali soil in the flooded area of Southwest Shandong.


Materials and methods
Experimental materials. The test soil samples were collected in He-ze City, Shandong Province, which is located in the Yellow River Basin at an altitude of 50 m in the southwest of Shandong Province, and is a typical Yellow River flooding area in Shandong Province. Its region has a warm temperate monsoon continental climate with strong seasonality in temperature, rainfall, wind and evaporation. The average annual temperature in this region is 13.7 °C, the extreme minimum temperature is − 16.5 °C, and the extreme maximum temperature is 40.5 °C. It belongs to the seasonal frozen soil region. The annual frost-free period is 210 days, and the annual average precipitation is 625 mm, which is concentrated in June to September. The annual average wind speed is 1.95 m s −1 , and the average wind speed from January to April is 2.33 m s −1 . The annual average evaporation is 874.82 mm, the maximum value appears in June, and the minimum value appears in January. The area is inland and lacks fresh water, so the Yellow River is the only source of irrigation water.
The local soil was severely damaged by salt, and its EC 1:5 reached 1.44 dS m −1 , PH < 8.5, belonging to weakly alkaline soil. At the same time, exchange sodium percentage (ESP) reaches 31.14%, far more than 15%, so these poor soils are usually classified as "saline-alkali soil" ( Table 1). The accumulation of salt, especially excessive Na + , led to the deterioration of soil properties in this area, with a salt content of 61.96 g kg −1 . The K s of the soil in this area was only 1.48 × 10 -5 cm s −1 , which belongs in the micro-permeable water class according to the code for engineering geological investigation of water resources and hydropower 33 .
Soil samples were collected from the vicinity of Dongming County, He-ze City, Shandong Province, located at 115.08° E, 35.38° N (Fig. 1). A number of undisturbed soil samples and scattered soil samples were taken from the test site for later measurement of the basic physical properties of the soil and preparation of soil samples for subsequent experiments. The clay (< 0.002 mm) in the soil samples is 4.95%, and the silt (0.002-0.02 mm) is 42.90%, sand (> 0.02 mm) in soil, reaching 52.15%. The soil texture of this region can therefore be classified as silty clay. Other basic physicochemical properties of the soil in this area are shown in Table 1.

Preparation of specimens.
A small number of soil samples were dried at 105 °C, crushed and sifted through 75 μm for PSD and XRD analysis. The other samples were air-dried, rolled and crushed and passed through a 2 mm sieve. The F-T cycles (N) of 0, 1, 5, 10, 15, 20 were set. The water content was 13%, 16%, 19%, and the dry density of remolded soil was 1.48 g cm −3 , 1.53 g cm −3 , 1.58 g cm −3 , a total of 54 groups of samples ( Table 2). The remolded sample were left in the shade for 24 h to ensure sufficient moisture diffusion. Then, the 77 × 40 mm soil sample was prepared by electrohydraulic demodding mechanism. The 61.8 × 40 mm ring knifes coated with vaseline were used to cut the samples and wraped with the plastic film for F-T cycle test and saturated hydraulic conductivity test. At the end of the above tests, we gently removed the soil samples from the Particle size distribution and X-ray diffraction before F-T cycles. The PSD analysis of soil can facilitate us to understand the composition of soil particles and the uniformity of particle size distribution. The BT-9300S laser particle analyzer (Dandong Baite Instrument Co., Ltd., Dandong City, China) was used to complete soil PSD analysis, and two grading indexes were used to analyze the uniformity of soil particles, namely uniformity coefficient (C u ) and curvature coefficient (C c ). Uniformity coefficient is used to reflect the distribution of different particle groups, and C c is used to judge the continuity of slope of cumulative curve of soil particle size distribution, and it is also an important index to judge whether soil particle size distribution is good. The expressions are as follows: It is found that the particle size of saline-alkali soil in this area is in the range of 1-100 μm, and the particle density curve was unimodal (Fig. 2). The relevant soil characteristic parameters in this area, C u = 7.14, C c = 0.33. According to the regulations of the Ministry of Water Resources of the People's Republic of China 36 , C u > 5, the distribution of soil grain size is not uniform. C c < 1, indicating that the soil is dominated by small particles. From this, we can preliminarily judge that the grain size distribution of saline-alkali soil in this area is poor, leading to its poor permeability.
(1)  www.nature.com/scientificreports/ The mineral composition of soil affects the physical and chemical properties of soil 37,38 , which we analyzed using a TD-3500 X-ray diffractometer (Dandong Tongda Technology Co., Ltd., Dandong City, China). The semi-quantitative analysis of soil minerals was carried out using Jade 6.5 software. Figure 2 shows the XRD map of the saline-alkali soil in this area. Table 3 shows the mineral composition of the local saline-alkali soil. The mineral composition of the soil includes quartz, albite, orthoclase, calcite, kaolinite, illite and chlorite. Primary minerals account for more than 90% of the total minerals, and quartz is the highest, accounting for 51.85% of the total minerals, followed by albite and calcite. Clay minerals (kaolinite, illite, chlorite) only accounted for 9.55% of the total mineral content of the soil, among which illite was dominant, reaching 5.68% of the total mineral content of the soil.

Freeze-thaw cycling experiments.
Freezing and thawing will change the physical properties and internal structure 39 of soil. F-T tests were carried out in a closed system and no moisture was supplied to the sample during the test for providing conditions similar to field conditions 40 . In this study, the freeze-thaw cycles were split into six: the initial state (no cycle) and sets ending after the 1st, 5rd, 10th, 15th and 20th cycle. The soil samples were subjected to a freezing temperature of − 15 °C for 12 h in order to obtain complete frost penetration. After freezing, they were allowed to thaw at a room temperature of about 10 °C for 12 h before being subjected to the next cycle. An ultra-low temperature refrigerated storage tank (DW-HL290) manufactured by Zhongke Meiling Cryogenics Co., Ltd was used for this purpose. On the basis of no more than 0.5 °C error, the refrigerated storage tank has a temperature adjustment range of − 5 to − 87 °C and the temperature is visible. Saturated hydraulic conductivity. Saturated hydraulic conductivity (K s ) represents the upper limit of water migration rate for soil, and it is one of the crucial hydraulic properties of soil 41 . Soil permeability is closely related to its microstructure 42 . In this study, according to the code for geotechnical testing 43 , variable head osmotic apparatus (TST-55) was used for the SHC test. Before the test, the soil samples were placed in a vacuum saturator for 24 h to ensure that the soil samples reached the saturation state. The flow rate was measured under the condition of variable head. Considering the effect of temperature on water viscosity, the water temperature was monitored in real time during the test. The expression of K s is as follows 44 : where K s is the saturated hydraulic conductivity (cm s −1 ), a is the cross-sectional area of the standpipe (m 2 ), L is the length of the sample (m), A is the cross-sectional area of the sample (m 2 ), h 1 and h 2 are the initial and final water head with respect to the outflow (m) and t is the time for the hydraulic head difference to decrease from h 1 to h 2 .
After using formula (3) to figure out K s , use k 20 = η T η 20 to calculate K S at standard temperature, where η T is the dynamic viscosity coefficient of water at T °C and η 20 is the dynamic viscosity coefficient of water at 20 °C.

Mercury intrusion porosimetry.
MIP is a relatively straightforward method, which can be used to derive the values of other important characteristic parameters, including total porosity, average pore diameter, median pore diameter, and most probable pore diameter, and the measured pore diameters range from 7 to 350 μm.
In present study, we measured soil porosity and pore size distribution using MIP. Pre-lyophilized MIP samples were directly used for the MIP analysis, we used a fully automatic PM-33-18 mercury intrusion analyzer (Conta Instruments Company, USA). Mercury will be injected into the soil sample under a certain degree of pressure. Therefore, according to the pressure applied to mercury and the volume of mercury invasion, the pore distribution of the soil sample can be obtained through the Washburn equation 45 . The expression is as follows: www.nature.com/scientificreports/ where P is the applied pressure, σ is the surface tension coefficient of the immersed liquid, θ is the contact angle between liquid and solid materials, and r is the cylindrical pore radius. Among them, the values of σ and θ were 0.485 N m −1 and 140° in this study, respectively 46 . Soil pores can be classified into inactive pores, capillary pores and aerated pores according to their function and size. The porosity of the inactive pore is less than 2 μm, and the permeability is the worst. As the diameter increases, the permeability of pores becomes better 20 . In order to facilitate the pore size classification of saline-alkali soil in the Yellow River floodplain of Southwest Shandong, China. Our study based on the Shear pore size division theory 47 , and other scholars 48,49 , combining with the pore characteristics of the local saline-alkali soil (Table 4) to divide the pore size of saline-alkali soil in this area.

Surface fragmentation of soil samples after F-T cycles.
In this study, in order to better observe the macroscopic effects of freeze-thaw cycles on soil, we selected several representative experimental groups to demonstrate. Figure 3 shows soil samples with a water content of 13%, 1.48 g cm −3 and different freeze-thaw cycles. With the increase of the number of freeze-thaw cycles, we found that the surface fragmentation of soil samples from A1 to F1 was more and more obvious. By observing F1-F3 (Fig. 4), it is found that under the same number of freeze-thaw cycles and dry density, high water content can make the surface of soil samples more broken. However, the change of dry density of a single factor has no obvious influence on the surface macrophenomena of soil samples (Fig. 4). In order to highlight their differences more intuitively, we made Figs. 3 and 4 binary image respectively to obtain Figs. 5 and 6. In the selected circular area in the figure, black is the soil matrix and white is the broken surface. Based on Figs. 5 and 6, we found that the area variation of white area was basically consistent with the variation law of the crushing surface in Figs. 3 and 4. At the same time, the Image J software was used to calculate the fractal dimension ( Table 5). The fractal dimension of pores mainly reflects the irregularity and complexity of the contact boundary between pores and solid particles. The larger the value, the more complex the distribution of pores and the more irregular the boundary between pores and solid particles.  Saturated hydraulic conductivity after F-T cycles. Figure 7 summarizes the changes of K s of waterbearing remolded soil caused by the changes of F-T cycles. Under the three dry densities of 1.48 (Fig. 7a), 1.53 (Fig. 7b) and 1.58 g cm −3 (Fig. 7c), the soil K s increased with the increase of the number of cycles and reached the  www.nature.com/scientificreports/ threshold value at N = 15, and then the increasing pressure generated by the expanding ice crystals reduced the porosity inside the soil, K s showed a decreasing trend. Taking 1.48 g cm −3 as an example (Fig. 7a), when N = 15, the K s of the three water contents of soil samples increased by 260.95%, 270.95% and 283.33%, respectively. We found that after the first cycle, the increment of K s decreased with the change of the cycle. Nevertheless, the increment of the first cycle was the largest, which might be because the freeze-thaw cycle affected the development of soil pores. Figure 8 shows the changes of K s with water content of water-remolded soil. F-T cycle before (N = 0) and after freezing and thawing cycle (N > 0) of soil samples have significant differences. When N = 0, the K s was hardly affected by the water content changes. After freezing and thawing cycle, K s of soil had approximate linear relationship with the increase of water content, the F-T cycle test with us observed is consistent with the regular pattern of soil sample surface broken. However, the soil water content increased from 13 to 19%, and the increase of K s was relatively small. Under the three dry densities, the maximum elevation of K s was 31.68% (Fig. 8a), 18.91% (Fig. 8b) and 32.63% (Fig. 8c) respectively. At the same time, we found that the K s of soil samples would be enhanced as long as they experienced F-T cycles, which indicated that F-T cycles would have an impact on soil structure.
The relationship between K s and dry density of water-bearing remolded saline-alkali soil is shown in Fig. 9. When the soil dry density is high, the K s decreases significantly, but the attenuation rate decreases. The higher the dry density of the soil, the higher the degree of internal consolidation of the soil, which will affect the pores in the soil, resulting in compression or even closure, thus blocking the seepage channel. In addition, we found that compared with the F-T cycles of 20 times (N = 20) and 10 times (N = 10), with the increase of dry density, the saturated water content curves of both will present a crossover phenomenon, and this crossover phenomenon is more obvious with the increase of water content.
Quantitative analysis of micropore characteristics. Figure 10a summarizes the pore size distribution of soil samples subjected to F-T cycles under the same dry density and water content conditions, and the pore diameter corresponding to the peak point in the curve is the most likely pore diameter. The soil samples with 13% moisture content and 1.48 g cm −3 dry density had more ultra-micropores (< 0.05 μm) and micropores (0.05-2 μm) and less macropores (≥ 20 μm) after F-T cycles, but the results were higher than those of the soil  www.nature.com/scientificreports/ samples without freezing-thawing cycle (N = 0). We found that the peak value of pore size distribution curve shifts to the right and increases with the freeze-thaw cycle. After 10 F-T cycles, the curve presents a bimodal distribution, and the pore size distribution of the second peak is more than 10 μm. When the freeze-thaw cycle reaches 15 times (E1), the peak value reaches the highest, and the porosity reaches 32.39% (Fig. 10b). Although the peak value of F1 is lower than that of E1, it is still significantly higher than that of A1. Among them, the first peak of E1 was 19.40% and 14.93% less than that of D1 and F1. The second peak was 6.25% and 13.33% higher than D1 and F1, respectively. This indicates that the second peak in the pore diameter distribution curve plays a decisive role after 10 F-T cycles. The distribution curves of D1 and F1 approximately coincide, and the porosity difference between them is less than 3%, which is consistent with the result of saturated water conductivity. Figure 11 shows the influence of water content and dry density on pore size distribution of soil samples. With the increase of soil water content, the dominant pore size moved to the right, gradually changing from micropores (0.05-2 μm) to pores (2-10 μm), and the content increased. On the contrary, the distribution of dominant pore size is not affected by the change of dry density, and the content will decrease with the increase of soil dry density. When the dry density increases from 1.48 to 1.58 g cm −3 , the first peak and second peak of dominant pore size decrease by 36.75% and 13.25% respectively. These results indicate that the K s of soil is closely related to its microstructure, while water content and dry density also affect the microstructure of soil samples after F-T cycles to varying degrees, and thus affect the K s of soil.

Discussion
We speculate that the degradation of permeability of saline-alkali soil is closely related to its microstructure. Therefore, the saturated water conductivity test of soil was carried out, and PSD, XRD, MIP and other advanced technologies were adopted to study the microstructure of saline-alkali soil. The relationship between microstructure and permeability of saline-alkali soil after freeze-thaw cycle was analyzed in order to provide a theoretical basis for the restoration of microstructure of saline-alkali soil.
Particle composition and mineral composition of saline-alkali soil based on PSD and XRD. Soil particle composition is an important factor determining soil structure characteristics 50 . The distribution and texture of soil particle size directly affect the basic physical properties of soil, especially the water www.nature.com/scientificreports/ absorption capacity of soil 51 . The mineral composition of the soil is also one of the important factors affecting the soil permeability. The high proportion of clay and clay minerals will reduce the soil permeability due to their hydrophilicity 14,20,52 . Our research shows that clay (particle size < 0.002 mm) only accounts for 4.95% in saline-alkali soil, and the proportion of clay minerals is 9.55%, both of which account for less than 10% of the total proportion. A study found that the permeability of saline soil in the Yellow River basin decreases with the increase of clay particle content 53 , and added Yellow River sediment to change the particle size distribution to remediate the saline-alkali soil. In addition, high clay mineral content can reduce soil permeability, and soil weathering degree is low when clay mineral content is less than 15% 23 . The soil samples used in this study had low content of clay particles and clay minerals. Based on these, we focused on the microstructure of soil and explored the microstructure of soil. Thus, MIP analysis is very necessary in our research.
Saturated hydraulic conductivity of saline-alkali soil based on freeze-thaw cycling experiments. The freeze-thaw cycle is a process of energy input and output 54 , in which the transport of water and salt will lead to changes in soil grain size and structure 50 . The degree of soil weathering by freezing is related to the number of freeze-thaw cycles and water content. After the initial freeze-thaw cycles, the crushing of coarse particles and the aggregation of fine particles in the soil will lead to a significant increase in porosity. After more freeze-thaw cycles, the soil structure will gradually aggregate and the particle size distribution tends to be stable 50 . According to the local daily freezing and thawing times, we set a total of 6 freeze-thaw cycles (including N = 0). It was found that the soil permeability coefficient increased non-linearly with the increase of the number of cycles. The reason is that ice lenses gradually form during freezing, and ice melts in the frozen soil during thawing, creating cavities that act as channels and increase soil permeability 55 . When the number of F-T cycles reaches 15 times, it reaches the peak value. Compared with N = 0, the K s increases by 209.77%.
The initial porosity of soil plays a decisive role before F-T cycles 29 , and the initial dry density of soil affects the initial porosity. Water to ice produces a volume change of about 9%, which affects the volume expansion of the pores during freezing 56 . Therefore, according to the standard compaction test, we determined the maximum dry density of the saline-alkali soil of 1.58 g cm −3 and the optimal water content of 16%, and the floating value can meet the conditions of the test area coverage. After the F-T cycles, soil is coupled by salt heave and frost heave, and soil pores change, resulting in more internal seepage channels, which means that soil porosity is the main influencing factor leading to the change of saturated water content. Freeze-thaw soils with high water www.nature.com/scientificreports/ content produce more expansive ice, which breaks the bonds between particles, in contrast, when soil water content is low, ice crystals only grow in soil pores 57,58 . This is consistent with our research results, the K s of soil is significantly affected by changes in dry density and water content. Among them, compared with B7 and B9, C7 and C9, D7 and D9, E7 and E9, and F7 and F9, the K s decreased by 42.25%, 53.52%, 57.36%, 49.80% and 41.15% respectively. Compared with B7 and B1, C7 and C1, D7 and D1, E7 and E1, and F7 and F1, the K s increased by 17.0%, 22.57%, 7.83%, 9.26% and 24.07% respectively. In addition, it has been shown that external stress of up  www.nature.com/scientificreports/ to 2 MPa can be generated during soil freezing 53 . Under the action of contact stress, the joints and frozen water in the soil will continue to grow and expand. As the circulation goes on, the growing ice crystals will produce pressure on the surface of the adjacent matrix and reduce the internal porosity of the soil 59 . Therefore, when the F-T cycle reaches 20 times, the K s of the soil decreases instead. At the same time, the formation of ice-penetrating crystals increases the pore water pressure, compressing the soil, which in turn increases the pore water pressure. During the F-T cycle test, the soil was always inside the constraint ring, and with the increase of soil internal pressure, the constraint of the ring on the soil was strengthened. Considering that high water content promotes frost heave, this explains the crossover in the saturation conductivity curve.
Pore size distribution of saline-alkali soil based on MIP after freeze-thaw cycles. Soil structure is an important factor affecting water conductivity [60][61][62] . The poor structure of saline-alkali soil is closely related to the low water conductivity and permeability of soil 63 , these are consistent with our experimental results. Soil microstructure is closely related to its K s and is strongly affected by freeze-thaw cycles. In the test, A1, A2 and A3, as soil samples without freeze-thaw cycles, have porosity of only 6.19-11.51%. The low porosity resulted in the low saturated water conductivity of the soil sample, which was only 1.48 × 10 -5 cm s −1 . This makes it difficult for agricultural irrigation water to penetrate the soil, and the remaining water forms surface runoff, which accelerates soil nutrient loss. In addition, the dense soil structure and low porosity promoted soil capillarity in the floodplain of southwestern Shandong Province, and the salt remained on the surface with the evaporation of water after the rise of groundwater, which was extremely unfavorable to the development of agricultural production.
The MIP results showed that there were a lot of ultra-micropores in the soil samples in this area. However, more ultra-micropores in the soil were not conducive to the remediation of saline-alkali soil 49 . The F-T cycle can reduce the content of ultra-micropores and micropores in soil samples. After 10 F-T cycles, a new peak value appears at 10 μm, indicating that freeze-thaw cycles have a strong effect on soil macropores. When the number of freeze-thaw cycles reaches the threshold (N = 15), the pore water and ice crystals undergoing the freezing and thawing process applied pressure on the surrounding soil and ring knife. In addition, ice-permeable crystals formed during freezing and thawing, which can exacerbate the effects of pressure. Pressure caused particles to condense, which made the pore volume and diameter smaller at N = 20. These also caused the aperture distribution curves of D1 and F1 to coincide.
Water content and dry density have great influences on F-T cycle. The water transport process becomes more complex under freeze-thaw cycles, and the complexity increases with the increase of water content. In the soil sample, the water was almost completely frozen into ice during the freezing process, and the soil particles were displaced due to volume expansion, and the soil particles were not completely restored to the original position during thawing 64 , resulting in the deviation of the pore content and the increase of soil porosity. Therefore, we can determine that the increase of water content will shift the peak value of the curve to the right and increase the content of the intrusion volume, which is consistent with the results obtained from the saturated water conductivity test. In addition, the increase of dry density will strengthen the cementation between soil particles, and reduce the porosity in the soil. The growth of ice crystals and the formation of ice-penetrating crystals in the soil are further inhibited, the intrusion volume of pores is correspondingly reduced, and the saturated water conductivity of the soil is reduced.
Effect of freeze-thaw cycles on salt discharge in saline-alkali soil. Under the action of freezingthawing cycle, the saline-alkali soil was affected by the coupling of salt heaving and frost heaving, pore water and ice crystals, then pore channels were generated in the soil, which facilitated the flow of water and the frost heav- Figure 11. Pore-size distribution cures with different water content and dry density or soils at the study site in the floodplain of southwestern Shandong Province: (a) The dry density is 1.48 g cm −3 , pore-size distribution cures with different water contents. (b) The water content is 19%, pore-size distribution cures with different dry densities. www.nature.com/scientificreports/ ing of ice. Therefore, in the early stage of freeze-thaw cycle, soil particles were broken, and the pore diameter and volume of soil became larger, correspondingly, K s would increase. After the freezing-thawing cycle reached 10 times, the impact of freezing-thawing cycle intensified, leading to the crushing of large soil particles. The second peak of pore size distribution curve appeared. When N = 15, the pore volume reached the maximum, and then K s reached the maximum. When N > 15, the pressure generated by ice crystals increase continuously, and the counterforce generated between the ice crystals and the soil matrix began to compress the soil interior. The internal pore channels of the soil were compressed, the pore volume and pore diameter correspondingly decreased, and K s decreased. In addition, high water content could enhance the effect of ice crystals on soil, thereby increasing pore distribution. The higher dry density resulted in a denser interior of the soil sample, which reduced the flow of water through the channels. In recent years, some scholars have tried to change the microstructure of saline-alkali soil with different ameliorants 65 . On this basis, we try to find a new ameliorating measure of saline-alkali soil. The former proposed that high soil porosity would improve its permeability 66 , and our research supports this view. The results show that the appropriate freeze-thaw cycle can greatly improve the pore distribution, soil porosity and saturated water conductivity of saline-alkali soil in the southwest of Shandong Province. This is also due to the weak internal cohesion of natural saline-alkali soil, and irreversible plastic deformation will occur after several times of frost heave, which will not be affected by the change of salt content. In a certain range, the increase of water content can enhance the effect of freeze-thaw cycle. Considering the natural water content of 14.07% and the optimal water content of 16% in the experimental area, this idea is feasible.
The extreme temperature in winter reaches − 16.5 °C in the Yellow Flood area of Southwest Shandong Province, which is a seasonal frozen soil area. Soil samples were taken from the topsoil layer (0-25 cm), which is included within the freezing depth of the soil in the area. Therefore, the F-T cycles can significantly improve the pore distribution and microstructure of crop surface soil. It is well known that after soil moisture evaporates, most of the salt and alkali remain in the surface of the soil 51 , which requires that we need to broaden the channels of water circulation in the soil and let the salt go with the water. Therefore, we can take the freeze-thaw cycle as the basis for the salt discharge in the saline-alkali soil project in this area. Based on the above research, in order to achieve the effect of improving the local soil microstructure and considering the influence of dry density on the pore distribution, we suggest that local farmers should properly plough the cultivated land in winter. Besides, burette can be used to increase the soil water content, in order to achieve the effect of promoting the freezing and thawing cycle.

Conclusions
The existence of saline-alkali soil seriously endangers the agricultural production and development in the floodplain of southwest Shandong Province. In this experiment, MIP and SHC test were used to explore the relationship between soil microstructure characteristics and K s under F-T cycles. The main reason of low permeability of saline-alkali soil is not the influence of grain size distribution and mineral composition. The low permeability of surface saline-alkali soil is closely related to its microstructure, and its pores are mainly composed of ultra-micropores and micropores. Under the action of F-T cycles, soil pores transition from micropores and ultra-micropores to small pores. When N = 10, mesopore particles begin to be broken, and such fragmentation is the most obvious when N = 15, which is consistent with the law presented by soil saturated hydraulic conductivity curve. High water content can improve the crushing effect of freeze-thaw cycle on soil particles, and then improve soil permeability, while the increase of dry density will play a certain inhibiting role. The results show that the permeability of saline-alkali soil is closely related to its microstructure, and the freeze-thaw cycle can improve the microstructure of saline-alkali soil to further improve its permeability, which provides a certain reference direction for the restoration of saline-alkali soil. Although the relationship between the microstructure of saline-alkali soil and the saturated hydraulic conductivity has been comprehensively analyzed by various methods in this experiment, there are still some limitations. In future studies, we need to consider the restoration effects of F-T cycles on other soil textures, and consider the coupling effects of F-T cycles on salt content, temperature and other factors in combination with soil microstructure.

Data availability
The data used to support the findings of this study are included within the article.