The research team, led by Peng Li, Yong Gao, and Xiaohong Dang of Inner Mongolia Agricultural University, focused on two adjacent mining areas in Ejin Horo Banner, Ordos City, near the border between Shaanxi Province and Inner Mongolia. The Lijiata mining area sits in an aeolian sandy landscape at the southern edge of the Mu Us Sandy Land, while the Buertai area lies just a few kilometers away on the loess plateau, where thick Quaternary loess deposits blanket ridge-and-gully terrain. Because the two sites share nearly identical climatic conditions—roughly 350 millimeters of annual precipitation and a mean annual temperature of 6.6 degrees Celsius—the design allowed the researchers to isolate the effects of soil type and subsidence disturbance from the confounding influence of weather.

From December 2024 through April 2025, the team deployed HOBO U23 automatic recorders at four fixed monitoring sites: subsidence and non-subsidence zones in each soil type. Sensors measured soil temperature and volumetric water content at six depths from 10 to 60 centimeters, logging data every hour. This continuous, high-resolution record captured the full arc of the freeze-thaw season, from initial freezing through stable freezing, alternating freeze-thaw, and final thaw. The researchers classified these stages using daily minimum and maximum soil temperatures relative to 0 degrees Celsius, then quantified water variability at each depth using coefficients of variation.

The timing results reveal a striking contrast between the two soils. Freezing progressed from shallow to deeper layers in both soil types, beginning in early to mid-December with little difference in onset dates. But the sandy soil froze as a coherent block: every layer from the surface down to 60 centimeters reached complete freezing on December 12. Loess soil behaved differently, with the upper 30 centimeters freezing completely on December 12 while the 40-to-60-centimeter layers lagged until late December. Loess then thawed earlier than sand—by as much as 35 days at some depths—and cycled through fewer daily freeze-thaw events, with counts ranging from 58 to 74 cycles in loess versus 75 to 93 in sand.

Subsidence left its own unmistakable fingerprint on this seasonal clock. In the loess subsidence area, complete freezing arrived approximately 6 days earlier than in the undisturbed reference zone, and thawing was delayed by about 2 days. The sandy subsidence area also showed delayed thawing, lagging the non-subsidence zone by 2 days. The explanation lies in how mining fractures the ground: surface subsidence destroys the original structural layering of the vadose zone, loosens the soil, and opens vertical fissures that increase infiltration pathways and the area of soil exposed to evaporation. Once ice forms, its thermal conductivity exceeds that of liquid water, accelerating water-heat exchange and speeding the freezing rate in the disturbed ground.

Soil temperature itself proved to be primarily a follower of air temperature, with the strongest atmospheric influence concentrated in the 0-to-40-centimeter layer and diminishing with depth. During freeze-thaw transitions, soil temperature exhibited a characteristic decrease-then-increase pattern and responded to air temperature with a noticeable lag. The subsidence zones also displayed distinctive thermal signatures: in the sandy subsidence area, surface temperatures during full freezing were higher while deep temperatures were lower, compressing the temperature differences among layers, whereas the loess subsidence area showed more pronounced temperature contrasts between depths.

Water dynamics diverged sharply between the soil types. In the loess region, soil water content outside the subsidence zone exceeded that inside it across nearly all layers, by margins ranging from 0.03 to 46.67 percent over the 0-to-60-centimeter profile. The non-subsidence loess zone recorded a maximum water content of 10.41 percent on December 13 and a minimum of 5.11 percent on January 21, while the subsidence zone swung from 11.62 percent down to just 1.35 percent. In the sandy region, the pattern partially reversed: water content in the 0-to-40-centimeter layers outside the subsidence zone ran 4.20 to 10.76 percent higher than inside, but the subsidence zone showed more active water movement in deeper layers and relative stability at the surface.

The variability analysis added another layer of nuance. Coefficients of variation ranked highest in the aeolian sand non-subsidence zone, followed by the sand subsidence zone, the loess subsidence zone, and finally the undisturbed loess zone. In the undisturbed loess profile, the upper 30 centimeters showed moderate variability while deeper layers were stable, but subsidence pushed the 0-to-10-centimeter and 30-to-40-centimeter layers into the strong-variation category. Sandy profiles showed strong variation from 0 to 30 centimeters under both conditions. These patterns reflect how the higher porosity and lower initial water content of sandy soil produce greater fluctuations during freeze-thaw cycles, while subsidence fissures amplify vertical hydraulic connectivity and promote preferential flow.

Perhaps the most technically significant result concerns the coupling relationship itself. Fitting polynomial curves to paired temperature and water measurements revealed significant nonlinear relationships at every depth in all four zones, with coefficients of determination ranging from 0.55 to 0.93, all statistically significant. In subsidence areas, soil water content initially rose with temperature, peaked, and then declined; in sandy subsidence sites, water increased steadily with temperature but at a gradually diminishing rate. Correlations were strongest in surface layers for most zones—the loess non-subsidence area peaked at R² of 0.93 in the 0-to-10-centimeter layer, while the sandy non-subsidence area reached 0.91 at 20 to 30 centimeters—but the sandy subsidence area broke the pattern, with correlations that did not vary systematically with depth, likely because wind erosion and subsidence fissures expose deep soil to environmental forcing.

The broader implications reach well beyond the Ordos Plateau. Because freeze-thaw processes regulate soil evaporation, water storage, and land-atmosphere energy exchange, any disturbance that shifts the timing of freezing and thawing ripples through the entire ecohydrological system of cold, arid regions. The study demonstrates that under comparable climatic conditions, soil structure—not climate—is the dominant regulator of hydrothermal coupling, and that mining-induced subsidence acts as a structural perturbation with measurable consequences for how water migrates and heat flows through the frozen vadose zone. As coal production continues across China’s ecologically fragile northern belt, these findings provide a scientific basis for restoration strategies that account for the hidden winter physics of disturbed ground, and they underscore that the legacy of underground extraction is written not only in sunken landscapes but in the frozen pores of the soil itself.

Subject of Research: Effects of coal mining-induced subsidence on soil water-heat coupling during freeze-thaw cycles in loess and aeolian sandy soils of Inner Mongolia

Article Title: Soil hydrothermal coupling dynamics in a Loess–Aeolian sandy mining subsidence area under freeze-thaw conditions

Article References: Li, P., Gao, Y., Dang, X., Liu, Y., Han, Y., Liu, J., & Li, X. (2026). Soil hydrothermal coupling dynamics in a Loess–Aeolian sandy mining subsidence area under freeze-thaw conditions. Frontiers in Environmental Science, 14, Article 1901630. https://doi.org/10.3389/fenvs.2026.1901630

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1901630

Keywords: soil freeze-thaw, mining subsidence, soil water-heat coupling, loess, aeolian sand, Inner Mongolia, seasonal frozen soil, coal mining, soil moisture dynamics, vadose zone, ecohydrology, Ordos Plateau