Accurate characterization of temperature fields and ice thickness evolution is critical for the rapidly changing Arctic sea ice system, when field-based buoy observations remain limited. Building on weak solution theory and solvability analysis for the coupled ice-water two-phase system, a physics-informed neural network (PINN) framework was developed for forward simulation and interface inversion of a Stefan free-boundary problem. The proposed approach enabled the simultaneous reconstruction of temperature fields in phases and identification of the interface trajectory. Methodologically, the framework combined rigorous theoretical analysis, a PINN implementation consistent with physical constraints, and observation-driven validation, thereby providing a systematic procedure for temperature field reconstruction and physically consistent inversion in ice-water two-phase Stefan free-boundary inverse problems. Validation proceeded from equivalent-flux benchmarks to fully coupled two-phase simulations. In the equivalent-flux stage, the water-phase impact on the ice was parameterized by a time-varying equivalent oceanic heat flux. With a three-stage training strategy and a multiplicative correction factor for the conductive flux, the ice temperature error was below 0.5 degrees C, and the absolute thickness error was under 2.5 cm. In the coupled stage, hard enforcement of the boundary conditions further stabilized convergence, keeping temperature errors below 0.3 degrees C in both phases and limiting the absolute thickness error to at most 4 cm. These results demonstrated that a physics-constrained PINN can deliver stable and accurate two-phase thermodynamic inversion from limited observations, and they motivate extensions toward a unified atmosphere-snow-ice-ocean inversion framework.

Ice cover significantly alters lake ecology, most directly affecting photosynthesis and dissolved oxygen. Accurate, rapid, and continuous in-situ monitoring of key ecological indicators (e.g., dissolved oxygen and chlorophyll) beneath the ice is crucial for protecting river and lake ecosystems. Using high-frequency sensors, the responses of chlorophyll-fluorescence and dissolved oxygen production to irradiance and thermal conditions were explored during the ice-covered period in Lake Wuliangsuhai, a shallow lake in Inner Mongolia, China. Based on the dissolved oxygen results, we estimated net ecosystem production and ecosystem respiration in January-February 2018. Our results showed that the photosynthetic activity of plankton was tightly controlled by photosynthetically active radiation. The penetration depth of photosynthetically active radiation was 90 cm (including 50 cm of ice). Oxygen production occurred mainly within the photic zone, while at greater water depths, oxygen consumption by respiration was greater than its production by photosynthesis. When irradiance was low (<35 mu molm(-2)s(-1)), chlorophyll-fluorescence and irradiance were positively correlated; however, with increasing radiation intensities, chlorophyll-fluorescence decreased, causing a decrease in net ecosystem production. Ecosystem respiration increased in parallel with water temperature. Although the metabolic processes showed a certain degree of lag (3-4 h) relative to the key physical drivers, the rapid changes in chlorophyll-fluorescence and dissolved oxygen concentration clearly proved that photosynthesis was primarily controlled by irradiance. The results reveal the metabolic characteristics of plankton and their physical driving mechanisms during the ice-covered period in shallow lakes, supporting the protection of lake ecosystems in cold and arid regions under climate warming.

Due to the complexity inherent in river ice dynamics, the utilization of remote sensing imagery represents the most crucial and effective method currently available for monitoring changes in river ice. In the Inner Mongolia segment of the Yellow River during winter, two distinct types of ice surfaces are observed: juxtaposed ice and consolidated ice. Additionally, certain areas of open water remain unfrozen. Rapid identification and classification of extensive ice formations and open water zones along this lengthy river section constitute critical information for informed decision-making in ice prevention and management strategies within the Yellow River basin. This paper takes the formation and characteristic analysis of different types of ice in the Yellow River channels in Inner Mongolia as the starting point. It employs a support vector machine (SVM) as the classifier and introduces an optimized model for classifying river ice types using high-resolution Sentinel-2 optical imagery. The model utilizes multi-band spectral features, along with multi-spectral fusion indices such as the normalized difference snow index (NDSI) and the normalized difference frozen surface index (NDFSI), as feature vectors. This approach attains an overall accuracy of 94.91% in classifying different types of ice and can significantly contribute to river ice monitoring by offering robust theoretical support. In the winter of 2023–2024, the proportion of juxtaposed ice on the Yellow River section in Inner Mongolia changed from 45% to 55%, the proportion of consolidated ice changed from 30% to 40%, and the proportion of open water changed from 9% to 19%. This research investigates the characteristics of river ice formations and develops a classification methodology for river ice patterns utilizing high-resolution Sentinel-2 imagery in conjunction with a supervised classification algorithm. The findings of this study are intended to offer technical support for the expedited interpretation of ice conditions in the Yellow River, thereby serving as a scientific basis for precise monitoring and effective disaster prevention and management related to river ice phenomena.

Lake ice phenology in Northeast China mediates regional climate interactions and sustains winter tourism and fisheries. However, sparse long-term observations limit our understanding of ice cover trends under climate warming. Here, we combine a one-dimensional thermodynamic lake model with machine-learning residual correction and train and validate the model with satellite-derived ice phenology data from 32 lakes during 2001–2014 to simulate lake ice phenology from 1901–2100 through historical hindcasts and CMIP6 projections (SSP126, SSP370, and SSP585). Results show a marked shift toward later freeze-up, earlier break-up, and shorter ice duration beginning in the 1970s. Across the three scenarios, regional air temperatures are projected to increase by 1.2–8.1 °C century−1, whereas the mean ice duration decreases by 13.6–49.8 d century−1. Under SSP585, historically extreme conditions are projected to become the new phenological state, implying increasing pressure on ecosystem management and seasonal planning for winter economic activities.

Sea ice plays an important role in the heat transfer into the Arctic Ocean whereas the presence of melt ponds on sea ice complicates the scenario. However, the refreezing pond is less focused and documented in comparison with the well-established seasonal variation. To better evaluate the effect of melt pond on the freezeup of sea ice, we conducted a series of observations with 81 melt ponds in the central Arctic during freezeup, 2012-2020. The melt ponds are categorized into five types based on the surface state to effectively investigate the various characteristics. The total albedo of each type is 0.14 (water pond), 0.20 (water-ice pond), 0.25 (ice pond), 0.39 (ice-snow pond), 0.74 (snow pond), respectively, showing the increase on albedo in August and September (0.0036 d-1) due to the changes of the surface state. The albedo dependence on the surface state, ice lid, pond depth and underlying ice is examined using both in-situ measurements and modified radiative transfer model, with result indicating the dominance of surface state followed by ice lid thickness. The total albedo of ice ponds decreases with increasing pond depth, and the raising of ice lid thickness reduces the albedo while rises that of ice-snow ponds. In addition, further analysis reveals the capacity of different ratios of spectral albedo on the distinction between snow-covered pond and unponded ice, potentially improving the melt pond retrieval algorithms.

Ice exhibits ductile deformation under uniaxial compression at low strain rates, a regime critical for ice-structure interaction and geophysical processes, with defects (gas bubbles and brine pockets) critically influencing microcrack development. To investigate the damage mechanisms governing the ice ductile deformation, uniaxial compression tests were performed on natural columnar ice sampled from a brackish lake (ice sample salinity averaged 0.9 ppt) under varying strain rates (10-6 s- 1 to 10-4 s- 1), temperatures (-12 degrees C to - 3 degrees C) and loading directions relative to the crystal columns. Acoustic emission (AE) signals were continuously monitored using four sensors, from which parameters such as count rate, energy, and amplitude were extracted, and three-dimensional AE source locations were determined to trace spatiotemporal microcrack evolution. On the basis of AE activity, microcrack development was divided into four characteristic stages: initial nucleation, stable growth, rapid propagation, and post-failure. The results revealed that low strain rates facilitated an extended stable stage and low-energy microcracking, while high strain rates triggered diffuse, high-energy microcracking and the absence of a stable stage. The b-value was determined using AE signal amplitudes to reflect cracking intensity and scales, with higher values linked to low-amplitude signals and small-scale crack dominance, and generally decreased with increasing strain rate. Temperature effects were linked to creep, with elevated temperatures enhancing creep and promoting dispersed low-energy microcracks, whereas lower temperatures suppressed creep and favoured localized, high-energy damage. Loading parallel to the ice columns caused earlier microcrack initiation and pervasive propagation than loading perpendicular. Overall, these findings establish a linkage between microcrack activity and the macroscopic mechanical behavior of ice, which is helpful for advancing understanding of ice-structure interaction and contributing to research in ice geophysics.

The physical properties of ice are primarily determined by its microstructure, such as gas bubbles and brine pockets within ice. However, the distributions and evolution patterns of these inclusions during ice freezing and melting remain poorly understood. Most studies therefore treat the ice microstructure as constant and ignore its seasonal variations. To investigate this issue, in situ experiments were conducted on a brackish lake to collect detailed information on the variations in the microstructure of ice using continuous sampling and a highresolution imaging system. During the freezing phase, the volume fractions of gas bubbles and brine pockets remained relatively stable and decreased, respectively. Furthermore, the size of gas bubbles and brine pockets in the ice surface, middle, and bottom layers of ice increased clearly during the melting phase of ice due to different reasons. The nearly 30 % increase in gas bubbles observed in the middle layer was driven by ice temperature, while the increase in the surface layer was influenced by the net shortwave radiation. Additionally, the variation in the size distribution of inclusions was attributed to the merging process, which primarily occurred among smaller inclusions rather than among larger inclusions. The number of both gas and brine inclusions in the middle layer was found to decrease by 10 to 20 % for an increase in ice temperature by 1 degrees C, while this phenomenon was not observed in the surface or bottom layers. This study could improve our understanding of how ice microstructure changes.

Rapid changes in Arctic summer sea ice exert substantial influences on the polar climate system, maritime navigation, and resource exploitation, while subseasonal-to-seasonal (S2S) prediction of sea ice state remains highly uncertain. Using daily observations and reanalysis data of sea ice concentration (SIC) and thickness (SIT) from 1979 to 2023, together with concurrent atmospheric and oceanic fields, this study develops a multivariate linear Markov model to perform S2S predictions of Arctic summer sea ice. Sensitivity experiments with different variable combinations, weighting strategies, and modal truncation schemes are conducted, and predictive skill is systematically evaluated against persistence and climatological baselines. Results indicate that the model exhibits stable forecast skill without pronounced error accumulation at extended lead times. SIC predictability is primarily governed by its intrinsic spatiotemporal persistence and is significantly modulated by oceanic thermodynamic forcing, particularly sea surface temperature and surface net energy flux, highlighting a pronounced oceanic memory effect. In contrast, local atmospheric dynamic variables provide limited incremental skill. For SIT, predictability is dominated by its own historical state, with SIC contributing marginal short-term improvement and air–sea coupling exerting weak influence. Overall, the proposed framework effectively extracts dominant predictable signals with clear physical interpretability, providing a computationally efficient statistical approach for S2S prediction of Arctic summer sea ice.

Arctic sea ice is a crucial component of Earth’s system and is experiencing dramatic variations. Satellite remote sensing is one of the most significant tools for monitoring Arctic sea ice. Efficiently extracting sea ice information from satellite-observed materials is a considerable challenge. In recent years, deep learning (DL) has been widely applied to Arctic sea ice remote sensing, enabling advances in classification, parameter retrieval, object detection, superresolution, and prediction. Beyond summarizing existing studies, this review article identifies a clear methodological evolution from early image-based feature learning to multilevel feature integration and, more recently, toward unified representation learning. Compared with existing reviews, this article contributes a stage-based synthesis of methodological evolution, a cross-task comparison of representative models, and practical guidance for framework selection in major Arctic sea ice applications. Through a critical synthesis of the literature, we highlight persistent challenges shared across applications, including limited generalization across regions and sensors, insufficient physical constraints, scarcity in high-quality labeled data, and limited treatment of uncertainty and interpretability. We further discuss emerging directions in DL-based sea ice remote sensing, including sea ice foundation models, physics-informed learning, alternative supervision strategies, uncertainty modeling, and explainable analysis specific to sea ice processes. These insights provide a consolidated perspective on advancing reliable and scientifically meaningful Arctic sea ice remote sensing.

Lake ice phenology plays a critical role in determining the hydrological and biogeochemical dynamics of catchments and regional climates. Lakes with complex shorelines and abundant aquatic vegetation are challenging for retrieving lake ice phenology via remote sensing data, primarily because of mixed pixels containing plants, land, and ice. To address this challenge, a new double-threshold moving t-test (DMTT) algorithm, which uses Scanning Multichannel Microwave Radiometer (SMMR) and Special Sensor Microwave/Imager–Special Sensor Microwave Imager/Sounder (SSM/I–SSMIS) sensor-derived brightness temperature data at a 3.125 km resolution and long-term ERA5 data, was applied to capture the ice phenology of Lake Ulansu from 1979 to 2023. Compared with the previous moving t-test algorithm, the new DMTT algorithm employs air temperature time series to assist in determining abrupt change points and uses two distinct thresholds to calculate the freeze-up start (FUS) and break-up end (BUE) dates. This method effectively improved the detection of ice information for mixed pixels. Furthermore, we extended Lake Ulansu's ice phenology back to 1941 via a random forest (RF) model. The reconstructed ice phenology from 1941 to 2023 indicated that Lake Ulansu had average FUS and BUE dates of 15 ± 5 November and 25 ± 6 March, respectively, with an average ice cover duration (ICD) of 130 ± 8 d. Over the last 4 decades, the ICD has shortened by an average of 22 d. Air temperature was the primary impact factor, accounting for 56.5 % and 67.3 % of the variations in the FUS and BUE dates, respectively. We reconstructed, for the first time, the longest ice phenology over a large shallow lake with complex surface cover. We argue that DMTT can be effectively applied to retrieve ice phenology for other similar lakes, which has not been fully explored worldwide.

Global warming reduces the thickness and duration of seasonal lake ice, increasing the risk of ice cover failure. To investigate the bending behavior of ice cover, six groups of full-scale cantilever beam tests were conducted on a brackish water lake during the winter of 2023–2024, covering the following three ice periods: growth, stable, and melt. A total of 16 upward-loaded beams and 24 downward-loaded beams were tested. The results showed that the flexural strength of brackish ice was 374.21 ± 99.93 kPa, and the effective elastic modulus was 2.77 ± 0.93 GPa. The square root of bulk porosity, fitted with an exponential function, is the optimal predictor of flexural performance. Both flexural strength and effective elastic modulus systematically decreased with increasing porosity, and empirical regression formulas were established. On average, downward-loaded flexural strength was approximately 17.3% to 38.8% higher than upward-loaded strength, whereas elastic modulus showed no significant difference between the two loading directions. Flexural mechanical properties during the melt period reduced significantly, with a strength and modulus about 33.0% to 61.1% lower than those in the growth and stable periods. Comparisons with existing datasets demonstrate that the mechanical properties of brackish ice are lower than those of freshwater ice but higher than those of sea ice. This study provides new in situ data on the full-scale flexural mechanical properties of brackish ice and offers an important basis for assessing ice loads in lakes and estuarine environments under climate change.

The formation and evolution of ice in the Yellow River represent complex dynamic processes. To elucidate the structural characteristics of ice crystals and their governing mechanisms in the Inner Mongolia reach, this investigation utilized high-resolution Sentinel-2 satellite imagery to systematically monitor spatiotemporal variations in open-water formations across diverse channel morphologies throughout the ice regime period. Systematic ice sampling was conducted across diverse channel morphologies of the Yellow River to quantify critical parameters, including crystalline structure characteristics, equivalent diameter distributions, density variations, and sediment content profiles. The results indicate the transformation of open water resulting from various river configurations during the freezing season exhibits distinct characteristics, which are significantly influenced by temperature variations. Ice crystal characterization exhibits that the crystalline structure predominantly manifests as two primary forms: columnar and granular ice formations, with their distribution varying systematically across different channel configurations. Ice crystal morphology exhibits heterogeneity in both form and dimensional characteristics. Columnar ice consistently exhibits larger equivalent diameters compared to granular ice formations. A progressive enhancement in the equivalent diameter of crystals is observed along the vertical axis corresponding to the thickness of the ice during the growth process. The ranges of variation in ice crystal size, ice density, and mud content within ice exhibit differences contingent upon the specific crystal structures present. Observational studies and comparative analyses of ice samples from the Inner Mongolia reach of the Yellow River reveal that channel morphology, ambient thermal conditions, and hydrodynamic parameters are the primary determinants governing the variability in ice microstructure and its associated physical characteristics. This investigation provides fundamental scientific insights and quantitative data that advance our understanding of river ice microstructural characteristics.

Melt ponds are usually modeled for light transfer as horizontally infinite water layers on level ice, and the albedo of floe is determined by a linear combination (LC) of melt pond and bare ice albedos weighted by their areal coverages. However, this method does not reflect the actual conditions because ice floes have a limited size. In the present study, an idealized two‐dimensional Monte Carlo (MC) model was employed to investigate the influence of melt ponds and floe size on the apparent optical properties (AOPs) of summer sea ice. The results showed that the albedo and vertical light transmittance of large floes mainly depend on the melt pond fraction and ice thickness, which is consistent to previous results. However, also the floe size plays an important role in the AOPs of small floes. Two parameters were proposed to present the accuracy of the LC method for small floes with lower sea ice concentration: the ratios of sea ice albedo and transmittance determined by the LC ( α line, T line ) to the values in the MC model ( α , T ), K α = α line / α , and K T = T line / T , respectively. Due to the lateral transmittance, K α , K T ≥ 1 and asymptotically approach 1 with floe size increasing to infinity. To reduce the biases in albedo and transmittance due to floe size, new parameterization formulas were provided for K α and K T with the distance into the marginal ice zone and in different melting stages. The results have potential to be implemented into future sea ice models to correct the AOPs of small sea ice floes obtained via the LC method.

By employing high-resolution imaging and image processing techniques, a quantitative analysis was conducted on the changes in volume fraction and size of microstructures, such as brine inclusions and air bubbles, within natural saline ice at different temperatures. This study revealed the distinct stratified distribution characteristics of ice microstructure parameters along the depth direction and elucidated the differential response mechanisms of various ice layers to temperature changes. The results indicate that the sizes of brine inclusions and air bubbles decrease progressively from the surface layer to the bottom layer, with the size distribution of microstructures being most concentrated in the bottom layer. Changes in the size of microstructures in the surface ice layer are primarily dominated by solar radiation, showing strong correlations (brine inclusions: r = 0.96, p < 0.01; air bubbles: r = 0.95, p < 0.02). In contrast, the size changes of microstructures in the middle ice layer show a more significant response to ice temperature, with strong linear relationships between the sizes of brine inclusions/air bubbles and ice temperature (brine inclusions: r = 0.70, p < 0.04; air bubbles: r = 0.69, p < 0.05). The temperature of the bottom ice layer, influenced by the stable lake water temperature, remains relatively constant, and no significant correlation was observed between its microstructure size changes and ice temperature. Derived from field experiments, this study provides quantified, layer-specific mechanisms of how saline ice microstructure responds to temperature. These mechanisms offer crucial observational constraints for refining the parameterizations of ice thermodynamics and albedo feedback in cryosphere and climate system models.

The gas and brine pores within sea ice act as critical defects influencing its mechanical properties. To investigate the effects of pore characteristics on the mechanical behavior of sea ice under uniaxial compression, granular sea ice specimens were developed numerically using discrete element method (DEM). Local parameters of the model were calibrated by matching the simulated stress-strain curve with experimental uniaxial compression data. Pore characteristics (porosity, vertical distribution, and size) within the numerical specimen were configured based on field observations. Uniaxial compression simulations were conducted at a strain rate of 5.71 × 10−3 s−1, and mechanical properties of uniaxial compressive strength, failure strain, Young′s modulus, crack propagation, and energy evolution were analyzed. Results showed that crack development within specimen became active only near peak stress, and the total crack count decreased with increasing specimen total porosity with shear cracks dominating the fracture patterns. Energy analysis revealed elastic strain energy predominated before peak stress, while dissipative energy increased rapidly after peak stress, exceeding elastic storage. Compressive strength, failure strain, and Young′s modulus of specimen decreased with increasing total porosity. Comparative analysis of specimens with different vertical pore distributions indicated that total porosity primarily affected sea ice strength, and vertical pore distribution governed crack localization. The spatial arrangement of pores showed negligible influence on sea ice mechanical behavior. Furthermore, the gas-to-brine ratio significantly affected sea ice mechanical response, with higher gas content reducing strength while promoting shorter, more numerous cracks. This study provides a mesoscale insight into pore-driven failure mechanisms in sea ice mesoscale.

A key challenge in lake ice modeling is quantifying the heat flux from water to ice. In shallow Central Asian lakes, where the seasonal ice cover mainly consists of columnar congelation ice, sunlight penetration enables strong interactions between ice and water. The evolution of ice cover in Lake Ulansu (Ulansuhai, Wuliangsuhai) in northern China was investigated via the High-resolution Thermodynamic Snow and Ice (HIGHTSI) model. Atmospheric forcing was provided by calibrated ERA5 reanalysis data, and the initial freeze-up dates were identified from remote sensing observations. A new parameterization of the water-ice heat flux (F-w), which is suitable for shallow lakes, was proposed as F-w = aQ(sw) + b, where Q(sw) represents the solar heating of water and a and b are fitted coefficients. The model showed high correlations (>0.9) and low errors (<5 cm for ice thickness; <2 degrees C for ice temperature) with respect to field observations. Throughout the ice season, long- and shortwave radiation promoted ice growth and melting, respectively. Surface melting and sublimation accounted for 9.5% and 9.8%, respectively, of the total ice decay, and the water-ice heat flux F-w =-17.5 +/- 13.0 W m(-2) was critical for simulation accuracy. Furthermore, despite the shallow depth, the lake released over 100 Wm(-2 )of heat into the atmosphere for 2 days after break-up. These findings highlight the climatic sensitivity and support sustainable water resource management of more than 10,000 shallow lakes in Central Asia.

Shallow lakes (<2 m depth) of Central Asian, receiving strong solar radiation and low precipitation, are sensitive to atmospheric forcing because of their low heat capacity, yet their under‐ice thermal conditions remain poorly investigated. We conducted the first complete ice season monitoring of Lake Ulansu (Ulansuhai, Wuliangsuhai), revealing unique thermal behavior. The lake was salinity stratified (<3‰), stabilizing the lower water layer and allowing the water temperature to reach 10°C before break‐up. The solar radiation absorbed by the water ( Q sw ) drove the water–ice heat flux, with approximately 82% of Q sw returning to the ice base, facilitating a rapid shift from convective mixing to stable stratification. These findings provide key insights into the thermal regimes of Central Asian shallow lakes, informing climate models and ecological assessments for more than 10,000 similar lakes in the region.

Ice accumulation is a natural phenomenon in cold climates, but it causes serious issues to various industries. To accurately detect the thickness of ice on a horizontal cold plate, a capacitance-coupled split-ring resonator is employed. The impact of the average height of static and dynamic freezing front on the resonator at -20 degrees C was investigated experimentally. The sensor is capable of detecting freezing fronts up to a maximum height of approximately 20 mm. The height of the static freezing front can be sensitively detected within the range of 0-5.4 mm, with an average error of the sensor less than 0.53 mm. Furthermore, the height of the dynamic static can be detected with a higher degree of accuracy within 0-8.4 mm, the average detection error is 0.37 mm. This technology can serve as a point of reference for the field of anti-icing.

Seasonal lake-ice runways provide low-carbon winter infrastructure for cold-region logistics, yet direct links between ice flexural mechanics and runway design remain scarce. To address this gap, we investigated seasonal ice from Huhenuoer Lake and proposed a mechanics-based methodology for determining key design parameters. On 14 March 2024, an ice block was hoisted and tested in a cold laboratory after several days. Full-depth investigation revealed that throughout the entire-depth range, the ice consists exclusively of columnar ice with a mean density of 0.89 g/cm3 and a mean grain size of 15.88 mm. Subsequently, four-point bending tests gave an average effective flexural elastic modulus of 3.31 GPa and flexural strength of 2813 kPa. Furthermore, statistical analysis showed both properties decrease with temperature and are lower in higher-porosity specimens. Based on these results, we established a systematic procedure to derive design flexural strength, effective flexural elastic modulus, and ice thickness. When applied to the potential Huhenuoer Lake-ice runway, this method yields values of 2800 kPa, 3.24 GPa, and 30 cm for the An-2 (design flexural strength, design effective flexural elastic modulus, and design ice thickness, respectively); the same method can be used to obtain the corresponding values for other aircraft. As lake ice is locally sourced, recyclable, and temporary, it can serve as a low-carbon material for construction purposes.