Volumetric changes induced by soil moisture phase changes can lead to pore system redistribution in freezing and thawing soil, which in turn affects soil strength and stability. The prefreezing water content and the number of freeze–thaw cycles (FTCs) affecting key factors of soil pore changes, and they determine the volumetric change magnitude and frequency during ice–water phase transitions. This study aims to reveal the effect of the prefreezing water content and the number of freeze–thaw cycles on the pore size distribution (PSD) of black soil, meadow soil and chernozem, which account for the largest arable land area in Heilongjiang Province, China. In situ soil samples with different prefreezing water contents were subjected to 1, 2, 3, 5, 10, and 20 FTCs, and then nuclear magnetic resonance (NMR) was used to quantify the PSD. It was shown that the pore sizes of the three soil types spanned multiple orders of magnitude, ranging from 0.001 to 100 μm overall. The inflection point of the cumulative porosity curves of all three soils occurred near 0.1 μm. For black soil and chernozem with high prefreezing water contents, when the number of FTCs reached 10 or 20, the soil self-weight led to thaw settlement, which reduced the difference in the total porosity of the soils with varying moisture contents. The initial FTC exerts the most significant influence on the pore structure. The impact of the prefreezing water content on soil pore structure diminishes as the number of FTCs increases. The plant root residues rendered meadow soil less sensitive to water content differences after the first FTCs but also limited the development of macropores during the late freeze–thaw period. The prefreezing water content alters the distribution of soil moisture before freezing and has a greater influence on the pore distribution of frozen-thawed soils compared to the cumulative effect of multiple FTCs.

Snow, as an important component of the cryosphere, holds a crucial role in the construction of polar infrastructure. However, the current research on the mechanical properties of snow is not comprehensive. To contribute to our understanding of the mechanical behaviors of snow in cold regions, uniaxial compression tests under step loading were performed on the snow. With the Maxwell model as the basis, different temperatures, densities, and loading rates were set to establish constitutive equations of snow. The changes in the elastic modulus and viscosity coefficient of snow with respect to three variables were investigated. The results show that the loading rate has no obvious effect on the elastic modulus and viscosity coefficient of snow. Both the elastic modulus and viscosity coefficient of snow follow an exponential function with respect to density, with an increase in density, resulting in a higher value. As temperature decreases, the elastic modulus and viscosity coefficient initially decrease and then increase, whereas no specific functional relationship between them was observed. Additionally, a new constitutive equation considering snow density is derived based on the Maxwell model.

The ice flood phenomenon frequently occurs in frigid locations of high latitude and high altitude, which triggers ice dam or ice jam flooding thus endangering personal and property safety. Hence, a scientific risk evaluation with enough consideration of each factor is a basic and necessary requirement for preventing ice flood disaster risks. This study establishes a risk evaluation system for ice flood disasters based on the catastrophe theory and utilizes the Pearson correlation coefficient to screen underlying indicators to evaluate the risk of ice flood in the upper Heilongjiang River region. Considering the correlation between different indicators, a hierarchical cluster analysis is invoked to simplify the indicator set and to select typical years. The results of the evaluation system indicate that the catastrophe membership values in the Mohe, Tahe, and Huma regions from 2000 to 2020 ranged from 0.86 to 0.93. Based on the membership values and the actual disaster situations, a four-level classification of risk ratings is conducted. The comparison between the results obtained from the catastrophe theory evaluation method and the fuzzy comprehensive evaluation method reveals similar risk levels, which verifies the effectiveness and practicality of the catastrophe theory applied to the ice flood risk evaluation and presents a novel method for the study of ice floods.

Ice is a common natural phenomenon in cold areas, which plays an important role in the construction of cold areas and the design of artificial ice rinks. To supplement our knowledge of ice mechanics, this paper investigates the mechanical properties of granular snow ice. The factors influencing the flexural strength of granular snow ice are analyzed through a three-point bending test. It is found that flexural strength is affected by strain rate. At low strain rates, flexural strength increases with increasing strain rate, whereas at high strain rates, flexural strength decreases with increasing strain rate. As temperature decreases, the flexural strength value of ice increases, but its brittleness becomes more pronounced, indicating that the strain rate corresponding to the maximum flexural strength is lower. Within the test temperature range, the tough-brittle transition range is from 6.67 × 10⁻⁵ s⁻¹ to 3.11 × 10⁻⁴ s⁻¹. At −5 °C, the strain rate corresponding to the maximum bending strength is 3.11 × 10⁻⁴ s⁻¹, while at −10 °C, it is only 6.67 × 10⁻⁵ s⁻¹. Flexural strength is influenced by crystal structure. At −20 °C, the average flexural strength of granular snow ice is 2.85 MPa, compared to 1.93 MPa for columnar ice at the same temperature. Through observation, we found that there are straight cracks and oblique cracks. The fracture toughness of granular snow ice was investigated by cutting prefabricated cracks at the bottom of the ice beam and employing a three-point bending device. It is found that fracture toughness decreases with increasing strain rate. Temperature also affects granular snow ice. At −15 °C, fracture toughness is 181.60 kPa·m^1/2, but at −6 °C, it decreases to 147.28 kPa·m^1/2. However, at varying temperatures and strain rates, there is no significant difference in the fracture patterns of ice samples, which predominantly develop upward along the prefabricated cracks.

Thermal conductivity is a crucial factor for the soil, which is significantly affected by environmental conditions. Based on the variation in the thermal conductivity and the influencing factors of silty clay obtained by the freeze–thaw cycle test, this paper adopted four machine learning models optimized by particle swarm optimization (PSO), including the artificial neural network model (ANN), random forest model (RF), support vector machine model (SVM), and extreme gradient boosting model (XGBoost) to predict the thermal conductivity of the soil. Meanwhile, mean absolute error (MAE), root mean square error (RMSE), and correlation coefficient(R²) were used to evaluate the accuracy of the models. The accuracy of the machine learning model and empirical model were also compared. Then, the Monte Carlo simulation was used to analyze the stability of the models. The research results showed that the predicted performance of the machine learning models is significantly better than the empirical models. Among all the machine learning models, the R² of the PSO-ANN model is above 0.95, while both RMSE and MAE values are below 0.02 (W·m⁻¹·K⁻¹). In addition, the stability order of the machine learning models is PSO-XGBoost, PSO-ANN, PSO-RF, and PSO-SVM. Therefore, comprehensively considering the accuracy and stability of the four machine learning models, the PSO-ANN model is recommended to predict soil’s thermal conductivity under freeze–thaw action.

Ground penetrating radar (GPR) has proven to be a very effective method for examining ice thickness. However, two preconditions must be met for this approach to be useful; the round-trip travel time of electromagnetic (EM) waves and radar transmission speed in the ice must be known. These issues are problematic because many factors affect radar transmission speed in ice, including impurities, physical properties such as porosity and density, and temperature. Results show that if these factors are not taken into account and a signal velocity of 0.17 m/ns in pure ice is used to estimate thickness, overestimates will result. We carried out a series of GPR surveys using dual channel host 200 MHz shielded antennas at the Toudaoguai Hydrological Station on the Yellow River, China, and collected samples to analyze ice impurities and physical properties. The results show that the ice crystal types include frazil, granular, and column at the Toudaoguai Hydrologic Station section. Our analysis of ice gas bubble and sediment content showed that the gas bubble volume content is between 11.95 and 13.0% in the frazil ice and between 7.9% and 8.6% in granular and columnar ice. At the same time, the ice sediment content ranged between 0.11‰ and 0.57‰, and the average was 0.24‰ in granular and columnar ice, which was about one-tenth that of the suspended sediment concentration in water. Additionally, a combination of GPR data as well as ice impurities, porosity, density, and temperature enabled us to provide insights on the variability of radar transmission speed and the equivalent dielectric permittivity in river ice. Our extensive observations reveal that radar transmission speed falls between 0.141 m/ns and 0.164 m/ns and that the equivalent dielectric permittivity of river ice increases in concert with ice temperature.