Spatial Variability of Soil Moisture in Mining Subsidence Area of Northwest China

of soil moisture in the - year subsidence area and the 2 - year subsidence area increased, the variability the of depth. Abstract 59 The current research only investigate the impact of coal mining on deep soil moisture 60 from the perspective of the absolute value of soil moisture. This study applied the 61 combined method of classical statistics and multi - dimensional geo - statistics to 62 analyze the changes of soil moisture of time and space from 0 - 10m in the mining face 63 of Nalin River No.2 Mine in Northwest China from the perspective of spatial 64 variability. The results of the study showed that in time distribution, on the whole, the 65 soil moisture in the partial areas of the 1 - year and the 2 - year subsidence area was 66 lower than that in the control area, and the variability increased, but as the subsidence 67 entered a stable period, the degree of variability decreased; vertically observed, in 68 space distribution, the 0 - 10m soil moisture in the control area had obvious distribution 69 rules with low spatial variability. However, the spatial variability of soil moisture in 70 the 1 - year subsidence area and the 2 - year subsidence area increased, and the 71 variability showed a trend of increasing continuously with the increase of depth. 72 component analysis, it was found that the change of soil texture 73 caused by coal mining subsidence, the change of microstructure of soil pores caused 74 by geotechnical deformation, as well as the preferential flow caused by changes in 75 groundwater level were the main reasons for the increasing spatial variability of soil 76 moisture. This study revealed the principals of spatial variability of soil moisture in 77 coal mining subsidence areas in Northwest China, which can provide a scientific basis 78 for the restoration of mining areas.

4 than that in the virgin area. Moreover, the soil water content around the ground 130 fractures decreased significantly, with a significant downward trend in the depth of 0 131 -90 cm, ranging from 9.27% to 15.47%; Wu, Tian, and Tang, (2019) took the Fuxin 132 subsidence area as the example to extend their research, finding that with the increase 133 of the distance from the crack, the soil moisture content increased, but the influence 134 was not obvious after 2m; Ma and Yang (2019) took the soil in the development area 135 of ground fractures mined by Ephedra in northern Shanxi as the research object, 136 finding that the moisture content of the soil near stepped ground fissures, except the 137 surface, is higher than that in the non-cracked area at all depths. The above scholars 138 believe that the ground fissures caused by coal mining are important factors in 139 affecting the changes of soil moisture in the subsidence area, but a consensus has not 140 been formed about the degree and scope of the influence. 141 On the other hand, some scholars believe that coal mining subsidence has no or supply comes from atmospheric precipitation, but not from groundwater. Mining 157 disturbance affects groundwater, but has a limited impact on soil water. 158 To sum up, the authors believe that the main reason for the differences in 159 previous studies is that the influence of coal mining on soil moisture was only studied Therefore, the authors took Nalin River No.2 mine, whose reserves are at 6 years, in Yu Shen fu coal mine area in the east of Mu Us Desert as the research object, constructed classic statistics and multi-dimensional geo-statistics methods. Moreover, 176 the study analyzed the variability of soil moisture in the unsaturated zone in the 177 subsidence area from the perspectives of time and space, and discussed the impact of 178 coal mining subsidence on soil water in the aeration zone from the perspective of 179 spatial variability rather than that of absolute soil water changes, so as to provide the 180 scientific basis for ecological restoration in mining areas.   196 According to the time of the mining completion, the study area was divided into 197 control area (CK), 1-year subsidence area (S1) and 2-year subsidence area (S2) with 198 partition sampling. According to the underground coal mining process, the 199 checkerboard distribution method was adopted. The grid design was 75m * 100m, and

224
Geo-statistics can be adopted to study the spatial distribution of soil heavy metals, 225 soil nutrients and soil moisture (Chartres, 1986  Semi-variance function, also known as variogram, is used to quantify the 230 randomness and spatial structure of variables, below is the calculation formula: (1) 232 In this formula, γ (h) is the variogram; h is the spatial distance between two In these formulas, α is the range; C0 is the nugget constant, which represents the 241 variance caused by random error; C is the space structure value caused by systematic 242 factors; C0 + C is the abutment value, representing the total variance of variables; C / 243 (C0 + C) is the spatial structure ratio. If C0/(C0+C) < 25%, it can be shown that the 244 data has strong spatial correlation; when 25% < C0/(C0+C) < 75%, it can be shown 245 that the data has medium spatial correlation; when C0/(C0+C) > 75%, that the data has 246 weak spatial correlation can be shown (Bogunovic, Pereira, & Brevik, 2017).

247
The standard for selecting the semi-variance function model is that the closer the 248 mean absolute error(MAE) and the root mean square error (RMSE)of the cross-check 249 result is to 0, the closer the regression fitting coefficient of determination R 2 is to 1, 250 the higher the accuracy of the model's simulation is.

251
Kriging interpolation method is based on the spatial autocorrelation, using the 252 original data of the regionalized variables and the structure of the variogram, with the 253 adoption of linear, unbiased, and optimal interpolation methods for the unknown 254 sampling points of the regionalized variables. The formula is as the following: In this formula, Z (x0) is the value of the point to be estimated; n is the number of 257 sampling points; Z (xi) is the value of the i sampling point; λi is a group of weight 258 coefficients; ∑λi = 1; the selection of λi ensures that the estimation of Z (x0) is 259 unbiased and the estimation variance is minimum.  showed that the CVs from large to small were S1, S2 and CK, with an average CV of 292 43.60%, 29.31% and 25.68% respectively, ranging from 24.7% to 65.53%, 20.32% to 293 39.04% and 13.37% to 36.78%, indicating that the spatial variability of soil moisture 294 in 1-year subsidence area was the highest. The above data show that there is no 295 significant difference between the subsidence area and the control area, but there are 296 still varying degrees of variability between different areas.

297
Compared with the CV of the same soil depth in different regions, the spatial 298 variability of CK increased at the depth of 0-3m and 4-10m. The variability of S1 299 gradually increased in the range of 0-5m and 7-9m, with the highest variability in 300 3-4m and 8-9m. S2 showed an increasing trend as a whole, but the variability was in 301 the middle of CK and S1, indicating that coal mining and other factors had a certain 302 impact on the soil moisture in the two-year subsidence area. But as the subsidence  Table2 shows the best-fitting models of CK, S1 and S2 in 0-10m soil layers. C0 is the subsidence area shows that the random variation of deep soil was larger than that 323 of surface soil, which may be caused by the difficulty of deep soil sampling.

324
According to C0+C, all soil layers in the control area were lower than those in 325 the subsidence area, which indicates that the variation of soil moisture in the control 326 area was small within the range of variation. However, considering the variation trend 327 of C0 in the three areas, it is found that the C value (structural variance, representing 328 the variation of non-random causes in the subsidence area) at 0-1m and 5-10m was 329 relatively larger than that in the CK, which indicated that the structural factors 330 including climate, coal mining subsidence, and soil texture have a greater impact on 331 the subsidence area.

332
The spatial structure ratio C0/(C0+C) represents the proportion of system were larger than 50%, and the degree of spatial variation was high. In conclusion, the soil moisture in the one-year subsidence area and 2-year subsidence area shows strong 342 or moderate spatial variability on the surface of soil (0-1m) and deep soil (5-10m), 343 which is consistent with the variation trend in the classical statistical results. 344 Meanwhile, it is shown that the contribution of random factors to the spatial 345 distribution of soil surface and deep layer is relatively small, and the spatial variation 346 is mainly caused by structural factors.

347
Based on the above results, it is found that the overall variation degree of 348 subsidence area is higher than that of the control area, and the variation caused by  (Figure 4 and Table 3). The interpolation accuracy of the control area is 355 generally higher than that of the subsidence area, and the interpolation accuracy of the 356 subsidence area is the lowest in 1 years. It can be concluded that as the spatial 357 variability of soil moisture increases, the accuracy of Kriging interpolation decreases. From the above analysis, it can be seen in Figure 5 that soil moisture has great 360 spatial variability in 0-1m and 5-10m. Therefore, Kriging interpolation method was 361 used to draw the spatial distribution map of soil moisture in CK, S1 and S2 layers 362 (0-1m, 5-6m and 9-10m).

363
Comparing the interpolation results of CK, S1 and S2 in 0-1m, 5-6m and 9-10m, 364 it was found that there were no significant spatial variability in 0-1m, and there were 365 one low value area and two high value areas of SM. The low value area was located in 366 the one-year subsidence area, while the high value area was located in the control area 367 and the two-year subsidence area. This is mainly because the control area was not 368 affected by the ground fissures caused by coal mining, while the two-year subsidence 369 area was gradually stable so that the soil moisture was restored. At 5-6m, the high 370 value area was mainly located in the control area and 1-year subsidence area, but the 371 soil moisture distribution in the subsidence area had poor spatial continuity so as to 372 indicate that the subsidence caused by coal mining had a certain impact on the soil 373 moisture distribution. In 9-10m, the distribution of high value and low value areas of 374 soil moisture were similar to that of 0-1m soil layer. The high value area was mainly 375 located in the northwest of the control area and the 2-year subsidence area. However, 376 the low value area and the high value area in the 2-year subsidence area had poor 377 spatial continuity, which referred to high spatial variability. On the whole, the 378 distribution of soil moisture is that the high value area was located in the control area, 379 followed by the 2-year subsidence area and the 1-year subsidence area, which proves 380 that the surface subsidence has a greater impact on the soil moisture; the areas with 381 obvious spatial variability of soil moisture were mainly in the subsidence area, 382 especially in the deep soil, whose reason is the control area has not been mined. The 383 surface subsidence and ground fissures caused by the disturbance of coal mining 384 changed the soil structure and the water transport channel, so the soil moisture 385 decreased and the spatial variability increased.

386
The results of soil water interpolation of CK, S1 and S2 in 0-1m, 5-6m and 387 9-10m were compared vertically, and then it was found that the soil moisture in the 388 three regions decreased initially and then increased, which was consistent with the 389 changing trend of water in the results of classic statistical analysis. At the same time, 390 it can be seen that in CK area, there was weak spatial variability in each depth, and 391 the vertical distribution of soil water had obvious regularity, showing a decreasing 392 trend from northwest to Southeast; in S1 and S2 areas, the soil moisture in each depth 393 layer was less than that in CK area, and there was no obvious regularity in the vertical 394 distribution, and the variation degree of soil moisture in each depth layer was also    In addition, soil moisture is also affected by its own water holding capacity,

519
In this study, the classical statistics and multi-dimensional geo-statistics method 520 were used to analyze the spatial and temporal distribution of 0-10m soil moisture in 521 the 1-year subsidence area, 2-year subsidence area and control area of Nalin River

522
No.2 mine. Through the above research, it is found that: