Method for identifying hydrological characteristics under influence of frozen soil degradation

Abstract

The invention discloses a method for identifying hydrological characteristics under the influence of frozen soil degradation. The method comprises the following steps: acquiring and preprocessing layered soil phase state data and hydrological flux data; calculating a depth-time-delay correlation matrix in the multi-year time-delay set, and screening out a significant depth; performing single inflection point segmented regression by taking the phase state data of the current year as an independent variable on each depth passing the test, and synchronously estimating a threshold value and post-threshold sensitivity of the depth; the three elements of time delay, threshold and sensitivity are normalized and synthesized into a hydrological response index in an equal weight mode, and the depth of a master control layer is judged with the maximum absolute value of the threshold sensitivity as the criterion. The method overcomes the defects that in the prior art, only the thermal state is reflected, the response strength is difficult to quantify, and the key depth is difficult to identify, achieves automatic and standardized diagnosis of the frozen soil-hydrological coupling structure, and can be used for frozen soil degradation monitoring and cold region water resource management.

Description

Technical Field

[0001] This invention relates to the fields of geoinformation science and hydrological monitoring technology, specifically to a method for identifying hydrological characteristics under the influence of permafrost degradation. This method is used to identify the coupling structure between soil ice changes and hydrological fluxes (runoff, evapotranspiration) under permafrost degradation conditions, and to determine the depth and spatial distribution of the main control layer. Background Technology

[0002] Due to global warming, permafrost in high-altitude and cold regions continues to degrade and local thaw zones are expanding. Shallow-to-deep hydraulic channels and recharge structures are being reorganized, resulting in longer hydrological process time lags, increased baseflow during winter / dry seasons, and altered runoff and evapotranspiration distribution patterns. These changes directly impact water resource security (such as "blue water" runoff resources) and ecosystem stability (such as "green water" evapotranspiration support) in cold regions, necessitating a unified diagnostic method capable of cross-regional comparisons and operational applications.

[0003] Currently, the commonly used methods for assessing the impact of permafrost on hydrological processes mainly include the following categories:

[0004] (1) Empirical models based on ground temperature or freezing depth only reflect changes in thermal state and lack comprehensive diagnostic capabilities for key coupling features such as hydrological response intensity, time delay, trigger threshold and main control layer depth.

[0005] (2) Based on single-layer or single-delay correlation / regression analysis, the linkage effect and time autocorrelation of multiple depths and multiple time delays are ignored, which can easily lead to one-sided conclusions and lack a unified index system to quantify the coupling strength.

[0006] (3) Numerical simulation based on distributed hydrological models has many sensitive parameters and a large amount of computation, making it difficult to directly and clearly invert the depth of the main control layer with clear physical meaning from the simulation results.

[0007] (4) The state estimation of multi-source remote sensing or reanalysis data mainly focuses on quantitative restoration and lacks an information diagnostic framework that integrates multiple elements and outputs standardized spatial products and structured classification results.

[0008] Therefore, it is necessary to propose an information diagnostic method that integrates multiple time delays, multiple depths, and multiple parameters. Summary of the Invention

[0009] The purpose of this invention is to propose a method for identifying hydrological characteristics under the influence of permafrost degradation. This method can simultaneously consider time lag, threshold activation, and post-threshold sensitivity within a unified framework, construct a hydrological response index that can be compared across regions, and determine the depth of the main control layer with clear physical meaning based on a rigorous statistical significance test. Ultimately, a standardized spatial product is formed to support rapid identification, regional comparison, and operational applications in the context of permafrost degradation.

[0010] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0011] A method for identifying hydrological characteristics under the influence of permafrost degradation, the method comprising the following steps:

[0012] S1. Acquire time-series data of stratified soil phase state and hydrological flux to characterize the soil freezing phase state, and perform quality control and alignment processing on the acquired data.

[0013] S2, within a pre-defined multi-year time lag set, calculates the statistical relationship between time-series data of stratified soil facies and hydrological flux based on depth, and performs significance tests considering time autocorrelation and multiple comparison corrections on the obtained statistical relationships to screen out significant depth-time lag pairs to form a significant candidate set. ;

[0014] S3, for salient candidate sets At each depth, using soil phase data for that depth in each year as the independent variable and hydrological flux as the dependent variable, a piecewise linear model with a single inflection point is fitted to simultaneously estimate the threshold corresponding to that depth. Postthreshold sensitivity ;

[0015] S4, the time delay, threshold and post-threshold sensitivity at each depth are transformed to the [0,1] interval by a normalization mapping that satisfies monotonicity, and the mapped time delay, threshold and post-threshold sensitivity are combined with equal weights to obtain the hydrological response index;

[0016] S5: Within the depth set where both the threshold and post-threshold sensitivity pass the significance test, select the depth with the largest absolute value of post-threshold sensitivity as the master control layer, and output the hydrological response index and the spatial distribution of the master control layer.

[0017] Further, in step S1, the stratified soil phase time series data includes soil ice content or equivalent phase indicator of soil ice content; the equivalent phase indicator includes unfrozen water conversion based on ground temperature, freeze-thaw index based on active or passive microwave, and phase index based on dielectric constant or resistivity; the hydrological flux time series data includes at least runoff and evapotranspiration.

[0018] Furthermore, in step S1, the process of performing quality control and alignment on the acquired data includes the following steps:

[0019] The raw data related to stratified soil phase and hydrological flux were extracted from distributed hydrological permafrost models, remote sensing or reanalysis products and ground station observation data.

[0020] Linear interpolation or Kalman smoothing was used to perform missing interpolation on the original data, and neighborhood or assimilation strategies were used to perform long missing interpolation on the original data. The imputed original data were then subjected to anomaly identification and removal, time scale alignment, spatial resampling, detrending and standardization, and corresponding daily or monthly stratified soil phase sequence and hydrological flux sequence were generated.

[0021] The stratified soil phase sequence and hydrological flux sequence for each day or month are aggregated by year and then separated by season to obtain the annual seasonal sequence.

[0022] Step S2 further includes:

[0023] Within a pre-defined multi-year time lag set, the statistical relationship between the time series data of stratified soil phase and the time series data of hydrological flux is calculated according to depth. The obtained statistical relationship is then subjected to a significance test considering time autocorrelation and multiple comparison correction to screen out significant depth-time lag pairs.

[0024] To overcome the time autocorrelation effect present in multi-year series, annual block permutation test or annual cyclic shift test is performed on each depth-delay pair in the depth-delay pair set to obtain the corresponding significance results.

[0025] Multiple comparison corrections are performed on the significance results obtained under multiple depths and time lags to control the overall misclassification rate;

[0026] The significant results are filtered out by a depth-delay combination of correction thresholds to form a significant candidate set, and a depth-delay significance mask reflecting the multi-year coupling structure is generated based on the significant candidate set.

[0027] Step S3 further includes:

[0028] For each depth in the significant candidate set, a piecewise linear model with a single inflection point containing a pre-threshold segment and a post-threshold segment is constructed. By searching for the inflection point position within the range of independent variable variation, the overall fitting error of the model is minimized, thereby determining the threshold corresponding to that depth.

[0029] After determining the threshold, linear fitting is performed on the pre-threshold segment and the post-threshold segment to obtain the linear response coefficients of the two segments. The difference between the coefficients of the two segments is used to characterize the degree of response enhancement after the threshold, thus obtaining the post-threshold sensitivity.

[0030] The confidence intervals of the threshold and post-threshold sensitivity corresponding to the depth are evaluated. When the threshold position is unstable or the sensitivity is not significant, the depth is marked as undetectable and distinguished by a mask in subsequent processing.

[0031] Step S4 further includes:

[0032] The optimal time delay, threshold, and post-threshold sensitivity corresponding to each depth are respectively input into a preset monotonic normalization function to transform them to the same normalization interval. The upper bound parameter of the monotonic normalization function is determined based on the statistical upper limit of multi-year samples.

[0033] The normalized components of the mapped optimal time delay, threshold, and post-threshold sensitivity are combined according to the equal weighting rule to form a hydrological response index that represents the multi-year scale permafrost-hydrological coupling intensity.

[0034] Among them, the normalized component of the optimal delay is used to reflect the coupling delay characteristics, the normalized component of the threshold is used to reflect the phase degree required for triggering, and the normalized component of the post-threshold sensitivity is used to reflect the response strength after the threshold.

[0035] The optimal time lag is the weighted average, median, or mode of the significant time lag set, and the upper bound of the normalization function is determined based on the upper quantile of the multi-year sample of the corresponding season.

[0036] Step S5 further includes:

[0037] Depths with significant threshold and post-threshold sensitivity are included in the available depth set; the master control layer depth is determined by comparing the absolute values ​​of post-threshold sensitivity at each depth; when multiple depths simultaneously meet the maximum condition, the one with higher goodness of fit is selected first, followed by the one with shallower depth or higher frequency of occurrence in multi-year sequences; the spatial distribution, significance mask, and uncertainty range of the master control layer depth are output.

[0038] Furthermore, the method also includes:

[0039] The multi-year series is divided into an early and late stage based on the detection of change points based on geothermal anomalies, phase anomalies or other driving factors; the hydrological response index and the depth of the main control layer are calculated for the early and late stages respectively; by comparing the differences between the two stages, the change in coupling strength and the migration of the depth of the main control layer are obtained to reflect the changes in the coupling structure during the permafrost degradation process.

[0040] The method for identifying hydrological characteristics under the influence of permafrost degradation of this invention can identify the coupling law between soil phase changes and hydrological fluxes (runoff or evapotranspiration) on a multi-year (interannual) scale and according to the four seasons. At the same time, it measures three key elements: multi-year time lag, threshold activation, and post-threshold sensitivity, constructs a multi-year hydrological response index that can be compared across regions, and determines the depth of the dominant layer in each season. This overcomes the shortcomings of existing technologies, such as only reflecting thermal state, only describing single-layer or single-time lag relationships, and difficulty in inverting the dominant depth. It realizes the automatic identification of multi-year memory effects under the background of permafrost degradation, enhances cross-regional comparability and engineering usability, realizes the automatic identification and diagnosis of hydrological response structure in permafrost areas, and provides standardized expression and operational output for it.

[0041] This invention can be embedded in the back-end computing module of a hydrological monitoring system and can run on a geographic information processing platform or a high-performance computing node, thereby realizing automated and standardized diagnosis of hydrological-thermal coupling structures under the background of permafrost degradation.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] First, it has strong comprehensive diagnostic capabilities: by normalizing and integrating the three key physical quantities of time delay, threshold and post-threshold sensitivity, a single and comprehensive hydrological response index (HRI) is generated, realizing a multi-dimensional comprehensive diagnosis of the permafrost-hydrological coupling relationship and overcoming the limitation of existing methods that only focus on a single factor.

[0044] Second, the physical mechanism is clear and the results are highly interpretable: the depth of the main control layer is determined by the criterion of "maximum absolute value of post-threshold sensitivity". The physical meaning is clear and it can accurately identify the key soil layer most sensitive to hydrological changes, providing direct physical insights for understanding and predicting hydrological processes under permafrost degradation.

[0045] Third, it has strong anti-interference ability and high reliability of results: By adopting annual block permutation / cyclic shift and combining it with false detection rate (FDR) control for statistical significance testing, it effectively overcomes the false significance problems caused by time series autocorrelation and multiple comparisons, and ensures the statistical robustness of diagnostic conclusions.

[0046] Fourth, the method is robust and easy to promote and apply: the core algorithm (matrix correlation, piecewise regression) is computationally efficient and insensitive to the setting of key parameters. Stable results can be obtained in different regions without complex optimization, and it has good transferability and business application potential.

[0047] Fifth, it has strong data adaptability: the method framework does not rely on a single soil phase data source, and allows the use of soil ice content and its various equivalent indicators (such as unfrozen water content, remote sensing freeze-thaw index, etc.) as input, supports multi-source data fusion, and broadens the applicable scenarios of the method.

[0048] Sixth, for those who need to assess multi-year coupling changes over different time periods, intertemporal difference and migration assessment (ΔHRI, Δz*) can be provided as an optional extension to describe the magnitude and direction of the degradation process. Attached Figure Description

[0049] Figure 1 This is a flowchart of the method for identifying hydrological characteristics under the influence of permafrost degradation according to the present invention. Detailed Implementation

[0050] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0051] The method of this invention can run in the background processing environment of a hydrological monitoring system or geographic information platform, and is applicable to high-performance computing nodes to achieve automatic identification of large-scale, long-term permafrost hydrological coupling information. The identification method includes the following steps:

[0052] S1. Acquire time-series data of stratified soil phase state and hydrological flux to characterize the soil freezing phase state, and perform quality control and alignment processing on the acquired data.

[0053] S2, within a pre-defined multi-year time lag set, calculates the statistical relationship between stratified soil facies data and hydrological flux data by depth, and performs significance tests considering time autocorrelation and multiple comparison corrections on the obtained statistical relationships to screen out significant depth-time lag pairs to form a significant candidate set. ;

[0054] S3, for salient candidate sets At each depth, using soil phase data for that depth in each year as the independent variable and hydrological flux as the dependent variable, a piecewise linear model with a single inflection point is fitted to simultaneously estimate the threshold corresponding to that depth. Postthreshold sensitivity ;

[0055] S4, the time delay, threshold and post-threshold sensitivity at each depth are transformed to the [0,1] interval by a normalization mapping that satisfies monotonicity, and the mapped time delay, threshold and post-threshold sensitivity are combined with equal weights to obtain the hydrological response index;

[0056] S5: Within the depth set where both the threshold and post-threshold sensitivity pass the significance test, select the depth with the largest absolute value of post-threshold sensitivity as the master control layer, and output the hydrological response index and the spatial distribution of the master control layer.

[0057] This invention follows a main workflow of "data acquisition and preprocessing—multi-year time lag—construction of in-depth statistical relationships—threshold and post-threshold sensitivity estimation—construction of hydrological response index—determination of the main control layer and spatial representation," and provides intertemporal extensions and parameter ranges. For ease of explanation, the Northern Hemisphere climate seasons are used: spring (March-May), summer (June-August), autumn (September-November), and winter (December-February of the following year, with February belonging to the following year). The following formulas are written under fixed seasonal conditions, and seasonal symbols are not repeated.

[0058] (1) Data acquisition and preprocessing

[0059] Time series data of stratified soil phase parameters and hydrological fluxes are obtained. Soil phase parameters can be soil ice content or equivalent phase indicators, including but not limited to: unfrozen water conversion based on geothermal temperature, freeze-thaw index based on active or passive microwave, and phase indicators based on dielectric constant or resistivity. Hydrological fluxes include at least runoff and evapotranspiration. Raw data can be obtained from distributed hydro-permafrost model outputs, remote sensing or reanalysis products, and ground station observations. Input data undergoes quality control and standardization: missing data interpolation (linear interpolation or Kalman smoothing is preferred for missing data, and neighborhood or assimilation strategies are used for long missing data), anomaly identification and removal, time scale alignment, and spatial resampling; detrending and standardization are performed when necessary to reduce the impact of non-stationarity on subsequent statistical estimations. Daily or monthly series are aggregated by year and split by season to obtain annual seasonal series. Indicates the first Year, Depth Soil phase quantity (or equivalent indicator quantity). Indicates the first Hydrological flux during this season of the year.

[0060] (2) Construction of multi-year time lag-depth statistical relationship

[0061] In multi-year time-delay set (Unit: Year, Optimal) Within ) calculate and The correlation coefficients were used to construct the depth-delay correlation matrix:

[0062]

[0063] The correlation coefficient can be either Pearson's correlation coefficient or Spearman's rank correlation coefficient; to suppress confounding effects, partial correlation can be used to control for one or more factors among temperature, precipitation, radiation, and vegetation index. Considering the autocorrelation of the annual series, this invention performs annual block permutation or annual cycle shift tests on the statistics to obtain... Values: Preferred block length is 3–5 years, and the number of replacements is 1000–5000; subsequently, multiple comparison corrections are performed using the Benjamini–Hochberg procedure to obtain the values. Value. Will satisfy (Preferred) , can target The depth-delay pairs (from sensitivity analysis) constitute a significant candidate set. This generates a corresponding saliency mask for subsequent masking and quality annotation. (The above...) The salience mask and its salience form a correlation matrix of "multi-year - depth - season".

[0064] (3) Threshold and post-threshold sensitivity estimation

[0065] by For independent variable, As the dependent variable, a piecewise linear model with a single inflection point is fitted to simultaneously estimate the threshold. Postthreshold sensitivity :

[0066] ;

[0067] Among the breakpoints Determined through multi-start point or grid search to minimize the sum of squared residuals; Defined as the increment of the post-threshold slope relative to the pre-threshold slope, it characterizes the intensity and direction of the response after exceeding the threshold; is the intercept of the baseline response before the threshold, and b is the slope of the baseline response before the threshold. This is the intercept correction term after the threshold stage. The uncertainty of the model parameters is preferably estimated using the bootstrap interval method; when the breakpoint is not significant or the model does not converge, the depth is marked as "undetectable", and the corresponding component is treated as zero in subsequent synthesis, while a significance mask is used to indicate this in the results.

[0068] (4) Construction of hydrological response index

[0069] To integrate the three elements of "memory lag, threshold, and post-threshold sensitivity" on a multi-year scale, these three elements are mapped to... Interval and equal-weighted multi-year hydrological response indices are synthesized. First, a monotonic mapping satisfying directionality is defined:

[0070]

[0071] in, For memory lag, Follow Increasing monotonically non-increasing, Follow Increasing monotonically non-increasing, Follow The sum is monotonically non-decreasing, and the range of all three is [missing information]. In a preferred embodiment, a linear mapping is used:

[0072]

[0073] and take .in Use a 30-year period or the upper limit of the data; The upper quantile (preferably P95) of the multi-year samples in this quarter is taken, and out-of-bounds values ​​are first pruned and then normalized; when the threshold or slope is undetectable, the corresponding component is counted as zero and masked. The larger the index, the lower the threshold, the shorter the time delay, the greater the post-threshold sensitivity, and the stronger the coupling at the multi-year scale. This index is used to comprehensively characterize the coupling strength at the multi-year scale, but it is not a necessary condition for the determination of the master control layer.

[0074] (5) Control layer determination and spatial representation

[0075] To reflect the dominant depth of the post-threshold response, the controlling layer is defined as the soil layer that is detectable by the threshold and has the largest absolute value of post-threshold sensitivity. The usable depth set is denoted as:

[0076] and All passed the significance test. ;

[0077] The main control layer is:

[0078] .

[0079] When there are tandem maxima, the one with the higher goodness of fit is preferred (e.g., ...). If the samples are larger and still appearing side-by-side, prioritize the shallower ones, or those that consistently appear more frequently over a longer period. For individual depths that could not be detected... or In this case, the depth is not included in the main control layer's decision. Output the main control layer for each season. Spatial distribution and its significance mask and uncertainty interval; simultaneously output multi-year hydrological response index. Spatial distribution and partition statistics are used for a comprehensive description of coupling strength.

[0080] (6) Intertemporal difference and migration assessment (optional, by season)

[0081] When it is necessary to assess multi-year coupling changes over different periods, the time axis is divided into early and late periods (e.g., 1964–1999 and 2000–2024) based on the change point detection of driving factors (geothermal anomalies and / or phase anomalies), and calculations are performed separately. , And define the difference and the migration:

[0082] ;

[0083] .

[0084] The dividing year is determined by the driving factor sequence rather than... This is determined inherently to avoid circular reasoning; for The robustness test was performed on the annual boundary shift. This inter-period step is an optional extension and does not constitute a necessary limitation on the core method of this invention.

[0085] In a preferred embodiment, the upper limit of the multi-year time delay is... Use a 30-year timeframe (1-year step); FDR threshold (0.10 sensitivity analysis can be performed); the block length for annual block replacement or annual cyclic shift is 3–5 years, and the number of replacements is 1000–5000; The upper bound is set at P95 of the multi-year sample data for this season. The identification method of this invention is applicable to different spatial resolutions; preferably, data from 1-10 km, such as 5 km resolution data, can be used. When the sample period is short, a degradation strategy is adopted: shortening the block length to ensure that the number of blocks after permutation is not less than 8, or using effective sample size correction and robust standard errors; for areas with strong seasonality, cyclic shifting with the same season constraint can be used to maintain seasonal structural consistency. The identification method of this invention does not depend on a specific platform and can be implemented in a general scientific computing environment; parameters can be adjusted without departing from the essence of this invention.

[0086] See Figure 1 This embodiment uses a typical watershed in the Qinghai-Tibet Plateau (geographically ranging from approximately 95°E to 102°E and 32°N to 37°N) as the application scenario to demonstrate the complete process of this method.

[0087] S101: Data Acquisition and Preprocessing

[0088] Obtain time-series data for this region from 1964 to 2024, a total of 61 years.

[0089] (1) Soil phase data: Daily soil ice content data provided by the cryosphere hydrological model were used, with a data depth ranging from the surface to 2 meters, including multiple layers (e.g., 0.0m, 0.2m, 0.4m…2.0m). The daily data were aggregated into an average series for spring, summer, autumn and winter to obtain stratified time series data. .

[0090] (2) Hydrological flux data: Daily runoff observation data released by national hydrological stations in the basin were used to calculate the total runoff for the four seasons and obtain the time series data of hydrological flux. .

[0091] (3) Preprocessing: The above sequence is standardized to eliminate dimensions, and linear interpolation is used for missing values.

[0092] S102: Construction of Multi-Year Lag-Deep Statistical Relationship

[0093] (1) Preset time delay set: (Year), that is, considering the time lag effect of up to 30 years.

[0094] (2) Statistical method: Calculate the value of each depth and each time delay Down, and Spearman rank correlation coefficient .

[0095] (3) Significance test: The p-value was calculated using the annual block permutation test (block length: 4 years, number of permutations: 2000), and multiple comparison correction was performed using the FDR (false detection rate) procedure. The significance level was set at [value missing]. Will satisfy Depth-delay pair Store in significant candidate set And generate the corresponding binary saliency mask.

[0096] S103: Threshold and Postthreshold Sensitivity Estimation

[0097] For significant candidate sets Each depth in :

[0098] (1) Piecewise regression: As the independent variable, Using [variable name] as the dependent variable, perform piecewise linear regression with a single inflection point. Determine the threshold that minimizes the sum of squared residuals through grid search. And obtain the postthreshold sensitivity. .

[0099] (2) Significance assessment: The bootstrap method (1000 iterations) was used to calculate the significance. and The 95% confidence interval is used. If the confidence interval contains zero or the model has not converged, the depth is marked as "undetectable".

[0100] S104: Construction of Hydrological Response Index

[0101] (1) Normalized mapping:

[0102] ;

[0103] ,in For all significant depth thresholds this season The 95th percentile;

[0104] ,in For all significant depths this season The 95th percentile;

[0105] For depths that are "undetectable", and Take 0.

[0106] (2) Index synthesis: Calculate the hydrological response index at each depth. .

[0107] S105: Control Layer Determination and Spatial Representation

[0108] (1) Define the available depth set: and All passed the significance test. .

[0109] (2) Select the main control layer: When there are multiple maxima, the goodness-of-fit factor should be chosen first. If the two are ranked higher, and still tied, the one with the lower elevation will be given priority.

[0110] S106: Product Output: Generate a spatial distribution map of the spring "Hydrological Response Index (HRI)" for the entire watershed (the maximum value can be used). (representing the pixel) and "master layer depth" "Spatial distribution map, with corresponding salience masking."

[0111] S107, Intertemporal Difference and Migration Assessment (Optional Extension)

[0112] 1. Period Division: Based on the regional annual average ground temperature sequence, the Pettitt change point detection method was used to detect the change point year as 1998. Based on this, the time axis was divided into the early period (1980-1997) and the late period (1999-2020).

[0113] 2. Calculate the difference: For each pixel within the watershed, calculate the difference between the early and late spring periods. and and calculate .

[0114] 3. Calculate migration: For each pixel, obtain the master control layer depth before and after the migration. and and calculate .

[0115] 4. Robustness test: Shift the change point years to 1996 and 2000, and repeat the above calculation. If... and If the spatial pattern is basically consistent with the results of 1998 as a turning point (e.g., Spearman correlation coefficient > 0.85), then the results are considered robust.

[0116] The identification method of this invention ultimately outputs, but is not limited to, the following: an HRI index raster in GeoTIFF format, a master control layer depth raster, a saliency mask raster, and a summary table in CSV format statistically analyzed by watershed sub-region. All calculations can be implemented using Python programs on servers or ordinary computers, and are suitable for large-scale batch processing on high-performance computing clusters. This invention overcomes the shortcomings of existing technologies that only reflect thermal state and are difficult to quantify response intensity and identify key depths, achieving automated and standardized diagnosis of permafrost-hydrological coupling structures, and can be used for permafrost degradation monitoring and cold region water resource management.

[0117] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0118] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for identifying hydrologic characteristics under the influence of permafrost degradation, comprising: The identification method is used for realizing automatic and standardized diagnosis of frozen soil-hydrological coupling structure, and comprises the following steps: S1, obtaining layered soil phase state time series data and hydrological flux time series data representing soil freezing phase state, and performing quality control and alignment processing on the obtained data; S2, in a set of pre-set multi-year time lags, calculate the statistical relationship between the layered soil phase state time series data and the hydrological flux time series data according to the depth, and perform a significance test and multiple comparison correction considering the time autocorrelation on the obtained statistical relationship, and screen out significant depth-time lag pairs to form a significant candidate set ; S3, for each depth in the set of significant candidates , fit a single-kink piecewise linear model to estimate the threshold value corresponding to the depth and the post-threshold sensitivity ;​ S4, converting the time lag, threshold value and post-threshold sensitivity of each depth to the interval [0, 1] through normalized mapping meeting monotonicity respectively, and performing equal weight synthesis on the mapped time lag, threshold value and post-threshold sensitivity to obtain a hydrological response index; S5, in the depth set in which both the threshold value and the post-threshold sensitivity pass the significance test, selecting the depth with the largest absolute value of the post-threshold sensitivity as the main control layer, and outputting the spatial distribution of the hydrological response index and the main control layer.

2. The method for identifying the hydrological characteristics under the permafrost degradation effect according to claim 1, characterized in that, In step S1, the layered soil phase state time series data includes soil ice content or an equivalent phase state indicator of soil ice content; the equivalent phase state indicator includes unfrozen water conversion based on ground temperature, freezing and thawing index based on active or passive microwave, and phase state index based on dielectric constant or resistivity; and the hydrological flux time series data at least includes runoff and evapotranspiration.

3. The method of identifying the hydrologic signature under permafrost degradation influence according to claim 1, wherein, In step S1, the process of performing quality control and alignment processing on the obtained data comprises the following steps: The layered soil phase state and hydrological flux related original data are extracted from a distributed hydrological frozen soil model, remote sensing or reanalysis products and ground station observation data; Linear interpolation or Kalman smoothing is used for short missing data interpolation, and neighborhood or assimilation strategy is used for long missing data interpolation; the interpolated original data are sequentially subjected to abnormality identification and elimination, time scale alignment, spatial resampling, detrending and standardization processing, and corresponding daily or monthly layered soil phase state sequence and hydrological flux sequence are generated; The daily or monthly layered soil phase state sequence and hydrological flux sequence are aggregated by year and then split by season to obtain seasonal sequence of each year.

4. The method of identifying the hydrologic signature under permafrost degradation influence according to claim 1, wherein, Step S2 further comprises: In a preset multi-year time lag set, the statistical relationship between the layered soil phase state time series data and the hydrological flux time series data is calculated according to depth, and the obtained statistical relationship is subjected to significance test considering time autocorrelation and multiple comparison correction to screen out a set of significant depth-time lag pairs; In order to overcome the time autocorrelation effect existing in multi-year sequence, annual block permutation test or annual circular shift test is performed on each depth-time lag combination in the depth-time lag pair set to obtain corresponding significance results; The multiple comparison correction is performed on the significance results obtained under multiple depths and multiple time lags to control the overall false rejection rate; The depth-time lag combinations passing through the corrected threshold value are screened out to form a significant candidate set, and a depth-time lag significance mask reflecting multi-year coupling structure is generated according to the significant candidate set.

5. The method of identifying the hydrologic signature under permafrost degradation influence according to claim 1, wherein, Step S3 further comprises: For each depth in the significant candidate set, a single inflection point segmented linear model containing a pre-threshold segment and a post-threshold segment is constructed, the inflection point position is searched in the variable change interval to minimize the overall fitting error of the model, so as to determine the threshold value corresponding to the depth; After determining the threshold, linear fitting is performed on the pre-threshold segment and the post-threshold segment respectively to obtain linear response coefficients of the two segments, and the difference between the two segment coefficients is used to represent the degree of response enhancement after the threshold, thereby obtaining the post-threshold sensitivity; The confidence interval of the threshold corresponding to the depth and the post-threshold sensitivity is evaluated, and when the threshold position is unstable or the sensitivity is not significant, the depth is marked as undetectable and is distinguished in a mask manner in subsequent processing.

6. The method of identifying the hydrologic signature under permafrost degradation influence according to claim 1, wherein, Step S4 further comprises: The preferred time lag, threshold and post-threshold sensitivity corresponding to each depth are respectively input into a preset monotonic normalization function, so as to be converted to the same normalization interval, and the upper limit parameter of the monotonic normalization function is determined according to the statistical upper limit of the multi-year sample; The normalized components of the mapped preferred time lag, threshold and post-threshold sensitivity are combined according to the equal weight rule to form a hydrological response index for representing the multi-year scale permafrost-hydrological coupling strength. The normalized component of the preferred time lag is used to reflect the coupling delay characteristic, the normalized component of the threshold is used to reflect the phase state degree required for triggering, and the normalized component of the post-threshold sensitivity is used to reflect the response strength after the threshold.

7. The method for identifying the hydrological characteristics under the permafrost degradation effect according to claim 6, characterized in that, The preferred time lag is the weighted average, median or mode of the significant time lag set, and the upper limit of the normalization function is determined according to the upper quantile of the corresponding seasonal multi-year sample.

8. The method for identifying the hydrologic signature under the influence of frozen soil degradation according to claim 1, wherein, Step S5 further comprises: The depths with significant threshold and post-threshold sensitivity are included in the available depth set; the main control layer depth is determined by comparing the absolute values of the post-threshold sensitivities of the depths; when multiple depths simultaneously satisfy the maximum condition, the depth with higher fitting goodness is preferentially selected, and then the depth with shallower depth or higher frequency in the multi-year sequence is selected; the spatial distribution of the main control layer depth, the significance mask and the uncertainty range are output.

9. The method of identifying the hydrologic signature under permafrost degradation influence according to claim 1, wherein, The method further comprises: Based on the anomaly of ground temperature, phase state or other driving factors, the multi-year sequence is divided into the early period and the late period; the hydrological response index and the main control layer depth of the early period and the late period are calculated respectively; the change amount of the coupling strength and the migration amount of the main control layer depth are obtained by comparing the differences between the two periods, which are used to reflect the change of the coupling structure in the permafrost degradation process.

Images