Spatiotemporal evolution characteristics of freeze–thaw erosion intensity in the high-cold mountainous areas of Northwestern Yunnan from 2017 to 2023

Abstract

Freeze–thaw erosion is a prevalent surface process in cold regions, driven by the cyclic freezing and thawing of soil and rock. This process results in land degradation, soil erosion, and a range of ecological and environmental challenges, posing significant threats to the ecological security and socio-economic development of high-altitude mountainous areas. This study focuses on the high-cold mountainous regions of northwestern Yunnan, employing multi-source remote sensing data and geographic information system (GIS) technologies. Eight key evaluation factors were considered, including annual freeze–thaw cycle duration, daily average phase-change water volume, annual precipitation, slope, aspect, vegetation coverage, soil type, and soil erodibility. By integrating expert scoring with principal component analysis, a comprehensive system was developed to identify and assess the intensity of freeze–thaw erosion zones in the region. The spatiotemporal evolution of freeze–thaw erosion intensity from 2017 to 2023 was systematically analyzed. The findings suggest that high-altitude regions face considerably greater erosion intensity, attributed to lower temperatures and frequent freeze–thaw cycles. In contrast, low-altitude areas are comparatively less affected. This reflects the intricate interplay of climatic and environmental dynamics. Over the 7-year period, areas of mild and moderate erosion expanded significantly, whereas regions of minimal and intense erosion showed a decline. Annual precipitation and freeze–thaw cycle duration emerged as the primary factors influencing the spatial distribution of erosion intensity. This study offers valuable insights and decision-making support for ecological conservation and disaster mitigation in the high-cold mountainous areas of northwestern Yunnan.

Introduction

Freeze–thaw erosion, a typical geological process, plays a significant role in shaping the morphology and ecological environment of high-cold mountainous areas. This phenomenon refers to the cyclic freezing and melting processes driven by temperature fluctuations between day and night and across seasons. These cycles alter the moisture content within rocks and soils, leading to fragmentation, erosion, transportation, and deposition. Freeze–thaw erosion is particularly severe in steep high-cold mountainous regions, where it stands as the third major type of erosion after hydraulic and wind erosion. It profoundly impacts geological stability, thermal and water exchange processes, and sediment yield rates^1,2,3. In the context of global climate warming, steep high-cold mountainous areas have experienced notable climatic changes, marked by increased frequency and intensity of freeze–thaw cycles. These intensified cycles accelerate the weathering and fracturing of steep and unstable geological bodies, producing fragments of various shapes and sizes. These fragments are subsequently transported and deposited along channels and debris flow slopes, extending toward mid- and low-mountain regions^4,5,6,7. This process provides an abundant material source for freeze–thaw debris flows and other related disasters in high-cold mountainous areas.

The manifestations of freeze–thaw erosion primarily include physical destruction, chemical erosion, and biological erosion. Physical destruction refers to the formation and expansion of cracks in rocks or soils caused by the cyclic expansion and contraction due to temperature fluctuations. Chemical erosion involves the enhancement of mineral chemical reactions during the freezing and melting of water, which accelerates the erosion process^8,9. Biological erosion pertains to the accelerated degradation of soil structure due to the influence of plant root systems during freeze–thaw cycles. Quantitative methods for studying freeze–thaw erosion processes and erosion rates generally include field observations, laboratory simulations, and mathematical modeling. For instance, monitoring equipment can record temperature and humidity data to establish the relationship between freeze–thaw cycle frequency and erosion amounts. Laboratory simulations under controlled freeze–thaw conditions are also used to measure erosion rates directly^10,11. Furthermore, mathematical models quantitatively describe the relationship between freeze–thaw processes and erosion amounts based on theoretical frameworks, employing numerical calculations or statistical analyses. The intensity of freeze–thaw erosion represents the actual extent of erosion occurring during these processes and is typically validated by comparing monitoring data with model results. Accuracy assessments, such as mean squared error (MSE) or relative error, are commonly used to evaluate the reliability of model predictions. These assessments ensure that the estimated intensity of freeze–thaw erosion aligns effectively with observed erosion amounts^12,13.

Currently, two primary methods are used to evaluate freeze–thaw erosion intensity: visual interpretation and weighted evaluation. Visual interpretation relies on the texture and color characteristics of different freeze–thaw erosion intensities observed in remote sensing imagery. However, this approach is highly subjective and susceptible to variations in the quality of remote sensing data and the timing of image acquisition¹⁴. The weighted evaluation method involves selecting a range of indicators related to freeze–thaw erosion intensity, such as climatic factors, topography, and vegetation coverage. By employing techniques like hierarchical analysis and expert scoring, the weights of these evaluation factors are calculated to facilitate a comprehensive assessment of freeze–thaw erosion intensity^15,16,17,18. Both methods have distinct advantages and limitations. The choice of method should be guided by the geographical characteristics of the study area and the availability and quality of evaluation data, ensuring the most suitable approach is applied to achieve reliable results.

In high-cold mountainous regions, freeze–thaw erosion exerts profound impacts beyond geomorphological changes and ecological degradation; it also significantly contributes to sediment accumulation in rivers and valleys¹⁹. These sediments, often regarded as waste materials, present a valuable opportunity to be repurposed as resources for soil improvement and planting substrate preparation. Recent studies have demonstrated the feasibility and environmental advantages of utilizing river sediments for this purpose, particularly in areas where soil degradation and sediment disposal pose critical challenges^20,21. Through appropriate sediment conditioning and the integration of essential nutrients, their potential to enhance soil fertility and substrate stability can be fully realized. Such practices not only mitigate environmental risks but also promote sustainable land management strategies. In this context, a robust methodology that incorporates physical, chemical, and biological indicators is essential for evaluating the quality of soil derived from sediment, ensuring its practical applicability.

Due to its unique geographical and climatic conditions, the high-cold mountainous areas of northwestern Yunnan represent one of the most severely affected regions by freeze–thaw erosion in China. In recent years, the intensification of climate change has further amplified the frequency and intensity of freeze–thaw cycles in this area, making it a focal point of academic research^22,23,24. Existing studies have indicated that factors such as temperature fluctuations, precipitation changes, and human activities significantly influence the freeze–thaw erosion process. However, the spatiotemporal evolution patterns of freeze–thaw erosion intensity across varying scales remain insufficiently explored^18,25,26.

This study seeks to systematically analyze the spatiotemporal evolution characteristics of freeze–thaw erosion intensity in the high-cold mountainous areas of northwestern Yunnan from 2017 to 2023 using multi-source data. By developing a comprehensive evaluation system, this study investigates the spatiotemporal patterns of change in freeze–thaw erosion intensity and its spatial distribution characteristics. Additionally, it explores the relationships between evaluation factors and freeze–thaw erosion intensity across temporal and spatial dimensions. The findings aim to provide valuable scientific evidence to support ecological environment protection and disaster prevention efforts in northwestern Yunnan, while also serving as a reference for freeze–thaw erosion studies in other high-cold mountainous regions.

Materials and methods

Study area overview

The high-cold mountainous areas of northwestern Yunnan are situated in the northwestern region of Yunnan Province, China, on the southeastern edge of the Tibetan Plateau (Fig. 1). With elevations generally exceeding 3000 m, the region is characterized by a cold plateau mountain climate, featuring complex and diverse terrain. Intersecting mountain ranges and deeply incised river valleys dominate the landscape, and the widespread presence of seasonal permafrost makes this area particularly susceptible to freeze–thaw erosion. The annual average temperature is below 0 °C, with long, harsh winters and short, cool summers^24,27,28,29.

Fig. 1

Location map of the study area.

Precipitation in the region is concentrated primarily in the summer months, while winters remain relatively dry, contributing to highly active freeze–thaw processes. The multi-year average freeze–thaw cycle spans approximately 150 days, with the main freeze–thaw period occurring from October to April of the following year, highlighting a prolonged freeze–thaw cycle³⁰. The area’s stratigraphy and geological structure are exceptionally complex, with strata spanning from the Paleozoic Cambrian to the Cenozoic Quaternary periods. Deep fault systems and an extensive network of rivers have resulted in significant topographic dissection and intense erosion³¹. The primary landforms in the region include modern glaciers, mid- to high-altitude mountains, and alluvial plains, exhibiting relief ranging from moderate to extreme. The interplay of complex lithology, steep mountainous terrain, intricate geological structures, and modern glacial activity provides ideal conditions for the initiation and progression of freeze–thaw erosion³². Given the harsh climate and challenging topography, the ecological environment in this area is fragile. Freeze–thaw erosion has emerged as a critical factor influencing surface morphology evolution and ecosystem stability.

Data sources and preprocessing

This study utilized multi-source datasets, including remote sensing, meteorological, and geographic information data, spanning the period from 2017 to 2023. To ensure consistency and enhance reproducibility, all datasets underwent rigorous preprocessing and standardization procedures, such as cloud removal, atmospheric correction, temporal and spatial resampling, and outlier elimination. The following sections provide detailed descriptions of the data sources and corresponding processing workflows.

The remote sensing datasets included Landsat 8, Sentinel-2, and Aqua satellite data, with spatial resolutions of 30 m, 10 m, and 0.25° respectively. Landsat 8 and Sentinel-2 data were primarily used for land cover classification and vegetation coverage extraction, as well as for spatial analyses of freeze–thaw erosion intensity. These processes were conducted on the Google Earth Engine (GEE) platform (https://code.earthengine.google.com/). Preprocessing involved radiometric calibration, atmospheric correction (using GEE’s built-in algorithms), and cloud/shadow masking based on quality assessment bands. To minimize noise and fill data gaps, pixel-level median composites were generated for each time frame. Aqua satellite data, derived from AMSR-E and AMSR2 sensors (https://gportal.jaxa.jp/gpr/), were employed to calculate annual freeze–thaw cycle durations and daily average phase-change water volumes. The ascending orbit data were consistently acquired at 13:30 local time, while descending orbit data were captured at 1:30 local time. Specifically, brightness temperature data from 18.7 GHz horizontal polarization (Tb18.7H) and 36.5 GHz vertical polarization (Tb36.5 V) were used for freeze–thaw cycle duration calculations. For phase-change water volume, brightness temperatures at 10.65 GHz and 36.5 GHz in both ascending and descending orbits were utilized. To harmonize the varying spatial resolutions (10 m, 30 m, and 0.25°), all datasets were resampled to a unified grid and standardized using z-score normalization, facilitating seamless intercomparison during the freeze–thaw erosion evaluation.

Meteorological data included daily temperature and precipitation records obtained from the China Meteorological Administration (https://data.cma.cn/site/index.html) and observational data from self-constructed weather stations at four altitude gradient bands. Data from seven national meteorological stations—Deqin, Gongshan, Zhongdian, Weixi, Nujiang Prefecture, Lijiang, and Huaping were included to analyze climatic drivers of freeze–thaw erosion. Preprocessing involved checking for missing or erroneous entries, which were addressed using linear interpolation or nearest-station substitution. Daily temperature values were aggregated into monthly means where required, while precipitation records were cross-validated against station data to ensure consistency. Outliers exceeding three standard deviations were carefully examined and corrected when verifiable errors were identified.

Geographic information datasets primarily included Digital Elevation Model (DEM) data and basic geographic vector layers, sourced from the China Geographic Information Resources Directory Service System (https://www.webmap.cn/main.do?method=index). The DEM data, with a 30-m resolution, was used for terrain analysis and slope and aspect calculations. The vector layers, produced in 2022 at a 1:250,000 scale in SHP format, were utilized for base map generation. DEM preprocessing steps included sink filling, artifact correction, and slope/aspect derivation, all conducted using ArcGIS software. DEM-derived products were co-registered to align with the spatial reference system of the remote sensing datasets. Additionally, the HWSD V2.0 database was used to compute the soil erodibility K-factor for evaluating freeze–thaw erosion intensity. Relevant attributes from the HWSD dataset were clipped to the study area, extracted, and standardized. Each raster layer (e.g., slope, aspect, K-factor) was resampled to match the resolution of the remote sensing data, ensuring compatibility for integration into the final analysis.

The comprehensive preprocessing and standardization workflow, encompassing radiometric and atmospheric corrections, cloud masking, temporal compositing, coordinate system alignment, and statistical outlier detection, ensured consistency across the multi-source datasets. These steps not only harmonized the datasets but also enhanced the reliability and reproducibility of the subsequent evaluations of freeze–thaw erosion intensity.

Definition of freeze–thaw erosion zones

The region where freeze–thaw erosion is prevalent, characterized by distinct terrain and landform features shaped predominantly by freeze–thaw processes, is defined as a freeze–thaw erosion zone. The lower boundary of this zone closely aligns with that of the periglacial zone, although the periglacial zone extends 100–300 m below the permafrost zone. Consequently, the spatial extent of the permafrost zone is smaller than that of the freeze–thaw erosion zone^33,34,35,36. In the context of the study area’s climatic and topographical conditions, the criteria for defining freeze–thaw erosion zones primarily include annual average temperature, winter minimum temperature, elevation, and slope. These criteria vary across regions, and no standardized method currently exists. Based on field investigations and multi-year observational data, areas within the study region with an annual average temperature below 0 °C and a winter minimum temperature below − 5 °C are identified as potential freeze–thaw erosion zones. Additionally, regions with elevations exceeding 2500 m and slopes steeper than 15° are classified as high-risk areas for freeze–thaw erosion. Using these criteria, remote sensing imagery and Digital Elevation Model (DEM) data were integrated to map the distribution of freeze–thaw erosion zones within the study area (Fig. 2).

Fig. 2

Distribution map of freeze–thaw erosion zones in the study area.

Construction of the freeze–thaw erosion intensity evaluation system

Establishment of evaluation units

The size of evaluation units significantly influences the spatial accuracy of assessment results. Selecting appropriate evaluation units enables more precise spatial distribution information, which is crucial for identifying localized freeze–thaw erosion characteristics. It also facilitates the integration of multi-source datasets, such as temperature, precipitation, and vegetation coverage, thereby supporting a comprehensive analysis of the driving factors behind freeze–thaw erosion^37,38. In this study, a grid size of 250 m × 250 m was chosen as the basic evaluation unit, based on the topographical characteristics of the study area and the spatial heterogeneity of freeze–thaw erosion. The digital elevation model (DEM) was initially utilized to partition the study area into multiple topographical units, delineating the freeze–thaw erosion zone and ensuring the accuracy of spatial analysis. To account for the complexity of the freeze–thaw erosion process and align with the spatial resolution of the available multi-source data, these topographical units were further classified according to slope gradients, aspects, and land cover types. Additionally, the spatial raster resolution was unified to maintain consistency across datasets. This methodological framework allowed for a detailed representation of the environmental conditions and freeze–thaw erosion characteristics within each evaluation unit. It effectively captured the spatial distribution patterns of freeze–thaw erosion across the study area and provided a robust foundation for subsequent analyses of freeze–thaw erosion intensity.

Selection of evaluation factors

The selection of evaluation factors is a critical component in constructing the freeze–thaw erosion intensity evaluation system, as it directly affects the scientific validity and reliability of the assessment results. Guided by the physical mechanisms of freeze–thaw erosion, its influencing factors, and previous research findings, this study identified eight key factors spanning four domains: climate, topography, soil, and land cover^{39,40,41,42,43}. The climate-related factors, including annual average freeze–thaw cycle days (AFTCD), annual average precipitation (AAP), and daily average freeze–thaw phase-change water (DAFTPCW), collectively represent the climatic drivers influencing freeze–thaw erosion. Topographical factors, represented by slope and aspect, reflect the spatial heterogeneity of terrain’s influence on erosion processes. Soil-related factors, namely soil type and soil erodibility, emphasize the sensitivity of rock and soil to freeze–thaw dynamics. Lastly, vegetation coverage, as the land cover factor, evaluates the stabilizing effect of surface vegetation on reducing erosion risk and maintaining surface stability^22,44,45. By comprehensively integrating these factors, the evaluation system establishes a robust scientific framework for assessing freeze–thaw erosion intensity. This approach ensures a thorough and accurate evaluation, providing a reliable basis for understanding the spatial and temporal dynamics of freeze–thaw erosion.

Evaluation factor calculation

Annual average freeze–thaw cycle days

The annual freeze–thaw cycle days refer to the number of days in a year during which the surface temperature undergoes both freezing and melting processes within a single day. Frequent freeze–thaw cycles exacerbate the mechanical degradation of soil and promote material migration, ultimately contributing to the erosion of soil and rock during the melting phase. This factor plays a crucial role in evaluating freeze–thaw erosion intensity^40,46.

Surface brightness temperature data provide a reliable basis for determining the freezing and melting states of the surface. The calculation involves two key steps. First, the daily freeze–thaw status is established by analyzing brightness temperature data recorded at different times during the day and night to identify whether freezing and melting processes occurred on a given day (Eqs. 1, 2, and 3). Second, the annual freeze–thaw cycle days are calculated by compiling the freeze–thaw status for each day throughout the year. This total represents the frequency of freeze–thaw cycles annually, serving as a critical indicator for freeze–thaw erosion analysis.

The calculation formula is presented as follows:

$$F = 1.47T_{{{\text{b}}36.5v}} + 91.69\frac{{T_{b18.7H} }}{{T_{b36.5V} }} - 226.77$$

(1)

$$T = 1.55T_{b36.5V} + 86.33\frac{{T_{b18.7H} }}{{T_{b36.5V} }} - 242.11$$

(2)

$$N_{AFTCD} = \mathop \sum \limits_{i = 1}^{365} f\left( {Tb_{i, day} ,Tb_{i, night} } \right)$$

(3)

In the formula, T_b36.5V represents the vertical polarization brightness temperature data, while T_b18.7H refers to the horizontal polarization brightness temperature data. The variable F is an indicator function used to determine whether a freeze–thaw cycle occurred on a given day. If a freeze–thaw cycle occurs, its value is 1; if no freeze–thaw cycle occurs, its value is 0. If F > T, it indicates a frozen state; otherwise, it indicates a melting state. To determine whether a specific day experienced a freeze–thaw cycle, both ascending and descending pass data need to be analyzed (the ascending and descending pass data correspond to local times of 13:30 and 1:30, respectively). If the descending pass data indicates that the surface is in a frozen state, while the ascending pass data indicates that the surface has melted, then that day is considered to have experienced a freeze–thaw cycle. By summing the number of days in a year that satisfy this condition, the total number of freeze–thaw cycle days for the year is obtained, known as the annual freeze–thaw cycle days.

Daily average freeze–thaw phase change water

The Daily Average Freeze–Thaw Phase Change Water refers to the amount of water phase change occurring in surface soil during freezing and melting processes within a single day. This indicator captures the latent heat exchange associated with the phase change of soil moisture during the freeze–thaw cycle, as well as the corresponding water migration. It is a critical factor in assessing freeze–thaw erosion intensity, as the phase change of water contributes to the physical disruption and structural degradation of soil particles, thereby intensifying erosion^4,46. During freezing, the expansion of water generates stress within the soil matrix, loosening the structure and elevating erosion risk. This process underscores the importance of understanding water phase changes in the context of freeze–thaw erosion. The AMSR2 sensor onboard the Aqua satellite provides microwave brightness temperature data, which can be utilized to monitor surface temperature and moisture variations. These data enable the inference of freezing and melting conditions of soil moisture, offering valuable insights into freeze–thaw dynamics (Eq. 4).

The specific calculation formula is presented as follows:

$$m_{vpt} = 3.0185\left( {\frac{{T_{bd10.65V} }}{{T_{bd36.5V} }} - \frac{{T_{ba10.65V} }}{{T_{ba36.5V} }}} \right) + 0.0008$$

(4)

In the formula, T_bd10.65V represents the descending vertical polarization brightness temperature at a frequency of 10.65 GHz (DV), while T_bd36.5V corresponds to the descending vertical polarization brightness temperature at a frequency of 36.5 GHz (DV); T_ba10.65V indicates the ascending vertical polarization brightness temperature at a frequency of 10.65 GHz (AV), and T_ba36.5V refers to the ascending vertical polarization brightness temperature at a frequency of 36.5 GHz (AV).

Vegetation coverage

Vegetation coverage refers to the proportion of the Earth’s surface covered by vegetation, typically expressed as a percentage. It serves as a key indicator for assessing ecosystem health and land use conditions. In the context of freeze–thaw erosion intensity evaluation, vegetation coverage is of critical importance. Vegetation stabilizes the soil through its root systems, enhances soil structure, and mitigates the separation and movement of soil particles caused by moisture changes during freeze–thaw cycles. As a result, vegetation plays an essential role in reducing the intensity and progression of freeze–thaw erosion^45,47.

In this study, vegetation coverage from 2017 to 2023 was estimated using the Google Earth Engine (GEE) platform combined with Landsat 8 satellite data (Eqs. 5 and 6). The calculation of vegetation coverage was based on the following formula:

$$NDVI = \frac{{\left( {NIR - RED} \right)}}{{\left( {NIR + RED} \right)}}$$

(5)

$$FVC = \frac{{NDVI - NDVI_{\min } }}{{NDVI_{\max } - NDVI_{\min } }}$$

(6)

In the formula, NIR refers to the reflectance in the near-infrared band (Band 5 of Landsat 8), and RED refers to the reflectance in the red band (Band 4 of Landsat 8); NDVImin and NDVImax represent the minimum and maximum NDVI values for the bare soil and fully vegetated areas in the scene, respectively.

Soil erodibility

Soil type and erodibility are critical factors influencing freeze–thaw erosion intensity. Different soil types exhibit varying degrees of resistance to freeze–thaw erosion due to differences in their physical structure and chemical properties. Clay soils, with fine particles and a dense structure, are more resistant to moisture infiltration and erosion during freeze–thaw processes, demonstrating strong erosion resistance. In contrast, sandy soils, characterized by larger particles and higher porosity, facilitate moisture infiltration, making them more vulnerable to structural damage and intensified erosion under freeze–thaw cycles. Additionally, soils with higher organic matter content possess more stable structures and exhibit greater erosion resistance, whereas soils with lower organic matter content are more prone to loosening and fragmentation during freeze–thaw cycles. The erodibility factor quantifies soil’s sensitivity to freeze–thaw erosion, reflecting its susceptibility to weathering, precipitation, and freeze–thaw processes. Soils with high erodibility are more easily damaged during freeze–thaw cycles, leading to increased soil loss and accelerated erosion processes^29,48,49. In this study, the soil erodibility factor was estimated using the algorithm derived from the Erosion Productivity Impact Calculator (EPIC) model, developed by Rao Liangyi et al. The EPIC model is specifically designed for assessing soil erosion and provides a robust framework for evaluating soil susceptibility in freeze–thaw environments^50,51 (Eq. 7). The specific calculation formula is as follows:

$$\begin{aligned} K & = 0.1317 \times \left\{ {0.2 + 0.3\exp \left[ { - 0.0265SAN\left( {1 - \frac{SIL}{{100}}} \right)} \right]} \right\} \times \left( {\frac{SIL}{{CLA + SIL}}} \right)^{0.3} \\ & \quad \times \left\{ {1 - \frac{0.25C}{{C + \exp (3.72 - 2.95C)}}} \right\} \times \left\{ {1 - \frac{0.7SNI}{{SNI + \exp ( - 5.51 + 22.9SNI)}}} \right\} \\ \end{aligned}$$

(7)

In the equation, $K$ represents the soil erodibility index, with units of $[{\text{t}} \cdot {\text{hm}}^{2} \cdot {\text{h}}/{\text{hm}}^{2} \cdot {\text{MJ}} \cdot {\text{mm}}]$, the larger the value, the stronger the soil erodibility, and vice versa; $SAN$ represents the percentage of sand content in the soil; $SIL$ represents the percentage of silt content in the soil; $CLA$ represents the percentage of clay content in the soil; and $C$ is the organic carbon content in the soil; $SNI = 1 - SAN/100$.

Evaluation method

In the assessment of freeze–thaw erosion intensity, a combination of comprehensive evaluation methods and weighted summation techniques was employed to accurately and scientifically represent the spatial distribution and evolutionary characteristics of freeze–thaw erosion intensity across the study area. Initially, the comprehensive evaluation method was used to integrate multiple evaluation factors, constructing a scientifically sound framework to quantify and assess the freeze–thaw erosion characteristics captured by different factors^16,42.

To determine the weight of each factor and elucidate their relationship with freeze–thaw erosion, both expert scoring and principal component analysis (PCA) were utilized. Expert scoring involved consulting specialists in geomorphology, soil science, and environmental science, who assessed the relevance of each factor based on their professional expertise and field observations. PCA was then applied to identify the factors that accounted for the majority of variance in freeze–thaw erosion within the study area, thereby quantitatively evaluating each factors contribution. By integrating the insights from expert scoring with PCA results, the final weights assigned to each factor were informed by both theoretical and empirical knowledge, while also reflecting the actual variance observed in freeze–thaw erosion. This approach ensured that the weight assignments were scientifically robust and aligned with real-world processes.

Subsequently, a weighted summation method was used to assign relative importance to each factor, based on the determined weights⁴¹. The specific weights of the evaluation factors were as follows: annual freeze–thaw cycle duration (0.16), daily average phase change water volume (0.17), annual average precipitation (0.11), slope (0.09), aspect (0.07), vegetation coverage (0.08), soil type (0.15), and soil erodibility (0.17)^4,40. Using the weighted sum formula (Eq. 8), the standardized values of each factor were multiplied by their respective weights and summed to derive the comprehensive evaluation value for each assessment unit, referred to as the freeze–thaw erosion intensity index.

Finally, ArcGIS software was employed to classify and analyze the intensity indices of each unit. Spatial interpolation techniques were applied to visualize the spatial distribution patterns of freeze–thaw erosion intensity, providing a clear representation of the erosion dynamics across the study area. These results offer a scientific foundation for ecological conservation and land management strategies.

$$I = \frac{{\sum\nolimits_{i = 1}^{n} {W_{i} I_{i} } }}{{\sum\nolimits_{i = 1}^{n} {W_{i} } }}$$

(8)

In the formula, W_i represents the weight corresponding to the evaluation index of each individual factor, I_i denotes the value assigned to each individual factor, and I is the comprehensive evaluation index. The larger the comprehensive evaluation index, the more severe the freeze–thaw erosion.

Freeze–Thaw Erosion Intensity Index

Based on the evaluation results, the Freeze–Thaw Erosion Intensity Index (FTEII) was developed to quantify erosion intensity, with values ranging from 0 to 1. This index represents a continuum of erosion severity, from slight to severe. To analyze the spatial distribution characteristics of freeze–thaw erosion in the high-cold mountainous areas of northwest Yunnan, the FTEII values were classified into six categories, reflecting varying degrees of erosion severity. The classification scheme was adapted from the method proposed by Li Zhiguang et al.¹⁶ from the Ministry of Water Resources in China and is defined as follows: Slight erosion (I ≤ 0.15), Light erosion (0.15 < I ≤ 0.30), Moderate erosion (0.30 < I ≤ 0.50), Intense erosion (0.50 < I ≤ 0.65), Extreme erosion (0.65 < I ≤ 0.75), and Severe erosion (I > 0.75).

Analysis of the spatiotemporal evolution characteristics of freeze–thaw erosion intensity

The analysis of spatiotemporal evolution characteristics is based on time series data of the Freeze–Thaw Erosion Intensity Index (FTEII). By utilizing annual FTEII values, the study investigates interannual variation patterns of freeze–thaw erosion intensity in the study area^25,43,52.

For the temporal analysis, the annual average FTEII was calculated for each year, and linear regression analysis was applied to evaluate the interannual variation trend (Eq. 9). For the spatial analysis, the coefficient of variation (CV) was employed to quantify the spatial variability of freeze–thaw erosion intensity across the study area (Eq. 10). This dual-dimensional approach provides a comprehensive understanding of the temporal dynamics and spatial heterogeneity of freeze–thaw erosion intensity.

$$Y = aX + b$$

(9)

In the equation, $Y$ represents the annual average of freeze–thaw erosion intensity, $X$ denotes the year, and $a$ and $b$ are the regression coefficient and intercept, respectively. By calculating the coefficient of determination $R^{2}$ and the significance level $p$ value, the significance of the interannual variation is determined.

$$CV = \frac{\sigma }{\mu } \times 100\%$$

(10)

In the equation, $\sigma$ represents the standard deviation of freeze–thaw erosion intensity, and $\mu$ denotes its mean value.

Contribution of evaluation factors to freeze–thaw erosion intensity

To further reveal the spatiotemporal correlation between each evaluation factor and freeze–thaw erosion intensity, this study employed the contribution rate method to assess the contribution of each factor to freeze–thaw erosion intensity[5.53]. First, the Comprehensive Evaluation Index (CEI) was calculated, which integrates the effects of various evaluation factors (Eq. 11). The formula for calculating the Comprehensive Evaluation Index is as follows:

$$CEI = \sum\limits_{i = 1}^{n} {w_{i} } \cdot x_{i}$$

(11)

In the formula, $CEI$ represents the Comprehensive Evaluation Index, $w_{i}$ is the weight of the $i$th evaluation factor, $x_{i}$ is the standardized value of the $i$th evaluation factor, and $n$ is the total number of evaluation factors.

Then, the contribution values of each evaluation factor were calculated (Eq. 12). The method for calculating the contribution values is as follows:

$$E_{i} = w_{i} \cdot x_{i}$$

(12)

In the formula, $E_{i}$ represents the contribution value of the $i$th evaluation factor.

Finally, based on the contribution values, the contribution rate of each factor was calculated (Eq. 13). The calculation formula is as follows:

$$CR_{i} = \frac{{E_{i} }}{{\sum\nolimits_{j = 1}^{n} {E_{j} } }} \times 100\%$$

(13)

In the formula, $CR_{i}$ represents the contribution rate of the $i$-th evaluation factor, and $\sum\nolimits_{j = 1}^{n} {E_{j} }$ is the sum of the contribution values of all evaluation factors.

Results

Distribution characteristics of evaluation factors for freeze–thaw erosion intensity

Freeze–thaw erosion intensity in high-cold mountainous regions is influenced by a combination of factors, including temperature, precipitation, topography, and vegetation. High-altitude areas with low temperatures and frequent freeze–thaw cycles exhibit higher erosion intensity, particularly on steep slopes and in regions with sparse vegetation. In contrast, low-altitude areas with dense vegetation coverage experience relatively weaker freeze–thaw erosion. A general spatial distribution pattern is observed: erosion intensity increases with elevation, steeper slopes, and reduced vegetation coverage.

The annual freeze–thaw cycle duration in northwestern Yunnan exhibits significant spatial variability. High-altitude regions are characterized by pronounced temperature fluctuations, resulting in frequent freeze–thaw cycles. The average AFTCD is approximately 150 days, primarily occurring from October to April of the following year. These cycles are concentrated in the mountainous and plateau regions of the northern and western study areas, while low-altitude regions experience shorter freeze–thaw cycles (Fig. 3a).

Fig. 3

Spatial Distribution Maps of Evaluation Factors for Freeze–Thaw Erosion Intensity in the High-Cold Mountainous Areas of Northwestern Yunnan.

DAFTPCW demonstrates a spatial distribution similar to AFTCD. High-altitude regions, with frequent freeze–thaw cycles, exhibit greater phase-change water volumes, ranging from 0.03 to 0.09 m³/m³. Areas with DAFTPCW values exceeding 0.05 m³/m³ account for 47.30% of the study area, primarily concentrated in the upper reaches of the Jinsha, Lancang, and Nu Rivers, as well as along steep slopes and mountainous regions. In contrast, low-altitude and gently sloping areas show lower DAFTPCW values, ranging from 0.03 to 0.05 m³/m³ (Fig. 3b).

Precipitation within the study area exhibits a distinct north–south gradient. The southern regions, influenced by monsoons, receive higher precipitation (750–1897.48 mm/yr), while the northern and central regions experience lower precipitation levels (541.58–750 mm). Precipitation impacts soil moisture content, thereby influencing freeze–thaw erosion intensity (Fig. 3c).

Slope is a significant factor affecting freeze–thaw erosion intensity. Steep slopes, primarily concentrated in the central and northern mountainous regions, account for 52.64% of the study area, resulting in higher erosion intensity due to poor soil stability. Gentler slopes, with average gradients below 15°, comprise 19.76% of the area and exhibit weaker erosion intensity (Fig. 3d).Slope aspect also plays a critical role. Sunny slopes, with greater solar exposure, experience faster melting and more frequent freeze–thaw cycles, leading to higher erosion intensity. In contrast, shady slopes, with less sunlight and slower thawing processes, exhibit reduced erosion intensity (Fig. 3e).

Vegetation plays a protective role in reducing freeze–thaw erosion by stabilizing soil and minimizing exposure to freeze–thaw cycles. Southern regions with higher vegetation coverage show lower erosion intensity, whereas the northern and high-altitude areas with sparse or absent vegetation experience greater erosion intensity (Fig. 3f).

The main soil types in northwestern Yunnan—southern paddy soil, permafrost, dark brown soil, red soil, and yellow–brown soil—exhibit varying responses to freeze–thaw processes (Figs. 3g). Permafrost soils experience the highest erosion intensity due to frequent freeze–thaw cycles, while southern paddy soil, though resistant, shows structural damage during winter freeze–thaw events. Dark brown soil demonstrates moderate erosion resistance but is vulnerable under extreme conditions. Red soil and yellow–brown soil, with loose structures, are highly prone to erosion, particularly in steep or high-altitude areas. The soil erodibility factor (K value) is higher in high-altitude regions, leading to increased erosion intensity, while lower K values in low-altitude areas correspond to stronger erosion resistance (Fig. 3h).

Overall, the spatial distribution of freeze–thaw erosion intensity is shaped by the interplay of multiple environmental factors, including climate, topography, vegetation, and soil properties. The distinct combinations of these factors create significant spatial variations in erosion intensity within the study area. These findings provide a foundation for further analysis of the temporal and spatial evolution characteristics of freeze–thaw erosion.

Spatial distribution of freeze–thaw erosion intensity

In the high-cold mountainous areas of northwestern Yunnan, the average freeze–thaw erosion intensity over the past 7 years was predominantly characterized by moderate erosion, with an erosion area of approximately 10,600 km², accounting for 24.44% of the total erosion zone. The spatial distribution of freeze–thaw erosion intensity varied significantly across the study area.

High-altitude regions (above 3000 m) exhibited notably higher freeze–thaw erosion intensities, encompassing approximately 9320 km² or 21.38% of the total erosion zone. These areas, primarily located in the northwest and northeast of the study region (Fig. 4), experienced frequent freeze–thaw cycles, resulting in weakened soil structures. Erosion intensities in these regions were predominantly classified as “severe” and "extreme."

Fig. 4

Spatial Distribution of Freeze–Thaw Erosion Intensity in the High-Cold Mountainous Areas of Northwestern Yunnan from 2017 to 2023 and the Average Annual Intensity.

In mid-altitude regions (2000–3000 m), freeze–thaw erosion intensities ranged from “moderate” to "intense," accounting for approximately 17,400 km² or 39.85% of the erosion zone. These areas were mainly distributed in the central and eastern parts of the study area, with fewer sporadic occurrences in the north and west. While mid-altitude regions were affected by freeze–thaw cycles, relatively high vegetation coverage played a protective role, mitigating the extent of erosion.

Low-altitude regions (below 2000 m) exhibited weaker freeze–thaw erosion intensities, characterized primarily as “slight” to “light” erosion. These areas accounted for 38.78% of the erosion zone. The soils in low-altitude regions were relatively moist, and the impact of freeze–thaw cycles was minimal, contributing to lower erosion intensity.

Spatial analysis revealed that high-value areas of freeze–thaw erosion intensity were concentrated in valleys and slopes in the northern part of the study area, particularly on south-facing slopes. Due to higher sunlight exposure and greater temperature fluctuations, erosion intensity on south-facing slopes was significantly higher than on north-facing slopes. These spatial distribution patterns reflect the combined influence of natural environmental factors and human activities. Strengthened monitoring and targeted management strategies are recommended for these high-risk areas to mitigate the increasing trend of freeze–thaw erosion.

Temporal and spatial evolution of freeze–thaw erosion intensity

From 2017 to 2023, freeze–thaw erosion intensity in the high-cold mountainous areas of northwestern Yunnan exhibited significant temporal and spatial variations (Table 1). Overall, the region experienced an increasing trend in freeze–thaw erosion intensity over the past seven years, with areas of slight and mild erosion transitioning to moderate and intense erosion. Areas with slight, mild, and extreme erosion showed a significant decreasing trend (P < 0.555), whereas areas with moderate, intense, and severe erosion demonstrated a notable increase (P > 0.8). Notably, the extent of mild and moderate erosion expanded considerably after 2020, while slight and intense erosion areas declined sharply (Fig. 5). Over the seven-year period, slight, mild, and moderate erosion areas exhibited relatively high stability between two observation periods. However, other regions displayed varying degrees of increased or decreased erosion intensity, reflecting the complex environmental dynamics at play in this region (Fig. 6). Key influencing factors include climate change, vegetation cover, topographical variations, and human activities. The temporal and spatial evolution of each erosion intensity category is analyzed in detail below.

Table 1 Areas of different freeze–thaw erosion intensities in the cold mountainous areas of Northwestern Yunnan from 2017 to 2023 (10³ km²).

Fig. 5

Changes in Freeze–Thaw Erosion Area in the Cold Mountainous Areas of Northwestern Yunnan from 2017 to 2023.

Fig. 6

Spatial Distribution of Changes in Freeze–Thaw Erosion Intensity in the Cold Mountainous Areas of Northwestern Yunnan from 2017 to 2023.

Slight freeze–thaw erosion intensity exhibited a consistent year-by-year decline, decreasing from 7.29 × 10³ km² in 2017 to 5.71 × 10³ km² in 2023. This trend likely reflects the effects of climate warming, as rising temperatures increased the frequency of freeze–thaw cycles, causing areas with slight erosion to evolve into regions with more intense erosion. Additionally, a sharp decrease in slight erosion areas was observed after 2019, suggesting that significant environmental changes, such as more frequent extreme weather events, may have had an intensified impact on the regional ecosystem during this period.

Light freeze–thaw erosion intensity exhibited relatively large fluctuations over the study period. The area of light erosion decreased from 9.17 × 10³ km² in 2017 to 8.84 × 10³ km² in 2019, followed by a surge to 1.06 × 10⁴ km² in 2020 and peaking at 1.38 × 10⁴ km² in 2021. These fluctuations were likely influenced by cyclical variations in annual precipitation and temperature. The significant increase in light erosion areas between 2020 and 2021 may be attributed to higher rainfall during this period, which enhanced soil moisture levels and expanded freeze–thaw erosion zones. Furthermore, vegetation recovery may have converted some moderately eroded areas into light erosion zones, contributing to the observed peak.

Moderate freeze–thaw erosion intensity reached its lowest value of 1.02 × 10⁴ km² in 2019 but gradually increased to 1.13 × 10⁴ km² by 2023. This pattern reflects the influence of interannual climate variability on freeze–thaw erosion. The decline in 2019 could be associated with cooler temperatures and lower precipitation, which temporarily reduced freeze–thaw activity. However, as climate conditions normalized, particularly with improvements in vegetation cover and soil structure, the area of moderate erosion began to recover after 2020, indicating a complex interplay between climatic and ecological factors.

Intense freeze–thaw erosion intensity exhibited significant volatility during the study period. In 2019, the area affected by intense erosion surged to 9.37 × 10³ km², a substantial increase compared to the 6.92 × 10³ km² recorded in 2018. This abrupt rise was likely driven by extreme climatic events in 2019, such as severe cold spells that triggered frequent freeze–thaw cycles, further degrading soil structures. Although the area of intense erosion declined in 2020 and 2021, it remained at relatively high levels, suggesting that freeze–thaw erosion in high-cold mountainous regions is highly sensitive to and fluctuates sharply in response to climate change.

Extreme and severe freeze–thaw erosion intensities also experienced notable fluctuations. The area of extreme erosion was 6.04 × 10³ km² in 2017, fluctuating before peaking at 6.92 × 10³ km² in 2020. A similar pattern was observed for severe erosion, with areas increasing during periods of intensified freeze–thaw activity. These variations were likely influenced by the increased frequency of freeze–thaw cycles and the intensification of extreme climatic events. Although both extreme and severe erosion areas showed a decline in 2021, reflecting potential natural ecological recovery, they fluctuated again in 2022 and 2023, highlighting the persistent instability of regional climate conditions and its impact on freeze–thaw erosion intensity.

The spatiotemporal evolution of freeze–thaw erosion intensity is the result of complex interactions between natural and human factors. Climate change, particularly fluctuations in temperature and precipitation, had a direct and significant impact on the interannual variability of freeze–thaw erosion. High-cold mountainous areas, being highly sensitive to climate change, experienced exacerbated freeze–thaw erosion during frequent extreme weather events. Vegetation cover changes also played a critical role, as vegetation recovery helped slow the expansion of light and moderate erosion zones. However, in areas with severe and extreme erosion, vegetation loss or human disturbance contributed to an increase in erosion intensity. Topographical differences, including elevation and slope aspect, further influenced the spatial distribution of erosion. South-facing slopes and complex high-altitude terrains were particularly prone to intense freeze–thaw erosion due to higher exposure to sunlight and temperature fluctuations.

To address the risks posed by freeze–thaw erosion, it is essential to adopt comprehensive management strategies that integrate climate adaptation, vegetation restoration, and regional environmental characteristics. Such strategies can help mitigate the ecological challenges posed by freeze–thaw erosion and promote sustainable development in high-cold mountainous regions.

The impact of evaluation factors on changes in freeze–thaw erosion intensity

The spatial and temporal variations in freeze–thaw erosion intensity are driven by a combination of natural and anthropogenic factors. In the northwestern Yunnan region, annual freeze–thaw cycle duration (AFTCD) and daily average phase-change water volume (DAFTPCW) are critical climatic factors influencing erosion intensity. Prolonged freeze–thaw cycles and increased phase-change water volumes disrupt soil structure and destabilize surface soils. Transition analysis of freeze–thaw erosion intensity between two periods (2017–2019 and 2021–2023) revealed significant changes in areas with light and moderate erosion, particularly transitions from light to moderate (2.82 × 10³ km²) and moderate to intense erosion (2.38 × 10³ km²). Climate warming, by increasing the frequency of freeze–thaw cycles, exacerbates erosion risks. Additionally, annual precipitation, another key climatic factor, influences freeze–thaw erosion intensity through its effects on moisture accumulation and soil water content. Increased precipitation accelerates soil degradation during freeze–thaw processes, particularly in areas with intense and extreme erosion levels. The transitions from intense to extreme erosion (1.43 × 10³ km²) and from extreme to severe erosion (1.51 × 10³ km²) underscore the role of high precipitation in intensifying erosion (Table 2).

Table 2 Transfer matrix of freeze–thaw erosion intensity in the high-cold mountain areas of Northwestern Yunnan from 2017 to 2023 (10³ km²).

The northwestern Yunnan region is part of the Three Parallel Rivers area, where river incision is particularly pronounced. The unique topography and rugged terrain significantly impact the stability of soils and surface materials. Soils on steep slopes are highly susceptible to freeze–thaw cycles, leading to elevated erosion intensity. Changes in erosion intensity at moderate and higher levels (e.g., from moderate to intense or from intense to extreme) highlight the compounding effects of slope and aspect on severe erosion. Sunny slopes, with greater sunlight exposure, experience more frequent freeze–thaw processes, further amplifying erosion intensity. In contrast, high vegetation coverage mitigates erosion by reducing surface moisture evaporation, slowing down freeze–thaw cycles, and stabilizing the soil. Notably, transitions from light to slight erosion (3.31 × 10³ km²) indicate that vegetation recovery has effectively weakened erosion intensity in certain areas.

Soil type and soil erodibility further determine soil stability under freeze–thaw conditions. In northwestern Yunnan, the soil is characterized by loose structures, fine particles, fractured rocks, and pronounced weathering, rendering it highly vulnerable to erosion, especially in areas with steep slopes and high precipitation. Regions with higher soil erodibility, such as those experiencing intense and extreme erosion, exhibit significantly elevated erosion intensity.

Moreover, the interaction between annual precipitation and freeze–thaw cycle duration is a critical driver of freeze–thaw erosion intensity. Increased precipitation enhances soil moisture content and prolongs freeze–thaw cycles by supplying additional water to surface and subsurface soils. This interaction accelerates soil structure breakdown and increases surface instability. For instance, regions with high precipitation and frequent freeze–thaw cycles experience significantly escalated erosion intensity, as the effects of these factors are mutually reinforcing. Additionally, areas with steep slopes and fine soil particles are more vulnerable to severe erosion due to the combined effects of prolonged freezing, increased thawing water volumes, and precipitation-driven soil saturation.

Future research could focus on quantifying these interactions under varying climatic and topographic conditions to better understand the complex dynamics of freeze–thaw erosion and to develop more effective management strategies.

Discussion

As of now, we have completed the assessment of freeze–thaw erosion intensity and analyzed its spatiotemporal evolution characteristics in the high-cold mountainous areas of northwestern Yunnan, with a focus on the changing trends and influencing factors of freeze–thaw erosion intensity. However, several issues warrant further discussion.

Firstly, the study utilized data from 2017 to 2023, resulting in a relatively short time series. This timeframe was selected based on the availability of data and the need to capture recent changes in freeze–thaw erosion characteristics in the study area. Although the short duration provides valuable insights into the region’s response to recent climate variability, the significant seasonality and interannual variability of freeze–thaw erosion in high-cold mountainous areas suggest that extending the time series could offer a more comprehensive understanding of long-term trends and their stability.

The annual average freeze–thaw cycle days (AFTCD) and daily average phase-change water volume (DAFTPCW), derived from surface brightness temperature data, served as core climatic drivers of freeze–thaw processes. By integrating these with other temporally varying factors such as vegetation cover, annual precipitation, and topography, the study enhanced the understanding of how changes in these factors during 2017–2023 impacted soil stability and erosion intensity. Notably, variations in vegetation coverage and precipitation played significant roles in influencing freeze–thaw erosion intensity by directly affecting soil structure and moisture content.

The study also considered the influence of soil factors, selecting soil type and soil erodibility as key variables. This integrated approach, which combines soil properties, is relatively rare in existing research. The application of the algorithm proposed by Rao Liangyi et al.⁵¹ within the EPIC model to estimate soil erodibility factors allowed for a more accurate reflection of the erosion susceptibility of different soil types under freeze–thaw conditions. By incorporating soil erodibility into the evaluation, this study deepened the understanding of the freeze–thaw erosion process, significantly enhancing the applicability and scientific validity of the model compared to traditional studies that focus primarily on climate and topography.

The study employed a spatial resolution of 250 m, effectively capturing the spatial heterogeneity of the complex topography in high-cold mountainous areas. To ensure consistency across datasets, resampling and interpolation methods were applied to standardize climate, topography, soil, and land cover data. For areas with missing data, approaches such as nearby station interpolation and geographically weighted regression were used. These methods ensured the spatiotemporal continuity and reliability of the data, reducing potential errors and enhancing the credibility of the results.

Field investigations were conducted to validate the assessment results (Fig. 6). In areas with high freeze–thaw erosion intensity, low vegetation coverage, exposed bedrock, and widespread rock debris slopes were observed. In contrast, areas with low freeze–thaw erosion intensity exhibited dense vegetation and minimal visible damage, with high ground cover. Regions with moderate freeze–thaw erosion intensity displayed a mix of features, including some rock debris and exposed bedrock, alongside slow vegetation growth. Once the ecological environment in these areas is damaged, the recovery of vegetation is slow or highly challenging.

Finally, by analyzing the contribution rates of each factor to the observed changes in freeze–thaw erosion intensity (e.g., unchanged, weakened, or strengthened), the study revealed the mechanisms through which different factors influence freeze–thaw erosion (Table 3). This comprehensive analysis offers insights into the driving forces behind freeze–thaw erosion and provides a foundation for developing strategies to mitigate its impacts in high-cold mountainous regions.

Table 3 Average contribution rates of evaluation factors to freeze–thaw erosion intensity in the high-cold mountain areas of Northwestern Yunnan (%).

The dominant role of soil erodibility

From the overall data, soil erodibility demonstrated the highest contribution rate across all groups, particularly in areas where erosion intensity remained unchanged or weakened. Contribution rates ranged from 25 to 30%, with a maximum value of 30.17%. These findings underscore the significant influence of soil physical properties on freeze–thaw erosion intensity. Higher soil erodibility indicates that soil particles are more susceptible to damage and transportation under the effects of freeze–thaw cycles. Consequently, soil erodibility emerges as a critical factor in the freeze–thaw erosion process, especially in regions with loose soil particles that are highly prone to erosion. This highlights the pivotal role of soil erodibility in shaping erosion intensity and its variability across different areas.

Significant increase in daily average phase-change water volume

As a key evaluation factor, the daily average phase-change water volume showed a significant increase in areas with enhanced erosion intensity, with its contribution rate peaking at 21.62%. This indicates that the frequent phase changes of water during freezing and thawing processes amplify the destructive effects of water on soil. In high-cold mountainous regions, these phase changes—characterized by freezing expansion and thawing contraction—exert substantial mechanical stress on the surface and soil, leading to intensified freeze–thaw erosion phenomena. As a result, the increase in daily average phase-change water volume has emerged as one of the primary driving forces behind the intensification of erosion intensity.

Gradual increase in annual average freeze–thaw cycle duration

The contribution rate of annual average freeze–thaw cycle duration exhibited a clear upward trend in areas where erosion intensity remained unchanged or increased. In regions with unchanged erosion intensity, the contribution rate gradually increased from 8.15% to 20.71%, while in areas with enhanced intensity, the highest contribution rate reached 21.02%. This trend suggests that a prolonged freeze–thaw cycle duration corresponds to a higher frequency of freeze–thaw actions, with each cycle causing incremental damage to the soil and, in turn, intensifying erosion. Consequently, the extension of annual average freeze–thaw cycle duration is positively correlated with freeze–thaw erosion intensity. This correlation is particularly evident in regions with frequent freeze–thaw cycles, where the cumulative impact of this factor becomes increasingly pronounced.

Relative stability of other factors

In contrast, the contribution rates of factors such as slope direction, slope, and vegetation coverage exhibited relatively small fluctuations across different intensity change areas. For instance, the contribution rate of slope direction consistently ranged between 5 and 9%, while the contribution rate of slope remained similarly stable. Although vegetation coverage is an important environmental factor, its direct impact on freeze–thaw erosion intensity appeared to be relatively limited. This may be attributed to the indirect nature of its influence, primarily through mechanisms such as soil coverage and reduction of surface runoff. Consequently, vegetation coverage does not emerge as the most critical factor driving freeze–thaw erosion intensity in high-cold mountainous areas.

Analysis of contribution rates of special factors

In the results of the factor contribution rate calculations, certain factors, such as soil erodibility and daily average phase-change water volume, exhibited a contribution rate of zero in areas where erosion intensity weakened. This suggests that during periods of reduced erosion intensity, the impact of soil properties or water phase changes on freeze–thaw erosion was minimal, while other factors played a more dominant role. These findings highlight the need for further in-depth research to uncover the specific geographical or climatic conditions and potential influencing mechanisms underlying these anomalies. Such investigations could provide valuable insights into the dynamics of freeze–thaw erosion under varying environmental conditions.

Comparison with international studies

Compared to related studies worldwide, this research introduces certain innovations in analyzing the spatiotemporal evolution characteristics of freeze–thaw erosion and its influencing factors in high-cold mountainous areas. For instance, Sadeghi et al.¹⁰ identified the frequency of freeze–thaw cycles and soil type as key factors influencing freeze–thaw erosion in high-cold regions of North America, findings that align with the results of this study. However, while Sadeghi et al. primarily focused on the effects of individual factors, this research adopts a more comprehensive approach by incorporating multiple factors, including soil erodibility, which provides a deeper understanding of the erosion process.

Similarly, Gao⁵³ highlighted the buffering effect of vegetation cover on freeze–thaw erosion in the European Alps, which corroborates our findings on the critical role of vegetation cover changes in mitigating freeze–thaw erosion intensity.

In other high-cold regions of Asia, such as Siberia and the Tibetan Plateau, studies have demonstrated that climate change, particularly shifts in precipitation patterns, significantly affects freeze–thaw erosion⁵⁴. Building on these findings, this study further reinforces these findings by identifying annual average precipitation as a key factor influencing the intensity of freeze–thaw erosion, thereby validating the universal significance of climatic factors in freeze–thaw erosion processes across diverse geographical contexts.

Furthermore, global research has frequently emphasized the potential impacts of human activities on freeze–thaw erosion, such as land use changes and vegetation management¹⁸. Although this study predominantly focuses on natural factors, changes in vegetation cover also indirectly reflect the influence of human activities. Future research could delve deeper into the interplay between human activities and freeze–thaw erosion, providing more robust insights for developing comprehensive and sustainable management strategies.

Limitations of the study and future directions

Although this study has made significant progress in assessing freeze–thaw erosion intensity and analyzing its spatiotemporal evolution, certain limitations remain. Firstly, the relatively short data time span constrains the ability to observe long-term trends. Future research should incorporate longer time series data to better capture the dynamics and stability of freeze–thaw erosion over extended periods. Secondly, while this study primarily relies on remote sensing data and GIS technologies, integrating field observation data in future studies could enhance the accuracy and robustness of the assessments.

Furthermore, international research highlights the substantial influence of microclimate changes and extreme weather events on freeze–thaw erosion processes^55,56. These dynamic factors are particularly relevant in high-cold mountainous regions where localized climatic conditions can vary significantly. Future research should explore the impacts of these factors to gain a deeper understanding of their role in shaping freeze–thaw erosion intensity and patterns. By addressing these limitations, subsequent studies can provide more comprehensive insights into the mechanisms driving freeze–thaw erosion and support the development of targeted management strategies.

Conclusion

This study evaluated the freeze–thaw erosion intensity in the high-cold mountainous areas of Northwestern Yunnan from 2017 to 2023, revealing the spatiotemporal evolution characteristics of freeze–thaw erosion and its influencing factors, and came up with the following main conclusions:

• 1. Freeze–thaw erosion intensity was influenced by multiple factors, including climate, topography, soil type, and vegetation cover. The study found that high-altitude areas experienced higher erosion intensity due to low temperatures and frequent freeze–thaw cycles, especially in steep and sparsely vegetated regions, exhibiting a clear spatial distribution pattern. In contrast, low-altitude areas with good vegetation cover showed weaker freeze–thaw erosion intensity.

• 2. Over the past seven years, moderate freeze–thaw erosion has predominated in Northwestern Yunnan, with the erosion area accounting for 24.44% of the total area. High-altitude areas exhibited significantly higher freeze–thaw erosion intensity, primarily concentrated in the northwestern and northeastern regions of the study area, while moderate to severe erosion predominated in mid-altitude areas. Low-altitude regions showed weaker erosion intensity, mainly exhibiting slight to mild erosion.

• 3. Between 2017 and 2023, significant spatiotemporal variations in freeze–thaw erosion intensity have been observed. The area of slight erosion decreased annually, while the areas of mild and moderate erosion significantly increased after 2020, reflecting the combined effects of climate change, vegetation cover, and human activities. The volatility in areas of intense and extreme erosion revealed the impact of extreme weather events on freeze–thaw erosion intensity.

• 4. Climate change was the primary factor influencing interannual variations in freeze–thaw erosion intensity, with high-cold mountainous areas being sensitive to fluctuations in temperature and precipitation. Changes in vegetation played an important role in mitigating mild and moderate erosion, whereas vegetation loss or human interference in areas of severe and extreme erosion exacerbated the increase in erosion intensity. In summary, the spatiotemporal evolution of freeze–thaw erosion intensity results from the interaction of various natural and human factors. Future management should integrate climate change, vegetation management, and regional environmental characteristics to effectively address the risks posed by freeze–thaw erosion.