An evaluation method and system for amplification effect of human activities in black soil area on groundwater drought

Abstract

The present application belongs to the technical field of hydrogeology and remote sensing application, and particularly relates to a method and system for evaluating the amplification effect of human activities on groundwater drought in black soil areas. The method comprises the following steps: obtaining groundwater storage anomaly, meteorological and hydrological and human activity multi-source time series data; performing trend analysis and ensemble empirical mode decomposition on the groundwater storage anomaly data; calculating the standardized groundwater drought index SGDI and identifying regional drought events; calculating the coupling drought amplification index CDAI; performing multi-factor contribution separation by using partial correlation analysis and random forest regression model; and dividing the farmland expansion dominant area, the climate change dominant area or the coupling dominant area based on the contribution rate threshold. The present application realizes the quantitative evaluation and driving contribution separation of the groundwater drought amplification effect under the coupling effect of farmland expansion and climate change, and provides scientific support for the sustainable utilization of groundwater resources, precise drought early warning and food security guarantee in black soil areas.

Description

Technical Field

[0001] This invention belongs to the field of hydrogeology and remote sensing application technology, and in particular relates to a method and system for assessing the amplification effect of human activities on groundwater drought in black soil areas. Background Technology

[0002] Groundwater is a crucial water source for maintaining stable crop yields in my country's core grain-producing areas (especially the black soil region of Northeast China). In recent years, climate warming has exacerbated the uneven spatial and temporal distribution of precipitation, significantly weakening the natural recharge capacity of groundwater; while excessive water extraction for human production and daily life continues to widen the gap between water supply and demand, both of which jointly drive the formation and exacerbation of groundwater drought.

[0003] Existing groundwater drought assessment technologies suffer from the following main shortcomings: First, they often employ the Standardized Groundwater Drought Index (SGDI) based on groundwater level or storage anomaly data combined with meteorological factors for statistical correlation analysis. However, the core analytical logic treats climate fluctuations and driving factors such as agricultural production as independent variables, only able to deduce the impact of a single factor or perform simple linear superposition, failing to characterize the dynamic interaction between multiple factors. Second, they lack the quantitative analytical capability for the nonlinear process of "coupling of farmland expansion and climate change – drought response," and cannot identify the amplification mechanism of farmland expansion and irrigation water withdrawal on the water consumption effect of climate drought during spring droughts. Third, they fail to quantitatively separate the contribution rates of farmland expansion and climate change to groundwater drought in the coupled system, resulting in a lack of regional targeting in water resource management policies.

[0004] Therefore, there is an urgent need for an assessment method that can scientifically quantify the amplification effect of groundwater drought in agricultural areas under the coupling effect of human activities and climate change, so as to achieve clear zoning of driving contributions and dynamic analysis at multiple scales, and provide a scientific basis for the sustainable utilization of groundwater resources in black soil areas and the safeguarding of national food security. Summary of the Invention

[0005] This invention provides a method and system for assessing the drought amplification effect of human activities on groundwater in black soil regions. By constructing a coupled drought amplification index (CDAI) and a multi-factor contribution separation model, it solves the technical problems of existing technologies that do not consider the coupling effect between the two, cannot quantify the amplification effect, and have ambiguous driving contributions.

[0006] Therefore, the first technical solution of this application discloses a method for assessing the amplification effect of human activities on groundwater drought in black soil regions, including the following steps:

[0007] Acquire multi-source input data, which includes: groundwater storage anomaly time series data, meteorological and hydrological time series data, and human activity time series data;

[0008] Trend analysis and signal decomposition are performed on the groundwater storage anomaly time series data, and trend component and periodic component are output; wherein, the trend analysis adopts a combination algorithm of nonparametric slope estimation and significance test, and the signal decomposition adopts empirical mode decomposition algorithm to decompose the original time series into several intrinsic mode functions and residual terms;

[0009] The standardized groundwater drought index (SGDI) time series is calculated based on the acquired groundwater storage anomaly time series data, and drought events are identified according to the judgment threshold.

[0010] The Coupled Drought Amplification Index (CDAI) is calculated based on the SGDI time-series data, using the following formula:

[0011] ;

[0012] in, This is the measured groundwater drought index calculated based on all input data. The simulated groundwater drought index is calculated based on meteorological and hydrological time series data after removing the aforementioned human activity time series data. The CDAI > 0 indicates the existence of a drought amplification effect.

[0013] Based on a machine learning regression model, the SGDI time series data, human activity time series data, and meteorological and hydrological time series data are subjected to multi-factor contribution separation. The contribution rate of each driving factor is output, and the dominant type partitioning is performed based on the contribution rate to output the regional dominant type classification results.

[0014] Preferably, the groundwater storage anomaly time series data is satellite gravity measurement remote sensing inversion data;

[0015] The meteorological and hydrological time-series data include air temperature, precipitation, and radiation flux;

[0016] The human activity time-series data includes farmland expansion time-series data, irrigation water use time-series data, and socio-economic data; and

[0017] The farmland expansion time series data is cultivated land classification data from optical remote sensing images;

[0018] The irrigation water time series data is spatialized raster data of statistical data.

[0019] Preferred,

[0020] The nonparametric slope estimation uses the Theil-Sen slope estimation to calculate the median slope, and the significance test uses the Mann-Kendall test. The trend was determined to be significant at that time;

[0021] The set-based empirical mode decomposition algorithm suppresses mode aliasing by adding white noise to the original sequence.

[0022] Preferably, the step of calculating SGDI and identifying drought events based on groundwater storage anomaly time-series data includes:

[0023] Using a preset time period as the baseline period, the mean μ and standard deviation σ of the groundwater storage anomaly values ​​within the baseline period are calculated. Based on the mean and standard deviation, the time-point data are standardized to obtain the SGDI time series.

[0024] When the SGDI value is ≤-1.0 at three or more consecutive time points, it is determined to be a regional groundwater drought event, and the start and end times and duration of the drought event are output.

[0025] Preferably, the multi-factor contribution separation and dominant type partitioning based on the machine learning regression model includes:

[0026] A random forest regression model was constructed using the annual rate of change of cultivated land area, annual precipitation, annual mean temperature and annual evapotranspiration as input variables and SGDI value as output variable.

[0027] The importance measure of each input variable is calculated as the contribution rate based on the random forest regression model.

[0028] When the contribution rate of farmland expansion factor is >60%, the output is farmland expansion-dominant type;

[0029] When the contribution rate of climate factors is greater than 60%, the output is climate change-dominated.

[0030] When the difference between the contribution rates of farmland expansion factor and climate factor is ≤20%, the output coupling is dominant.

[0031] Preferably, the method further includes: calculating the geometric centroid coordinates of the drought region at different time scales based on the spatial distribution data of the drought event, and outputting the temporal migration trajectory data of the geometric centroid coordinates.

[0032] The second technical solution of this application discloses an assessment system for the amplification effect of human activities on groundwater drought in black soil regions, including:

[0033] The data input module is configured to acquire multi-source input data, including groundwater storage anomaly time series data, meteorological and hydrological time series data, and human activity time series data.

[0034] The spatiotemporal analysis module is configured to perform trend analysis and signal decomposition on the groundwater storage anomaly time series data, and output trend component and periodic component;

[0035] The drought identification module is configured to calculate the SGDI index based on the groundwater storage anomaly time series data and identify drought events according to the judgment threshold.

[0036] The coupling amplification effect calculation module is configured to calculate the coupling drought amplification index (CDAI) based on the SGDI index, wherein the CDAI calculation formula is:

[0037] ;

[0038] in, This is the measured groundwater drought index calculated based on all input data. The simulated groundwater drought index is calculated based on meteorological and hydrological time-series data after removing the aforementioned human activity time-series data.

[0039] The driving contribution separation module is configured to perform multi-factor contribution separation on the SGDI index and the time series data of human activities and meteorological and hydrological data based on a machine learning regression model, and output the contribution rate of each driving factor.

[0040] The dominant type partitioning module is configured to perform dominant type partitioning based on the contribution rate and a preset threshold, and output the dominant type classification results for the region.

[0041] Furthermore, the system also includes:

[0042] The validation module is configured to perform leave-one-out cross-validation and output model accuracy metrics.

[0043] The visualization output module is configured to generate and output SGDI spatial distribution map, CDAI spatial pattern map, and drought centroid migration trajectory map.

[0044] And, a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements an assessment method for the amplification effect of human activities on groundwater drought in the black soil region.

[0045] A computer-readable storage medium storing a computer program that, when executed by a processor, implements a method for assessing the amplification effect of human activities on groundwater drought in the black soil region.

[0046] Beneficial effects

[0047] (1) Precise quantification of amplification effect: Through the coupled driving assessment method, the degree of groundwater drought amplification under the superposition of farmland expansion and irrigation can be quantitatively characterized, avoiding the neglect of multi-factor interaction effect by the traditional single factor assessment method, and the assessment logic is more in line with the actual drought driving characteristics.

[0048] (2) Clear driving contribution: The multi-factor coupled contribution rate decomposition model is adopted to achieve quantitative separation of the contribution rates of climate and human activities to groundwater drought for the first time. It can clearly distinguish the dominant driving types in different regions and provide a more targeted basis for water resource management.

[0049] (3) Multi-scale support for decision-making: covering time trends, spatial distribution, and centroid migration analysis, providing targeted solutions for groundwater management and drought prevention in different regions and at different time scales.

[0050] (4) Accurate identification of drought events: The criteria of SGDI≤-1.0 and continuous duration≥3 months are adopted to effectively exclude short-term random fluctuations, ensure that the identified drought events have actual hydrological significance and have a good spatiotemporal matching degree with historical agricultural production reduction records.

[0051] (5) Provides a basis for targeted regional management: By dividing the thresholds of 60% and 20%, the dominant driving types of different regions are clarified, so that water resource management policies are transformed from "one-size-fits-all" to "regional policies", significantly improving management efficiency. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 The spatial variation rate (K value) of groundwater drought index (SGDI) in the black soil region from April 2002 to December 2022.

[0054] Figure 2 This represents the long-term variation of the Groundwater Drought Index (SGDI) in the Black Soil Region from April 2002 to December 2022. Detailed Implementation

[0055] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0056] The first embodiment of this application discloses a method for assessing the amplification effect of human activities on groundwater drought in black soil regions, comprising the following steps:

[0057] 1. Acquire multi-source input data, including: groundwater storage anomaly time series data, meteorological and hydrological time series data, and human activity time series data;

[0058] The time-series data on groundwater storage anomalies are derived from satellite gravity measurement remote sensing inversion data, specifically groundwater storage anomaly (GWSA) data retrieved by the GRACE (Gravity Recovery and Climate Experiment) and its subsequent mission, GRACE-FO. This data, based on satellite gravity measurement technology, inverts changes in terrestrial water storage by detecting variations in the Earth's gravity field, and then subtracts components such as surface water, soil water, and snow cover to obtain the groundwater storage anomaly. The data has a monthly temporal resolution, and the spatial resolution reaches 0.05° after downscaling. Compared to traditional groundwater level observation well networks, satellite gravity measurement offers advantages such as complete spatial coverage and lack of geographical limitations, making it particularly suitable for monitoring groundwater storage in large-scale areas (such as the entire Northeast Black Soil Region).

[0059] The meteorological and hydrological time-series data includes near-surface air temperature, precipitation rate, downward shortwave radiation flux, and downward longwave radiation flux. These data can be derived from regional surface meteorological element-driven datasets in China (such as datasets merging ERA5 reanalysis data with station observations), with a temporal resolution of 3 hours or daily scales, and are aggregated to obtain monthly-scale data. Through multi-source data fusion and artificial intelligence interpolation techniques, the accuracy of key variables such as radiation and precipitation is significantly improved, providing reliable input for subsequent evapotranspiration calculations and drought index construction.

[0060] The human activity time-series data includes farmland expansion time-series data, irrigation water use time-series data, and socioeconomic data. Farmland expansion data is based on optical remote sensing imagery such as Landsat, with farmland information extracted through supervised classification or deep learning algorithms, achieving a spatial resolution of 30 meters and a temporal resolution of annually. Irrigation water use data is sourced from local statistical yearbooks (such as the *Heilongjiang Provincial Statistical Yearbook*) and publicly available agricultural water use bulletins from the Ministry of Agriculture and Rural Affairs. After spatial processing (such as raster interpolation based on farmland area distribution), it forms raster data with a spatial resolution matching that of the groundwater data. This application spatially matches socio-statistical data with remote sensing data, achieving a quantitative expression of human activity intensity in geographic space, laying the foundation for subsequent coupled analysis.

[0061] It should be noted that the core of the groundwater drought amplification effect assessment in this invention relies on the aforementioned groundwater data, meteorological data, and human activity data. Compared with existing technologies, no additional auxiliary data (such as topographic, hydrogeological, and land use coordination data) are involved in model construction and result analysis. Existing multi-source data already covers the time, space, and driving factor dimensions required for the assessment, and can fully support the entire technical process of groundwater drought identification, amplification effect quantification, and driving contribution separation in the Northeast Black Soil Region.

[0062] 2. Perform trend analysis and signal decomposition on the groundwater storage anomaly time series data, and output trend component and periodic component; wherein, the trend analysis adopts a combination algorithm of nonparametric slope estimation and significance test, and the signal decomposition adopts empirical mode decomposition algorithm to decompose the original time series into several intrinsic mode functions and residual terms, so as to distinguish the influence of meteorological factors and human activities on groundwater storage, and provide a data foundation for subsequent contribution analysis and regional dominant type classification.

[0063] Trend analysis: A combined algorithm of Theil-Sen slope estimation and Mann-Kendall test was used. Theil-Sen slope estimation is a nonparametric statistical method that calculates all pairwise data pairs... slope Take all The median is used as the trend slope of the sequence. This indicates that reserves are on the rise. <0 indicates a downward trend. Compared to traditional linear regression, this method is insensitive to outliers. Even when there are extreme outliers in the groundwater storage data (such as abnormal fluctuations in strong El Niño years), it can still robustly estimate the long-term trend, avoiding the trend estimation bias caused by outliers in the least squares method.

[0064] The Mann-Kendall test is used to determine the significance of a trend: first, construct an ordered series S, then calculate the statistic based on the variance Var(S) of S. ;when (when α=0.05) =1.96), indicating that the trend is significant at the 95% confidence level. This combined method can clearly define the trend direction, rate of change, and significance of reserves. The irrigation method does not require the assumption that the data follows a normal distribution, making it suitable for the non-normal distribution characteristics often observed in hydrological time series.

[0065] Signal decomposition: The Ensemble Empirical Mode Decomposition (EEMD) algorithm is used to separate the trend and periodic terms of the groundwater storage sequence, distinguishing the dominant roles of human activities and climate fluctuations. EMD involves repeatedly adding white noise of different amplitudes to the original sequence x(t) (to suppress mode aliasing in traditional EMD), and then repeatedly performing EMD decomposition on the noisy sequence to obtain several intrinsic mode functions (IMFs) and residual terms r(t). Based on component characteristics: high-frequency IMFs are classified as periodic terms (corresponding to dominant climate fluctuations, due to seasonal or interannual periodic variations in climate elements such as precipitation and evapotranspiration); the residual term r(t) is classified as a trend term (corresponding to dominant human activities, due to the long-term cumulative trend of farmland expansion and irrigation water withdrawal), achieving quantitative separation of the components corresponding to the two types of driving factors.

[0066] 3. Spatial Distribution Analysis: This study characterizes the spatial pattern of groundwater storage changes in different geomorphic regions and matches it with the characteristics of high-intensity irrigation areas. First, based on annual groundwater storage raster data, Kriging interpolation is used to construct a continuous spatial distribution field of storage changes. For geomorphic regions such as the Songnen Plain, Sanjiang Plain, and Liaohe Plain, indicators such as the mean and coefficient of variation of storage changes within these regions are extracted to compare the intensity of changes and spatial heterogeneity in different areas. Then, high-intensity irrigation area data (obtained by coupling CLCD farmland data and irrigation data) is overlaid. Through spatial overlay statistics, the matching relationship between irrigation areas and areas with high or low storage changes is clarified, and the degree of influence of irrigation activities on the spatial pattern of storage in different geomorphic regions is analyzed.

[0067] 3. Calculate the standardized groundwater drought index (SGDI) time series based on the acquired groundwater storage anomaly time series data, and identify drought events according to the judgment threshold;

[0068] (1) Definition of the base period and calculation of parameters: For example, taking April 2002 to December 2022 as the complete base period, calculate the base period mean (μ) and base period standard deviation (σ) of the GWSA anomaly data during this period, as follows:

[0069] ;

[0070] In the formula, σ represents the GWSA anomaly at time t within the base period, in cm. n is the total number of time nodes in the base period (249 monthly nodes from April 2002 to December 2022), in cm. σ is the standard deviation for the base period, in cm.

[0071] SGDI standardization calculation: Based on the baseline parameters, the GWSA anomalies at each time node are standardized to obtain the SGDI sequence, as shown in the following formula:

[0072] ;

[0073] in, Let t be the groundwater drought index at the t-th time point. The smaller the value, the more severe the groundwater drought.

[0074] (2) Verification and determination of drought event judgment criteria

[0075] Based on the characteristics of groundwater drought in the target area, a target threshold is set by combining the SGI threshold with the time of historical drought events. When the SGDI of the target area is less than or equal to this threshold and the duration of this state is greater than or equal to a preset time, it is judged as a regional groundwater drought event. Taking the Northeast Black Soil Region as an example, based on the historical groundwater drought characteristics of the Northeast Black Soil Region, and using the SGI threshold with the time of historical drought events, the final setting is: when the SGDI of a certain area is less than or equal to -1.0, and the continuous duration of this state is greater than or equal to 3 months, it is judged as a regional groundwater drought event.

[0076] 1) Threshold rationality: An SGDI of less than or equal to -1.0 corresponds to a GWSA anomaly value that is one standard deviation below the mean during the baseline period, which is consistent with the degree of GWSA anomaly during historical drought periods;

[0077] 2) Duration requirement: A duration of 3 months or more can exclude short-term random fluctuations (such as temporary water shortages caused by less rainfall in a single month) and ensure that the identified drought events have a real impact.

[0078] 4. Calculate the Coupled Drought Amplification Index (CDAI) based on the SGDI time-series data. The calculation formula is as follows:

[0079] ;

[0080] in, The measured groundwater drought index is calculated based on all input data (climate + human activities), i.e., the SGDI obtained by direct observation calculation in step 3-(1); The simulated groundwater drought index is calculated based on meteorological and hydrological time series data after removing the aforementioned human activity time series data. It is the "natural state" drought index assuming no human activity (farmland expansion, irrigation water extraction) interference.

[0081] The calculation method is as follows: set the farmland expansion data and irrigation water data in the input data to zero or replace them with historical benchmark values ​​(such as the benchmark level in 2002), keep the meteorological and hydrological data unchanged, and re-execute the SGDI calculation process of step 3-(1) to obtain the result.

[0082] A CDAI value greater than 0 indicates a drought amplification effect, meaning that human activities (farmland expansion and irrigation) exacerbate the depletion of groundwater by climate drought; the higher the CDAI value, the stronger the amplification effect. This step overcomes the limitations of traditional methods that only calculate the impact of single factors separately, enabling a comparison between the coupled scenario of "climate drought + human activities" and the single scenario of "climate drought only," accurately quantifying the degree of drought amplification by human activities. For example, if a certain region... =-1.2 (moderately arid climate), while =-1.8 (severe actual drought), then CDAI=0.5, indicating that human activities have amplified the drought intensity by 50%.

[0083] 5. Based on a machine learning regression model, perform multi-factor contribution separation on the SGDI time series data, the human activity time series data, and the meteorological and hydrological time series data, output the contribution rate of each driving factor, and perform dominant type partitioning based on the contribution rate to output the regional dominant type classification results.

[0084] Partial correlation analysis preprocessing: Before constructing the regression model, partial correlation analysis is performed on the input variables such as the annual change rate of cultivated land area, annual precipitation, annual average temperature, and annual evapotranspiration to eliminate multicollinearity among variables. The purpose of this processing is that variables such as farmland expansion and irrigation water use, and temperature and evapotranspiration, often exhibit statistical correlation (collinearity), and direct regression can lead to unstable coefficient estimates. Partial correlation analysis, by controlling for the influence of other variables, calculates the net correlation between the two variables, ensuring the independence and accuracy of subsequent contribution rate calculations.

[0085] Random forest regression modeling: Using the annual rate of change of cultivated land area, annual precipitation, annual mean temperature, and annual evapotranspiration as input variables (features), and the SGDI value as the output variable (objective), a random forest regression model is constructed. Random forests generate multiple decision trees through an ensemble learning mechanism and average the prediction results. Random forests can capture nonlinear interactions between factors (such as the synergistic effect of rising temperatures and irrigation water intake), and can quantitatively calculate the contribution rate of each factor to the variation of SGDI through feature importance measures (such as based on the reduction of Gini impurity or permutation importance).

[0086] The importance measure of each input variable is calculated as the contribution rate based on the random forest regression model.

[0087] When the contribution rate of farmland expansion factor is >60%, the output is farmland expansion-dominant type;

[0088] When the contribution rate of climate factors is greater than 60%, the output is climate change-dominated.

[0089] When the difference between the contribution rates of farmland expansion factor and climate factor is ≤20%, the output coupling is dominant.

[0090] In this implementation, the 60% and 20% thresholds are set based on management decision-making needs. The 60% threshold ensures the absolute control of the dominant factor, facilitating the formulation of targeted policies (e.g., strict control of arable land development in areas dominated by farmland expansion, and enhanced climate adaptability in areas dominated by climate change). The 20% coupling zone threshold identifies areas where the effects of both factors are roughly equal, indicating the need for comprehensive control measures. The effect of this zoning technology is that it breaks through the traditional "one-size-fits-all" management model, achieving regional precision and type-differentiation in groundwater resource management.

[0091] In a further embodiment, the method further includes: calculating the geometric centroid coordinates of the drought region at different time scales based on the spatial distribution data of the drought event and the spatial distribution data of the drought event obtained by SGDI drought identification, and outputting the temporal migration trajectory data of the geometric centroid coordinates.

[0092] Multi-dimensional verification: ① Consistency verification: compare the spatiotemporal matching degree between drought events and historical records of agricultural yield reduction and groundwater over-extraction; ② Model accuracy verification: adopt leave-one-out cross-validation to ensure that CDAI calculation error is <15% and contribution rate separation accuracy is ≥80%.

[0093] Spatial centroid analysis: Tracking the migration trajectory of the spatial centroid of drought at different time scales (March, June, September, and December) further reveals the spatial propagation pattern of drought driven by the coupling of meteorology and human activities, which is a spatial supplement and verification of the aforementioned spatiotemporal change analysis results.

[0094] Application scenarios: The assessment results will be applied to: ① precise early warning of groundwater drought; ② differentiated regional water resource allocation; ③ farmland development and climate adaptation policy formulation.

[0095] The second embodiment of this application discloses an assessment system for the amplification effect of human activities on groundwater drought in black soil regions, comprising:

[0096] The data input module is configured to acquire multi-source input data, supporting batch import, projection transformation, resampling, and cropping of multi-source heterogeneous data (NetCDF, GeoTIFF, CSV, etc. formats). The multi-source input data includes groundwater storage anomaly time-series data, meteorological and hydrological time-series data, and human activity time-series data.

[0097] The spatiotemporal analysis module is configured to perform trend analysis and signal decomposition on the groundwater storage anomaly time series data. It has a built-in Theil-Sen slope estimation, Mann-Kendall test, and EEMD decomposition algorithm library, and outputs trend slope, significance level, periodic term, and trend term components.

[0098] The drought identification module is configured to calculate the SGDI index based on the groundwater storage anomaly time series data and identify drought events according to the judgment threshold; it can automatically calculate the SGDI time series, identify drought events based on the -1.0 threshold and the 3-month duration standard, and output the start and end time, duration and intensity level of the drought event.

[0099] The coupled amplification effect calculation module is configured to calculate the coupled drought amplification index CDAI based on the SGDI index and generate a spatial distribution map of the amplification effect.

[0100] The driving contribution separation module is configured to perform multi-factor contribution separation on the SGDI index and the time series data of human activities and meteorological and hydrological data based on a machine learning regression model, and output the contribution rate of each driving factor.

[0101] The dominant type partitioning module is configured to perform dominant type partitioning based on the contribution rate and a preset threshold, and output the dominant type classification results for the region.

[0102] In a further embodiment, the system may also include:

[0103] The validation module is configured to perform leave-one-out cross-validation, which evaluates model accuracy by successively removing one sample from the training model and then predicting that sample, ensuring that the CDAI calculation error is <15% and the contribution rate separation accuracy is ≥80%. The technical advantage lies in ensuring the reliability and generalization ability of the evaluation results through rigorous statistical validation.

[0104] The visualization output module is configured to generate and output SGDI spatial distribution maps, CDAI spatial pattern maps, dominant type zoning maps, and drought centroid migration trajectory maps, supporting dynamic timeline display and interactive queries.

[0105] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the evaluation method as described in Embodiment 1.

[0106] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the evaluation method as described in Embodiment 1.

[0107] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that it may also include input / output interfaces, network interfaces, etc.

[0108] The processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0109] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function, etc.; the data storage area may store data created based on the use of the computer device, etc. Furthermore, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart memory cards (SMCs), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0110] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the evaluation method described in this invention.

[0111] The technical means and effects of this application will be described in detail below through specific embodiments.

[0112] Example 1

[0113] Using the Northeast Black Soil Region from April 2002 to December 2022 as the research period, the above methods and systems were applied, combined with the spatial distribution map of the SGDI trend K value ( Figure 1 ), Overall SGDI time series plot ( Figure 2 The following core results were obtained.

[0114] The SGI (Solar Growth Index) in the Northeast Black Soil Region shows a significant downward trend with fluctuations: from 2002 to 2010, the SGDI mainly fluctuated around 0; after 2010, the fluctuation range widened and continued to shift downwards; and from 2018 to 2022, the SGDI repeatedly reached extreme low values ​​below -2 (corresponding to...). Figure 2 The period below the red dotted line). Spatially, the SGDI trend K value shows significant divergence ( Figure 1In the central Songnen Plain core area of ​​the study region, the K value is between 4 and 6 (red high value area), and the SGDI decline trend is the most significant; in the southern and eastern areas (parts of the Sanjiang Plain and Liaohe Plain), the K value is between -2 and 0 (blue low value area), and the SGDI decline is slower. Moreover, the high K value area has a spatial matching degree of 81% with the core area of ​​"18% expansion of cultivated land from 2002 to 2022" shown by CLCD data, which confirms the driving role of farmland expansion.

[0115] Between 2002 and 2022, based on SGDI below -1 (e.g.) Figure 2 (Indicated by the red dashed line) and lasting longer than 3 months, a total of 7 regional drought events were identified. The longest duration was between 2018 and 2019, reaching 14 months; the second longest was between 2021 and 2022, reaching 12 months. The absolute value of the mean SGDI during the drought periods ranged from 1.5 to 2.8. Regions experiencing severe drought were... Figure 1 The high-value regions of the K-value completely overlap.

[0116] The above applications verify that this method can accurately quantify the amplification effect and clearly distinguish the driving type, providing a scientific basis for formulating differentiated groundwater management policies in the black soil region (such as strictly controlling the increase of irrigated area in the Songnen Plain and implementing climate-adaptive irrigation system in the Sanjiang Plain).

[0117] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for assessing the amplification effect of human activities on groundwater drought in black soil regions, characterized in that, Includes the following steps: Acquire multi-source input data, which includes: groundwater storage anomaly time series data, meteorological and hydrological time series data, and human activity time series data; Spatiotemporal variation analysis was performed on the groundwater storage anomaly time series data to extract trend and periodic components. The trend analysis used a combination algorithm of nonparametric slope estimation and significance test, and the signal decomposition used an ensemble empirical mode decomposition algorithm to decompose the original time series into several intrinsic mode functions and residual terms. The standardized groundwater drought index (SGDI) time series is calculated based on the acquired groundwater storage anomaly time series data, and drought events are identified according to the judgment threshold. The Coupled Drought Amplification Index (CDAI) is calculated based on the SGDI time-series data, using the following formula: ； in, This is the measured groundwater drought index calculated based on all input data. The simulated groundwater drought index is calculated based on meteorological and hydrological time series data after removing the aforementioned human activity time series data. The CDAI > 0 indicates the existence of a drought amplification effect. Based on a machine learning regression model, the SGDI time series data, human activity time series data, and meteorological and hydrological time series data are subjected to multi-factor contribution separation. The contribution rate of each driving factor is output, and the dominant type partitioning is performed based on the contribution rate to output the regional dominant type classification results.

2. The method according to claim 1, characterized in that, The groundwater storage anomaly time series data is obtained from satellite gravity measurement remote sensing inversion data; The meteorological and hydrological time-series data include air temperature, precipitation, and radiation flux; The human activity time-series data includes farmland expansion time-series data, irrigation water use time-series data, and socio-economic data; and The farmland expansion time series data is cultivated land classification data from optical remote sensing images; The irrigation water time series data is spatialized raster data of statistical data.

3. The method according to claim 1, characterized in that, The nonparametric slope estimation uses the Theil-Sen slope estimation to calculate the median slope, and the significance test uses the Mann-Kendall test. The trend was determined to be significant at that time; The set-based empirical mode decomposition algorithm suppresses mode aliasing by adding white noise to the original sequence.

4. The method according to claim 1 or 3, characterized in that, The calculation of SGDI and identification of drought events based on groundwater storage anomaly time series data includes: Using a preset time period as the baseline period, the mean μ and standard deviation σ of the groundwater storage anomaly values ​​within the baseline period are calculated. Based on the mean and standard deviation, the time-point data are standardized to obtain the SGDI time series. When the SGDI value is ≤-1.0 at three or more consecutive time points, it is determined to be a regional groundwater drought event, and the start and end times and duration of the drought event are output.

5. The method according to claim 1, characterized in that, The multi-factor contribution separation and dominant type partitioning based on the machine learning regression model includes: A random forest regression model was constructed using the annual rate of change of cultivated land area, annual precipitation, annual mean temperature and annual evapotranspiration as input variables and SGDI value as output variable. The importance measure of each input variable is calculated as the contribution rate based on the random forest regression model. When the contribution rate of farmland expansion factor is >60%, the output is farmland expansion-dominant type; When the contribution rate of climate factors is greater than 60%, the output is climate change-dominated. When the difference between the contribution rates of farmland expansion factor and climate factor is ≤20%, the output coupling is dominant.

6. The method according to claim 1, characterized in that, The method further includes: calculating the geometric centroid coordinates of the drought region at different time scales based on the spatial distribution data of the drought event, and outputting the temporal migration trajectory data of the geometric centroid coordinates.

7. An assessment system for the amplification effect of human activities on groundwater drought in black soil regions, characterized in that, include: The data input module is configured to acquire multi-source input data, including groundwater storage anomaly time series data, meteorological and hydrological time series data, and human activity time series data. The spatiotemporal analysis module is configured to perform trend analysis and signal decomposition on the groundwater storage anomaly time series data, and output trend component and periodic component; The drought identification module is configured to calculate the SGDI index based on the groundwater storage anomaly time series data and identify drought events according to the judgment threshold. The coupling amplification effect calculation module is configured to calculate the coupling drought amplification index (CDAI) based on the SGDI index, wherein the CDAI calculation formula is: ； in, This is the measured groundwater drought index calculated based on all input data. The simulated groundwater drought index is calculated based on meteorological and hydrological time-series data after removing the aforementioned human activity time-series data. The driving contribution separation module is configured to perform multi-factor contribution separation on the SGDI index and the time series data of human activities and meteorological and hydrological data based on a machine learning regression model, and output the contribution rate of each driving factor. The dominant type partitioning module is configured to perform dominant type partitioning based on the contribution rate and a preset threshold, and output the dominant type classification results for the region.

8. The evaluation system according to claim 7, characterized in that, The system also includes: The validation module is configured to perform leave-one-out cross-validation and output model accuracy metrics. The visualization output module is configured to generate and output SGDI spatial distribution map, CDAI spatial pattern map, and drought centroid migration trajectory map.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for assessing the amplification effect of human activities on groundwater drought in black soil areas according to any one of claims 1-6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for assessing the amplification effect of human activities on groundwater drought in black soil areas according to any one of claims 1-6.

Images