E ﬀ ect of Soluble Salt Loss via Spring Water on Irrigation-Induced Landslide Deformation

: Landslide exposes the previously blocked groundwater discharge. High concentrations of soluble salt form salt sinters that can be observed near discharge passages. Based on existing laboratory investigation results of soil leaching and shearing reported in the literature, the e ﬀ ect of the soluble salt loss via spring water on irrigation-induced landslide deformation was studied under large-scale conditions. During our ﬁeld investigation of landslides in the Heitai terrace of the Yellow River’s upper reaches in Gansu Province, China, 35 spring outlets were found, and the Heitai terrace was divided into ﬁve subareas, based on the di ﬀ erence in spring ﬂow. Deformation data for the terrace were obtained by small baseline subset technology (SBAS-InSAR). These data were analyzed in combination with the amount of soluble salt loss, to explore the correlation between the deformation of the landslide and the soluble salt loss in the loess irrigation area. The results showed that the cumulative deformation and the loss of soluble salt were increasing continuously in the terrace. Although the increasing intensity of each subarea was di ﬀ erent, the changing intensity of the two during the corresponding monitoring period was highly consistent. The statistical analysis revealed a strong positive correlation between the accumulated loss of soluble salt via spring water and the accumulated displacement of the terrace edge ( p < 0.01). After the slope k between the two was tested by the Grubbs test and t -test, the k was no abnormality ( α = 0.05) and di ﬀ erence (Sig > 0.05), further providing the basis for conﬁrming the existence of this positive correlation. When the loss of soluble salt in rock and soil increased gradually, the accumulated deformation of the terrace edge also increased continuously. The ﬁndings of this study are of great signiﬁcance for understanding the formation mechanism of landslides and the identifying landslide revival in irrigation areas of the Loess Plateau. of spring, it is changing slowly, and ( b ) the two or more branch seepage paths caused by the collapse deformation of soil. At this location, evidence of soil collapse can be clearly found, which is common on landslide mass. Y.Z., S.Z., J.M. and Y.Y.; resources, R.Z. and X.M.; data curation, Z.Z.; writing of the original draft, Z.Z.; review and editing, R.Z., Z.Z., X.M., Y.Z., S.Z., J.M. and Y.Y.; supervision, R.Z.; project administration, R.Z. and X.M.; funding R.Z.


Introduction
The area covered by loess in China is approximately 631,000 km 2 , which accounts for 6.6% of the country's total area [1]. Irrigation-induced landslides occur in loess areas, and the Heifangtai terrace in Yongjing County, northwest China, is typical irrigation-induced landslide area. The types of landslides include loess-bedrock planar slide, loess flow, loess flowslide, and loess slide [2]. Several scholars have reported that rising groundwater levels aggravate the saturated area of loess and that the matrix suction inside a slope decreases under saturated water conditions [3,4]. This causes the affected The interferometric synthetic aperture radar (InSAR) technology extracts surface deformation data through the relationship between the radar wave phase difference and spatial distance in two SAR observations [35]. This technology is suitable for large-scale surface deformation monitoring and has the advantages of having high precision and all-time, all-weather usage [36]. Gabriel et al. [37] proposed the differential InSAR (DInSAR), which contributed greatly to the development of time series InSAR technologies such as the persistent scatter InSAR (PSInSAR) [38] and small baseline subsets (SBAS-InSAR) [39]. Time series InSAR technology can obtain more accurate surface deformation data; it has been widely used in landslide and ground instability monitoring, and it has achieved remarkable results [40][41][42][43]. In this study, the cumulative displacement data of the Heitai terrace was obtained by using SBAS-InSAR technology. There are two main reasons: On the one hand, SBAS-InSAR technology generates differential interferograms in spatial, temporal baselines and small Doppler center frequency difference data pairs, which solves the spatial incoherence problem caused by the long spatial baseline of conventional InSAR and increases the time sampling frequency. On the other hand, SBAS algorithm can identify the coherent point target by setting the temporal baseline and spatial baseline of generating interferograms, so as to reduce the orbit error and incoherence noise.
In summary, so far, too many scholars have analyzed the effect of soluble salt loss and pore

Study Area
The Heifangtai terrace is the fourth terrace of the Yellow River and is situated at the junction of the Yellow River (to the south) and Huangshui River (to the north), approximately 42 km to the west of Lanzhou City. The surface of the terrace is almost flat, but it is cut by many gullies, including the Yehu gully, Shancheng gully, and Moshi gully. Taking the Hulang gully (the first tributary of the Yellow River) as the boundary, the eastern area is the Heitai terrace, with an area of~12.0 km 2 , and the western area is the Fantai terrace, with an area of~1.7 km 2 . The area is located at an elevation of approximately 1730 m and is 130 m higher than the present riverbed of the Yellow River. The unique tectonic landform makes the Heifangtai terrace an island; there is almost no hydraulic connection with the outside world [44,45]. The climate is temperate and semi-arid, with an average annual temperature of 8.4 • C, average annual precipitation of 310 mm, and an evaporation capacity of approximately Water 2020, 12, 2889 4 of 17 1689 mm. Approximately 70-80% of the annual rainfall is concentrated during the period from July to September [46]. It is difficult to recharge groundwater effectively; therefore, the nature of Heifangtai is a dry platform. In the early 1960s, Heifangtai was confirmed as a settlement site, and irrigation became a necessary measure to ensure immigrants' agricultural production. Since 1964, 22 irrigation projects have been built. The irrigated area is approximately 7.55 km 2 , which is 55.13% of the total area. The volume of annual irrigation increased to~8.8 × 10 5 m 3 in the 1980s, from~7.22 × 10 5 m 3 [7], and subsequently increased to approximately 6.0 × 10 6 m 3 in recent years. After over 40 years of flooding irrigation, the groundwater level rose by at least 20 m by the end of 2012 [47]. The groundwater level is currently rising at a rate of 0.18 m/year [48], and massive loess landslides have been induced around the Heifangtai terrace [11,49]. According to incomplete statistics, more than 140 loess landslides have occurred at the edge of the terrace, whereby 3-5 landslides currently happen each year.
The lithological profile of the Heifangtai terrace (from the bottom to the top) is mainly composed of four units. The basement is comprised of sand-mudstone off the Hekou Group (Lower Cretaceous). A clay layer (~4-17 m) is situated above a gravel layer (~2-5 m), which both formed during the Middle Pleistocene. The upper loess stratum (~25-50 m) is composed of Lishi loess and Malan loess of the Middle Pleistocene and Late Pleistocene, respectively [7]. We have also provided some basic physical properties of the loess and silty clay in the study area (Table 1). The rising groundwater level has increased the number of discharge outlets of the platform margin spring. A field investigation of the Heitai terrace revealed a total of 35 spring discharge outlets, which are derived from loess pores, gravel layer pores, and bedrock fractures. Based on the fact that spring flow is closely related to the geographical location of the spring discharge outlet distribution, the study area was divided into five subareas (zones I-V; Figure 1).

Spring Water Sampling and Spring Flow Measurements
Based on the terrace edge field investigation, it was found that the drainage system around the terrace was composed of 35 descending springs. Because the spring water flows out from loess pore, sand-gravel interlayer, and bedrock fissure, the soluble salt content has differences in these three types of spring water. According to the actual terrain of the spring dew point distribution, 15 spring samples of three water types were collected at relatively flat positions of the channel in the five subareas at each period ( Figure 1). The sampling tools include 500 mL pure water bottle, label paper, marker, and absorbent paper. Before sampling, the container was rinsed with spring water, at least 2 or 3 times. Two bottles were sampled for each spring water dew point; we wiped off the spring water on the bottle wall with absorbent paper, put a label paper on it, and sealed the containers after sampling.
The spring flow was determined, indirectly, by using Equation (1): where V represents the flow velocity (m/s), B represents the channel width (cm), and h represents the water depth (cm). Parameters B, h, and V were measured as follows. Firstly, the width and depth of the flat bottom of the spring channel were measured. Secondly, when the water surface was gentle, the water-surface velocity was measured 10 times, using a handheld surface velocimeter. Thirdly, considering that the water depth is of a centimeter-level, the average water-surface velocity was used to replace the water-flow velocity.

Chemical Component Analysis of Samples
The degree of mineralization is expressed by the total salt content of water and was calculated in this study by using the additive anion and cation method. The salt ions of the rock and soil on the Heitai terrace are mainly composed of NO 3− , Cl − , SO 2− , HCO 3− , NH 4+ , Ca 2+ , Mg 2+ , Na + , and K + . The lixiviation of rock and soil by infiltrating irrigation water is a soluble salt source to spring water. Therefore, the spring water samples contained the same ions as the rock and soil. Table 2 shows the methods used to determine the various ions in the spring water samples.

Statistical Analysis
The normal distribution was tested by Shapiro-Wilk (W test) and Quantile-Quantile (Q-Q) Plot test methods before the statistical analysis of the data. The bivariate correlation test method was used to judge the correlation between each group of variables, and the Pearson' r correlation coefficient R was used to evaluate the degree of correlation between the two variables, where R 2 represents the goodness of fit of the regression equation, and the significance level was set was α = 0.05. Grubbs test was used to judge the abnormal value, and t-test was used to test the difference of value.

Variation Characteristics of the Soluble Salt Loss
There are two main sources of soluble salt to spring water in the study area. The first is the leaching of substances from rock and soil, which occurs due to the groundwater flow while irrigation water infiltrates. The second potential source is the salt contained within irrigation water itself. The laboratory test results of the chemical composition of the spring water samples showed that soluble salt was composed of easily soluble salt (mainly NaCl and KCl, followed by Na SO), medium soluble salt (mainly CaSO and MgSO), and insoluble salt (CaCO). The loss of easily soluble salt is the most serious and is followed by the medium soluble salt and then insoluble salt. The loss of soluble salt from soil changes the soil's microstructure, which then changes the soil stability. By calculating a mass balance of the soluble salt from the study area, it is possible to determine the main source of salt discharge, which can then provide a basis to explore the terrace's stability.

Statistical Analysis
The normal distribution was tested by Shapiro-Wilk (W test) and Quantile-Quantile (Q-Q) Plot test methods before the statistical analysis of the data. The bivariate correlation test method was used to judge the correlation between each group of variables, and the Pearson' r correlation coefficient R was used to evaluate the degree of correlation between the two variables, where R 2 represents the goodness of fit of the regression equation, and the significance level was set was α = 0.05. Grubbs test was used to judge the abnormal value, and t-test was used to test the difference of value.

Variation Characteristics of the Soluble Salt Loss
There are two main sources of soluble salt to spring water in the study area. The first is the leaching of substances from rock and soil, which occurs due to the groundwater flow while irrigation water infiltrates. The second potential source is the salt contained within irrigation water itself. The laboratory test results of the chemical composition of the spring water samples showed that soluble salt was composed of easily soluble salt (mainly NaCl and KCl, followed by Na SO), medium soluble salt (mainly CaSO and MgSO), and insoluble salt (CaCO). The loss of easily soluble salt is the most serious and is followed by the medium soluble salt and then insoluble salt. The loss of soluble salt from soil changes the soil's microstructure, which then changes the soil stability. By calculating a mass balance of the soluble salt from the study area, it is possible to determine the main source of salt discharge, which can then provide a basis to explore the terrace's stability.
In recent years, due to the variation in planting patterns and plant varieties, the volume of annual irrigation water applied to Heitai terrace has ranged from~590.91 × 10 4 m 3 to~779.00 × 10 4 m 3 (the Heifangtai Water Management Office, Yongjing County, Linxia Hui Autonomous Prefecture, Gansu Province). The mineralization degree of irrigation water was calculated as 0.236 g/L, and the total amount of salt that was brought into the soil on Heitai terrace due to irrigation was estimated to be between~0.139 × 10 4 t to~0.184 × 10 4 t. The annual spring flow was estimated to range from 91.46 × 10 4 m 3 to~201.36 × 10 4 m 3 . The mineralization degree of spring water was calculated to be 15-29 g/L, and the amount of salt discharged via the spring water was between~1.372 × 10 4 t tõ 5.839 × 10 4 t, which is~7-42 times the amount of salt brought in. According to the mass balance results of the annual amount of soluble salt, it was concluded that the loss of soluble salt via spring water occurred mainly through the lixiviation of soluble salt in soil following the infiltration of irrigation water. An increase in the irrigation intensity and the accumulation of irrigation time on Heitai terrace means that the total amount of salt loss continually increases. In order to better understand the spatial and temporal distributions of soluble salt loss on Heitai terrace, the spring flow and mineralization degree in zones I, II, III, IV, and V were measured in the field and in the laboratory, respectively, from September 2016 to April 2017. The results are shown in Tables 4 and 5. Figure 3 shows the differences in the soluble salt loss of each subarea. Although the cumulative loss increased gradually over time for all subareas, there was a clear difference in the variation of the accumulation loss of soluble salt in each subarea. The greatest variation occurred in zones V and IV, followed by zones III and II. During the monitoring period, the cumulative loss of soluble salt in zone I was relatively low and only about 1/23 of that of zones II and III, and 1/45 that of zones IV and V. We also find the loss of soluble salt increased significantly in February 2017; this is because the temperature of Heitai is lower in December, January, and early February, while its thawing period in the middle of February. Therefore, the dew point of frozen spring water melts in the process of temperature rising, and a large amount of water is discharged, leading to the increase of soluble salt loss. We can also know that this phenomenon takes place less on sunny slopes.     Figure 4 presents the displacement of Heitai terrace along the radar line-of-sight (LOS), which was obtained by using SBAS technology, from September 2016 to April 2017. According to the standard deviation of the total study area's coherent points, we are setting the relative stability threshold of displacement range-6 to 6 mm. A negative value indicates that the deformation moves away from the sensor along the LOS direction, and a positive value indicates that the deformation moves along the LOS direction toward the sensor. It can be seen that the accumulated displacement change of the terrace was mainly concentrated between 6 and 6 mm, but the displacement exhibited obvious spatiotemporal distribution characteristics at the terrace edge. With increasing time, the accumulated displacement of zones II, III, IV, and V gradually increased to within 18 mm, and the local cumulative displacement was >18 mm, which indicates intensive deformation characteristics. The cumulative displacement of zone I was within 12 mm during the monitoring period and exhibited no obvious change.

Correlation Analysis of the Cumulative Salt Loss and Cumulative Displacement
The cumulative loss of soluble salt via spring water presented the same spatial and temporal distribution characteristics as the accumulated displacement of the terrace edge. The SPSS software was used to verify the normal distribution hypothesis of the experimental data. The results of the Q-Q Plot test basically showed a straight line, and W test results were sig >0.05. Therefore, we consider that the data meet the normal distribution. We attempted to use the bivariate correlation method to explore the correlation between the two. The relationship between the cumulative loss of soluble salt via the spring water was positively correlated with the accumulated displacement of the terrace in zones II, III, IV, and V. The two variables exhibited a good linear relationship in zones II, III, IV, and V (R 2 = 0.865, R 2 = 0.936, R 2 = 0.908, and R 2 = 0.973 (p < 0.01)) ( Figure 5). Although we do zoning research based on the difference of spring flow in different geographical locations. We had to admit that, at the boundaries of each subarea, the spring water might come from the groundwater in two adjacent areas. This led to the "false" of soluble salt loss in this area.
However, there was no linear relationship in zone I, but there were positive cumulative displacements. The positive cumulative displacement that appears in the Zone I caused by the orbital SAR image that does not express the deformation on the downslope of the region well. The positive cumulative displacement in a certain period of time also indicates that there is a slight change in the slope itself during the imaging period. For the total study area, the displacement change in a certain period of time cannot affect the evolution trend of the entire time series, which is in the acceptable range. The accumulated displacements of the other four areas are larger than the set stability threshold. Combined with field verification, it is believed that the terrace in these subareas has obvious signs of deformation.
Using the Grubbs test method, the outlier of slope k between the accumulated loss of soluble salt and the cumulative displacement in zones II, III, IV, and V was tested. At a detection level of α = 0.05, the calculated G < G (4) showed that there was no abnormality for slope k in the study subareas. According to the accumulated loss of soluble salt and displacement of zones II, III, IV, and V in the monitoring period, the central scatter diagram (Figure 6) was obtained. In Figure 6, there were obvious differences in the distribution characteristics of data points in each subarea. However, a large number of data points were of a banded distribution, and there was a positive linear correlation between them (R 2 = 0.600, p < 0.01). Taking the slope value as the detection value and expressing it as k, a t-test was used to distinguish the difference between k and k. First, the results showed that the concomitant probability of the t-test was 0.288, and the upper and lower limits of the 0.95 confidence level were −0.104 and 0.044. Second, the concomitant probability of the t-test was greater than the significance level of 0.05. The results indicate that there was no significant difference between slope k and k in each study subarea, further confirming the strong positive correlation between the cumulative loss of soluble salt and displacement.
However, there was no linear relationship in zone Ⅰ, but there were positive cumulative displacements. The positive cumulative displacement that appears in the Zone I caused by the orbital SAR image that does not express the deformation on the downslope of the region well. The positive cumulative displacement in a certain period of time also indicates that there is a slight change in the slope itself during the imaging period. For the total study area, the displacement change in a certain period of time cannot affect the evolution trend of the entire time series, which is in the acceptable range. The accumulated displacements of the other four areas are larger than the set stability threshold. Combined with field verification, it is believed that the terrace in these subareas has obvious signs of deformation. Using the Grubbs test method, the outlier of slope k between the accumulated loss of soluble salt and the cumulative displacement in zones II, III, IV, and V was tested. At a detection level of α = 0.05, the calculated G < G (4) showed that there was no abnormality for slope k in the study subareas. According to the accumulated loss of soluble salt and displacement of zones II, III, IV, and V in the monitoring period, the central scatter diagram (Figure 6) was obtained. In Figure 6, there were obvious differences in the distribution characteristics of data points in each subarea. However, a large number of data points were of a banded distribution, and there was a positive linear correlation between them (R 2 = 0.600, p < 0.01). Taking the slope value as the detection value and expressing it as k, a t-test was used to distinguish the difference between k and k. First, the results showed that the concomitant probability of the t-test was 0.288, and the upper and lower limits of the 0.95 confidence level were −0.104 and 0.044. Second, the concomitant probability of the t-test was greater than the significance level of 0.05. The results indicate that there was no significant difference between slope k and k in each study subarea, further confirming the strong positive correlation between the cumulative loss of soluble salt and displacement.

Discussion
Subsidence and deformation were very common in the Loess irrigation area. As mentioned, it was confirmed that the soluble salt loss in rock and soil can affect the shear strength of its, resulting in subsidence and deformation of soil [51,52]. In this study, SBAS-InSAR technology was used to

Discussion
Subsidence and deformation were very common in the Loess irrigation area. As mentioned, it was confirmed that the soluble salt loss in rock and soil can affect the shear strength of its, resulting in subsidence and deformation of soil [51,52]. In this study, SBAS-InSAR technology was used to monitor the long-term ground deformation in the Loess irrigation area, i.e., Heifangtai, and the soluble salt content lost in the soil was obtained in the corresponding period. Through statistical analysis, it was found that there was a strong positive correlation between the two, proving that the change of soluble salt content of loess had an effect on its shear strength to a certain extent. However, the statistical analysis results also show that no positive correlation was found between the two variables in zone I. We offer two possible explanations for this observation. On the one hand, it can be seen from the elevation distribution of groundwater on Heitai terrace (Figure 7) that the groundwater level was relatively high in the western and southwestern areas. Although groundwater domes were observed in the western area of the terrace, there was minimal groundwater flow into zone I, which is similar to the findings of Peng et al. [49]. On the other hand, there was a serious drying-up of springs in zone I during the monitoring period. According to Table 4, the minimum and maximum spring flows of the other zones were approximately 15.7 and 91 times that of zone I (average of 34 times), and the loss of soluble salt in zones I, II, III, IV, and V was 2.01 × 10 8 g, 44.13 × 10 8 g, 45.84 × 10 8 g, 80.18 × 10 8 g, and 92.53 × 10 8 g, respectively ( Figure 3). Based on these two aspects, we believe that the total amount of soluble salt brought out by spring water was very small in zone I; hence, more salt was retained in the zone's pore solution. The irrigation frequency and volume of the terrace also increased due to the reform of the planting patterns and plant varieties; however, the salt supplied by irrigation water to the Heitai terrace is still weak. On the contrary, the infiltration of irrigation water has a serious effect on the dissolution and lixiviation of soluble salt in rock and soil. This changes the shear strength and other parameters of loess, which then affect the stability of the landslide mass [18,53,54]. In addition, these changes may cause the formation of collapsible pits and cracks, which can become dominant channels for the rapid infiltration of water. This accelerates the rising groundwater level and leads to the reactivation of landslide mass and the regression damage of terrace edges [6,11,55,56].
Moreover, we found differences in the mineralization degree of spring water in each study area that also fluctuated during the monitoring period (Table 4, Figure 8). Li et al. [28] analyzed the mechanism of loess's strength weakening under groundwater leaching, by means of ring shear, ion chromatograph, and zeta potentiometer. With the increased of the soil immersion time, the soluble salt in soil was dissolved continuously, leading to the destruction of soil structure, and ion concentration in the soil pore water increases. The soil shear strength and shear strength indexes, ϕ', appeared as "spoon-shape" changes in this process. Wen and He [29] reported that the physicochemical interaction between water and salt was the dissolution and leaching of soluble salt, which greatly affected the c' or ϕ' of red mudstone. After the soluble salt in the red mudstone was leached many times, the downward trend of the residual shear strength became very weak. Zhang et al. [27] had studied the influence of the pore water chemistry on soil shear behavior in Heitai, including the soil shear behavior with NaCl concentration in pore water and the shear behavior of soil by desalination effect of soil. The results showed that the shear strength increased with increasing NaCl concentration, and then decreased with further increase of NaCl concentration; the long-term irrigation can decrease the NaCl concentration and then can lower the soil shear strength in Heitai terrace. Other scholars had also confirmed that the weakening of the shear characteristics of soil is mainly caused by the change of the parameter c' or ϕ' caused by the dissolution loss of soluble salt through laboratory experiments, and they believe that this weakening tends to zero with the increase of leaching times [25,30]. Therefore, the changing trend of the soluble salt content of spring water was analyzed. The fluctuation in the mineralization degree is related to the degree of leaching. A landslide mass begins to deform slowly due to the weakening of the soil shear strength until the settlement deformation changes the seepage path; hence, a soil that has experienced little dissolution again suffers from the main leaching of groundwater. The soluble salt is brought out from the terrace, which again increases the mineralization degree. This continuous cycle eventually leads to the landslide deformation instability or resurrection, which has been extensively verified at the landslide site ( Figure 9). There are also relevant reports that the continuous loss of salt ions caused by long-term seepage leaching further aggravates soil structure degradation and strength. This eventually leads to the large deformation of the loess surface, which may cause slope instability [18,29,57,58]. The results of the present study, as well as the laboratory test results of other investigations, demonstrate that the mechanical behavior of rock and soil is affected by the chemical composition of the pore fluid to a certain extent. However, it is necessary to make a more quantitative assessment of these behaviors, to gain a deeper understanding of the potential processes, in order to evaluate their role in the occurrence and movement of loess landslides. At the same time, the water content of the soil will inevitably change during the leaching process of soluble salt in the Loess irrigation area, but this is only limited to the shallow unsaturated area of loess. Although the water content of rock and soil can also affect its shear behavior [6,24,25], we did not quantify the impact of water-content change on the deformation of shallow loess, and this is a deficiency of this work.
to gain a deeper understanding of the potential processes, in order to evaluate their role in the occurrence and movement of loess landslides. At the same time, the water content of the soil will inevitably change during the leaching process of soluble salt in the Loess irrigation area, but this is only limited to the shallow unsaturated area of loess. Although the water content of rock and soil can also affect its shear behavior [6,24,25], we did not quantify the impact of water-content change on the deformation of shallow loess, and this is a deficiency of this work.  . Site of landslide mass settlement deformation: (a) the main path of spring, it is changing slowly, and (b) the two or more branch seepage paths caused by the collapse deformation of soil. At this location, evidence of soil collapse can be clearly found, which is common on landslide mass.

Conclusions
As mentioned, the shear strength can be used as an important evaluation index of landslide stability, which is closely related to factors, such as the soluble salt content or pore salt solution concentration in rock and soil. The study aimed to explore a new method to quantify the relationship between the loss of soluble salt from rock and soil and its deformation behavior. This was investigated via a field investigation of the spring flow on Heitai terrace, laboratory determination of Figure 9. Site of landslide mass settlement deformation: (a) the main path of spring, it is changing slowly, and (b) the two or more branch seepage paths caused by the collapse deformation of soil. At this location, evidence of soil collapse can be clearly found, which is common on landslide mass.

Conclusions
As mentioned, the shear strength can be used as an important evaluation index of landslide stability, which is closely related to factors, such as the soluble salt content or pore salt solution concentration in rock and soil. The study aimed to explore a new method to quantify the relationship between the loss of soluble salt from rock and soil and its deformation behavior. This was investigated via a field investigation of the spring flow on Heitai terrace, laboratory determination of spring water's mineralization degree, and analysis of surface deformation data obtained by using SBAS technology. Data analysis and interpretation led to the following conclusions.
The main source of salt in the groundwater of the Heitai terrace is the leaching of soluble salts from rock and soil by irrigation water. The loss of soluble salt is concentrated in the northern, northeastern, eastern, southeastern, and southern areas of the terrace. This is consistent with the spatial distribution of intense deformation in the Heitai terrace area, and there is a positive correlation between them.
According to the salt ion concentration change in spring water, the loss of soluble salt via spring water was estimated. Along with the results of a large number of landslide field investigations, our findings suggest that the landslide mass began to deform slowly due to the weakening of the soil shear strength. We infer that, until the settlement deformation changes the seepage path, the soil with a low degree of lixiviation again suffers from groundwater's main leaching. Hence, the concentration of pore salt ions decreases in soil, and soluble salt is brought out from the terrace. This continuous cycle eventually leads to landslide deformation instability or resurrection.
Based on the positive correlation between the two, the relative deformation of slope can be recognized by detecting the loss of soluble salt in water during the period. This will provide a new idea for the early identification of landslides in the future.