Method and system for dynamic period management of groundwater monitoring data based on multiple time scales

Abstract

The application relates to a kind of multi-time scale based groundwater monitoring data dynamic period management method and system, belong to groundwater monitoring and data analysis technical field.The groundwater data time series is decomposed into different scale periods by db4 wavelet decomposition, the identified period is verified by wavelet power spectrum significance test, and the core period of the groundwater level data in the region is determined according to the number of significant sites.According to the time scale of the core period, the driving factors are classified according to the influence, the water level of each core period of each site is subjected to Spearman correlation test with the main driving factor components, and the rank correlation coefficient of water level and each driving factor component is calculated.A contribution matrix of the region is formed by constructing period-leading factor-contribution weight;Hierarchical management and zoning management are carried out.The application greatly improves the analysis accuracy and regional adaptability.

Description

Technical Field

[0001] This invention relates to a dynamic periodic management method and system for groundwater monitoring data based on multiple time scales, belonging to the field of groundwater monitoring and data analysis technology. Background Technology

[0002] Existing technologies have many limitations in statistically analyzing time-series data from groundwater monitoring stations, making it difficult to meet the needs of precise and regional groundwater management. Specific problems include:

[0003] (1) Single analysis methods are insufficient to account for the nonlinearity and multi-scale periodicity of data: Groundwater systems are affected by seasonal precipitation brought by monsoon climate, the heterogeneity of aquifers caused by complex geological structures, and the superimposed effects of high-intensity human activities (such as agricultural irrigation and industrial mining). As a result, the data exhibit significant strong nonlinear correlations and multi-scale periodic fluctuations. For example, a sudden increase in agricultural mining can lead to abrupt changes in groundwater levels, which is a nonlinear feature; while changes in groundwater levels include periodic fluctuations at different scales such as monthly, annual, and interdecadal. Traditional linear analysis methods (such as the Pearson correlation test) cannot capture this nonlinear correlation, and single periodicity detection methods (such as Fourier transform) are also difficult to identify multi-scale periodicity features at the same time, which can easily lead to the loss of key information and a one-sided understanding of the dynamic changes of groundwater.

[0004] (2) Lack of targeted regional adaptation design: The hydrogeological conditions of different regions are complex and diverse. Some regions have brackish water interaction, some regions are dominated by fractured aquifers with groundwater recharge lag, and some regions have special precipitation distribution due to special weather conditions. Existing general data analysis algorithms do not take into account the unique hydrogeological conditions of different regions, resulting in deviations in period identification and correlation analysis. For example, in the mountainous area of ​​central Shandong, due to the lag in the recharge of fractured aquifers, the water level fluctuation cycle during the rainy season is not completely synchronized with the precipitation cycle, and general algorithms may misjudge the water level fluctuation cycle during the rainy season; when analyzing the correlation between precipitation and water level, general methods ignore the impact of differences in soil water retention capacity in different regions on the correlation between the two, leading to misinterpretation of the correlation.

[0005] (3) Insufficient analysis of multi-scale driving mechanisms: Groundwater dynamics are the result of the combined effects of multiple driving factors at different time scales. For example, seasonal precipitation mainly affects monthly water level changes, climate anomalies such as El Niño mainly affect interannual water level fluctuations, while long-term extraction policies play a dominant role in interdecadal water level trends. Existing methods mostly analyze the relationship between groundwater and driving factors at a single time scale, failing to distinguish the contributions of factors at different scales. This makes it impossible to clarify the influence weight of each driving factor at different time dimensions, resulting in a lack of hierarchical basis for water resource management strategies and making it difficult to achieve precise control. Summary of the Invention

[0006] The purpose of this invention is to overcome the above-mentioned shortcomings and provide a dynamic periodic management method and system for groundwater monitoring data based on multiple time scales, which improves the analysis accuracy and regional adaptability.

[0007] The technical solution adopted in this invention is as follows: The method for dynamic periodic management of groundwater monitoring data based on multiple time scales includes the following steps: S1. Select a region covering multiple typical hydrogeological units as the research object, obtain monthly average water level data from multiple groundwater monitoring stations over many years, as well as driving factor data during the same period, and perform data preprocessing. S2. The time series of groundwater data is decomposed into periods of different scales using db4 wavelet decomposition. The identified periods are verified by wavelet power spectrum significance test. The core period of groundwater level data in the area is determined based on the number of stations with strong significance. S3. Based on the time scale of the core cycle, the driving factors of each core cycle are classified according to the magnitude of their impact as: seasonal factors, interannual factors, and interdecadal factors. S4. Perform Spearman correlation test on the water level of each station in each core cycle and each component of its main driving factors, and calculate the rank correlation coefficient between the water level and each component of the driving factors. S5. Take the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor, and use the absolute value of the Spearman rank correlation coefficient as the contribution weight of the dominant factor to construct the contribution matrix of the region by period-dominant factor-contribution weight. S6. Implement time-scale hierarchical management based on the dominant factors of different cycles, generate regional difference reports using contribution matrices, and implement zoning management based on regional differences.

[0008] In the above method, the data preprocessing in step S1 includes data cleaning, missing value imputation, and outlier imputation; Missing values ​​were filled using a "precipitation-weighted interpolation method", the calculation formula of which is: missing value = Σ (water level of the three adjacent months × proportion of precipitation in the corresponding month), where the proportion of precipitation in the corresponding month = precipitation in that month / total precipitation in the three adjacent months; Outlier imputation is handled using the "moving median of adjacent 3 months" correction method, which calculates the median of the month before, month after, and month of the outlier (after removing the outlier) as the correction value.

[0009] The driving factor data include precipitation, temperature, agricultural output, El Niño index, and the commissioning time of water conservancy projects.

[0010] The db4 wavelet decomposition scale described in step S2 is set to 8 layers, ranging from 1 to 256 months, specifically divided as follows: layers 1-2 are used to capture short-term random fluctuations, mainly corresponding to decadal fluctuations; layers 3-4 are used to identify seasonal cycles, corresponding to cycles of 12-24 months, which match the precipitation cycle of the region; layers 5-6 are used to extract interannual cycles, corresponding to cycles of 3-8 years, mainly associated with El Niño climate anomaly cycles; layers 7-8 are used to analyze interdecadal trends, corresponding to cycles of 10-20 years, mainly reflecting the impact of long-term mining on groundwater levels.

[0011] The seasonal factors mentioned in step S3 include monthly precipitation and irrigation extraction, the interannual factors include annual average temperature and El Niño index, and the interdecadal factors include cumulative extraction and the commissioning time of water conservancy projects.

[0012] The core of the regional disparity report in step S6 is to divide the area into hydrogeological units and conduct comparative analysis of core data, specifically including: (1) Divide the region into hydrogeological units such as plain irrigation area and mountain fissure aquifer, extract the contribution matrix data of monitoring stations in each region, and calculate the average contribution weight ratio of each region in different periods. (2) Compare the core dominant cycles, dominant factor types, and weight values ​​of each region to clarify the core differences between regions (e.g., some regions have a high proportion of seasonal cycles, while some regions have a high proportion of interdecadal cycles). (3) Summarize the water level driving characteristics of each region, analyze the core reasons for the differences (hydrogeological conditions, human activities, etc.), and present them in a concise table and text summary to form a report.

[0013] Another objective of this invention is to provide a dynamic periodic management system for groundwater monitoring data based on multiple time scales, including a data acquisition and preprocessing module, a multi-scale period division module, a significance calculation module, a core period screening module, a Spearman correlation analysis module for driving factors, a contribution matrix construction module, and a hierarchical and regional management module. The data acquisition and preprocessing module is used to select areas covering multiple typical hydrogeological units as research objects, acquire monthly average water level data from multiple groundwater monitoring stations over many years, as well as driving factor data from the same period, and perform data preprocessing. The multi-scale period partitioning module uses db4 wavelet decomposition to decompose the groundwater data time series into periods of different scales; The significance calculation module verifies the identified periodicity through wavelet power spectrum significance calculation. The core period selection module determines the core period of groundwater level data in the area based on the number of highly significant stations. The Spearman correlation analysis module for driving factors is used to classify the driving factors according to the magnitude of their influence in each core cycle based on the time scale of the core cycle, and to perform Spearman correlation tests on the water level of each station in each core cycle and each component of its main driving factors, and to calculate the rank correlation coefficient between the water level and each component of the driving factors. The contribution matrix construction module is used to construct a contribution matrix for the region by taking the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor and the contribution weight of the dominant factor with the absolute value of the Spearman rank correlation coefficient. The hierarchical and regional management module performs hierarchical management on a time scale based on the dominant factors of different periods, generates regional difference reports using contribution matrices, and performs regional management based on regional differences.

[0014] The beneficial effects of this invention are: 1. Significantly improved analysis accuracy: Compared with the single wavelet analysis method, this invention can more accurately identify the driving factors at different periodic scales by performing Spearman correlation test in different periods. It corrects the misjudgment of "precipitation dominates all scales" in the traditional method and clarifies the dominant role of agricultural mining on the seasonal scale and long-term over-extraction on the interdecadal scale.

[0015] 2. Significantly Enhanced Regional Adaptability: A preprocessing method with regional adaptability was designed, and the data processing workflow was optimized for the hydrogeological characteristics of different regions. This improved the reliability of analysis results for stations with a data missing rate >20% to 85%, far exceeding the approximately 60% of existing general methods.

[0016] 3. Facilitates tiered and zoned management: The constructed contribution matrix clearly shows the influence weights of driving factors at different scales, directly serving the formulation of "tiered control strategies": In the short term (within 1 year), seasonal water level fluctuations can be stabilized by adjusting the amount of water extracted during the irrigation period; in the medium term (3-5 years), emergency water supply plans can be formulated based on El Niño index predictions to address the impact of climate anomalies; in the long term (more than 10 years), total extraction limits can be controlled to curb the trend of continuous water level decline.

[0017] 4. The method is highly universal and easy to promote: The method of this invention can be transferred to various hydrological zones. Only parameters such as wavelet decomposition scale need to be adjusted to adapt to the hydrological characteristics of different regions. The groundwater management system built based on this method can be directly applied to groundwater monitoring and control scenarios. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The present invention will be further described below with reference to specific embodiments.

[0020] Example 1: A dynamic periodic management method for groundwater monitoring data based on multiple time scales, comprising the following steps: S1. A region encompassing multiple typical hydrogeological units was selected as the research object. Monthly average water level data from multiple groundwater monitoring stations over several years, along with concurrent driving factor data, were obtained, and data preprocessing was performed. The study selected a region encompassing two typical hydrogeological units: plain irrigation areas and mountain fissure aquifers. Monthly average water level data from 15 groundwater monitoring stations from 2004 to 2024 were selected, along with driving factor data such as precipitation, temperature, and agricultural extraction during the same period.

[0021] Data preprocessing (regional adaptation optimization): (1) Data source: Groundwater level data comes from the monitoring network of the regional hydrological center with a monitoring accuracy of ±0.01m, including the monthly average water level record of each station; among the driving factor data, precipitation data comes from the regional meteorological station with an accuracy of ±0.1mm, temperature data is the monthly average temperature of the same period, and agricultural extraction data comes from the extraction ledger of the regional water conservancy bureau, which records the monthly groundwater extraction volume of each irrigation area.

[0022] (2) Cleaning rules: First, the raw data is thoroughly checked and values ​​that jump due to instrument failure are removed. Data with daily water level fluctuation > 1m are set as jump values. Such data are usually caused by monitoring instrument failure or human recording errors and do not reflect the real dynamic changes of groundwater. Then, the stations are screened and stations with continuous observation duration ≥ 15 years are retained. Finally, 12 valid stations are selected from 15 stations to ensure that the data has a sufficient time series length to support periodic analysis.

[0023] (3) Missing value imputation: To address the data missing problem at mountain stations during the rainy season (June-September), the "precipitation weighted interpolation method" proposed in this invention is used for imputation. The specific calculation formula of this method is: missing value = Σ (water level of 3 adjacent months × proportion of precipitation in the corresponding month), where the proportion of precipitation in the corresponding month = precipitation in that month / total precipitation in 3 adjacent months. For example, water level data for a certain mountain station in July 2010 is missing. The water levels for the adjacent months of June, August, and September are 52.3m, 51.8m, and 51.5m respectively. Rainfall in June was 180mm, in August 220mm, and in September 150mm. The total rainfall for these three months is 180 + 220 + 150 = 550mm. Therefore, the proportion of rainfall in June is approximately 180 / 550 ≈ 0.327, in August it's approximately 220 / 550 ≈ 0.4, and in September it's approximately 150 / 550 ≈ 0.273. Thus, the missing water level value for July 2010 is approximately 52.3 × 0.327 + 51.8 × 0.4 + 51.5 × 0.273 ≈ 51.9m. Compared to the traditional mean method, this method reduces the filling error by 32% and better reflects the impact of rainy season rainfall on groundwater levels.

[0024] (4) Outlier handling: Outliers are identified using the IQR method. IQR is short for Interquartile Range, which is the difference between the upper quartile (Q3) and the lower quartile (Q1), i.e., IQR = Q3 - Q1. It is commonly used to identify outliers in data. The principle is to determine data that exceeds Q3 + 1.5IQR or is lower than Q1 - 1.5IQR as outliers. The IQR of water level data for each station is calculated. For the identified outliers, they are verified in conjunction with the mining records of the same period to confirm whether the sudden drop in water level is caused by reasonable reasons such as over-extraction during the peak irrigation period. If it is a reasonable abrupt change, the "moving median of adjacent 3 months" is used for correction. That is, the median of the month before, the month after, and the month of the outlier (after removing the outlier) are calculated as the correction value to retain the trend of the abrupt change but remove the measurement error. For example, the water level data of a certain station in May 2015 was 48.2m. Calculations showed Q1=50.1m, Q3=51.3m, IQR=1.2m, and Q1-1.5IQR=50.1-1.8=48.3m. This data was lower than this value and was therefore considered an outlier. Combined with the extraction records, it can be seen that this month was the peak irrigation period, indicating over-extraction. The water level in April was 50.5m, and the water level in June was 49.8m. Therefore, the correction value is the median of 49.8 and 50.5 (after removing the outlier in May), i.e., (49.8+50.5) / 2=50.15m.

[0025] S2. The db4 wavelet decomposition method is used to decompose the groundwater data time series into periods of different scales. The identified periods are verified by wavelet power spectrum significance test. The core period of the groundwater level data in this area is determined based on the number of stations with strong significance. Wavelet Basis and Scale Selection: Based on the characteristics of groundwater level data—"sharp short-term fluctuations and stable long-term trends"—the db4 wavelet was selected as the basis function. The db4 wavelet, short for Daubechies 4 wavelet, is an orthogonal wavelet basis function with compact support properties, proposed by Ingrid Daubechies. It exhibits good time-frequency localization performance in the multi-scale decomposition of non-stationary time series and is suitable for capturing abrupt changes and periodic features in groundwater level data. The db4 wavelet's compact support property allows it to well adapt to abrupt changes in the data, such as sudden drops in water level caused by irrigation; simultaneously, its good time-frequency resolution helps to accurately identify periods at different scales.

[0026] The decomposition scale is set to 8 layers, with each layer corresponding to a period range of 1-256 months. Specifically, the layers are divided as follows: Layers 1-2 are used to capture short-term random fluctuations, mainly corresponding to decadal fluctuations caused by human activities such as irrigation and water diversion; Layers 3-4 are used to identify seasonal cycles, corresponding to a 12-24 month cycle, which matches the precipitation cycle of the region; Layers 5-6 are used to extract interannual cycles, corresponding to a 3-8 year cycle, mainly associated with climate anomaly cycles such as El Niño; Layers 7-8 are used to analyze interdecadal trends, corresponding to a 10-20 year cycle, mainly reflecting the impact of long-term extraction and other factors on groundwater levels.

[0027] Periodicity Verification: The identified periods were verified using a wavelet power spectrum significance test (α=0.05). The wavelet power spectrum represents the energy intensity of different periodic components; a higher power value indicates a more significant period. The pass rate was assumed to be the proportion of effective monitoring stations with significant groundwater level periodicity characteristics determined by the wavelet power spectrum significance test at the set significance level (α=0.05), relative to the total number of effective groundwater monitoring stations in the study area. The test results showed that the pass rate for the 12-month period was 92%, meaning that 11 out of 12 effective stations showed significant power values ​​for this period; the pass rate for the 48-month period was 75%, with 9 stations showing significant power; and the pass rate for the 144-month period was 68%, with 8 stations showing significant power. Therefore, 12 months, 48 ​​months, and 144 months were identified as the core periods for groundwater level data in this area.

[0028] S3. Based on the time scale of the core cycle, the driving factors of each core cycle are classified according to their magnitude of influence as: seasonal factors, interannual factors, and interdecadal factors. Driving factor stratification: Based on the time scale of the core cycle, driving factors are divided into the following three categories: Seasonal factors: including monthly precipitation and irrigation output, which mainly affect water level fluctuations over a 12-month cycle; Interannual factors: including average annual temperature and El Niño index, mainly affect water level changes over a 48-month cycle; Interdecadal factors: These include cumulative extraction volume and the commissioning time of water conservancy projects, and mainly affect water level trends over a 144-month period.

[0029] S4. Perform Spearman correlation tests on the water level of each station for each core cycle and each component of its main driving factors, and calculate the rank correlation coefficient between the water level and each component of the driving factors: The Spearman Rank Correlation Coefficient (ρ) measures the strength and direction of the monotonic correlation between two variables, ranging from -1 to 1. The closer the absolute value is to 1, the stronger the monotonic association between the variables. The significance level (p) is a probability value used in statistical hypothesis testing to determine the reliability of the results. When the p-value is less than 0.05, the correlation between the variables is generally considered statistically significant.

[0030] Periodic testing: 12-month cycle: Spearman correlation tests were performed on the water level components of each station over the 12-month cycle with monthly precipitation and irrigation output. The calculated rank correlation coefficient between water level and monthly precipitation was ρ = 0.68 (p = 0.002), indicating a significant positive correlation between the two; the rank correlation coefficient between water level and irrigation output was ρ = -0.73 (p = 0.001), indicating a significant negative correlation between the two. By comparing the absolute values ​​of ρ, it can be seen that agricultural output has a more significant impact on seasonal fluctuations.

[0031] 48-month cycle: The water level component of the 48-month cycle was tested with the annual average temperature and the El Niño index. The results showed that there was a significant negative correlation between water level and the El Niño index (ρ=-0.52, p=0.03), revealing the inhibitory effect of El Niño years (drought) on water level; while the correlation between water level and annual average temperature was not significant (p=0.07).

[0032] 144-month cycle: The water level component over the 144-month cycle was examined in relation to the cumulative extraction volume and the commissioning time of water conservancy projects. It was found that the water level and the cumulative extraction volume had a strong negative correlation (ρ=-0.81, p=0.0005), confirming that long-term over-extraction is the main cause of the decline in water level. The correlation between water level and the commissioning time of water conservancy projects was relatively weak (p=0.12).

[0033] S5. Take the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor, and use the absolute value of the Spearman rank correlation coefficient as the contribution weight of the dominant factor to construct a contribution matrix for this region consisting of period, dominant factor, and contribution weight: Based on the results of the periodic test, the contribution weight of each driving factor in the corresponding period (i.e., the absolute value of the Spearman rank correlation coefficient) is calculated, and the following matrix is ​​constructed: .

[0034] S6. Implement tiered management based on the dominant factors of different periods, generate regional difference reports using contribution matrices, and manage by region based on these differences: A regional disparity report is generated using a period-dominant factor-contribution weight contribution matrix. The core principle is to use hydrogeological units as the basis for zoning, conduct multi-dimensional comparative analysis around the three core elements of the contribution matrix, and combine monitoring station data from different sub-regions within the study area (such as plain irrigation areas and mountainous fractured aquifers) to quantify differences and summarize characteristics. The specific steps are as follows: (1) Determine the report zoning and data basis. Divide the comparative sub-regions according to the typical hydrogeological unit types in the study area (such as plain irrigation area, mountain fissure aquifer, freshwater and saltwater interaction area, etc.), extract the contribution matrix data of all monitoring stations in each sub-region, calculate the average contribution weight of the dominant factor at different period scales in each sub-region, and form the standardized contribution matrix of each zoning (to eliminate single-site data bias).

[0035] (2) Conduct multi-dimensional comparative analysis of differences, with the standardized contribution matrix as the core, and compare the differences in groundwater level driving characteristics of each sub-region from three dimensions: Cycle Scale Dimension: Compare the differences in the proportion of core cycles (seasonal / annual / decadal) in each partition, that is, the proportion of the sum of the contribution weights of different cycles in the total weight of the partition; Dominant Factor Dimension: Compare whether the dominant factors of each partition are consistent under the same period scale. If they are inconsistent, identify the characteristic dominant factors of each partition. Contribution weight dimension: Compare the differences in contribution weight values ​​of the same period-dominant factor combination in each region to quantify the strength of the factor's influence. [Example Reference] In the comparison between plain irrigation areas and mountain fissure aquifers, the contribution of the 12-month seasonal cycle in the plain irrigation area accounts for 65%, while the contribution of the 144-month interdecadal cycle in the mountain area accounts for 58%, which intuitively reflects the core difference in cycle scale.

[0036] (3) Quantitative representation and feature summarization of regional differences: Quantify and label the above comparison results (e.g., use proportion and weight difference to represent the degree of difference) and summarize the features to clarify the core laws of groundwater level change in each sub-region: such as a certain zone being "seasonal factor-dominated", a certain zone being "decadal factor-dominated", or a certain zone having the characteristics of multi-factor balanced influence. At the same time, analyze the core reasons for the formation of differences (e.g., hydrogeological conditions, intensity of human activities, climate response characteristics, etc.).

[0037] (4) Compile a regional difference report. The report should include four parts: basic information, comparative analysis results, summary of difference characteristics, and causal analysis. It should be presented in tables and visualization charts (such as bar charts to compare the periodic contribution ratio of each region and radar charts to show the differences in factor weights). At the same time, the report should clearly define the core feature labels of the contribution matrix of each region (such as "plain irrigation area - agricultural irrigation-dominated - significant seasonal fluctuations" and "mountain area - cumulative mining-dominated - significant long-term trend changes") to provide direct conclusion support for subsequent regional management.

[0038] Report output example (with examples): .

[0039] Example 2: A dynamic periodic management system for groundwater monitoring data based on multiple time scales, including a data acquisition and preprocessing module, a multi-scale period division module, a significance calculation module, a core period screening module, a Spearman correlation analysis module for driving factors, a contribution matrix construction module, and a hierarchical and regional management module. The data acquisition and preprocessing module is used to select areas covering multiple typical hydrogeological units as research objects, acquire monthly average water level data from multiple groundwater monitoring stations over many years, as well as driving factor data from the same period, and perform data preprocessing. The multi-scale period partitioning module uses db4 wavelet decomposition to decompose the groundwater data time series into periods of different scales; The significance calculation module verifies the identified periodicity through wavelet power spectrum significance calculation. The core period selection module determines the core period of groundwater level data in the area based on the number of highly significant stations. The Spearman correlation analysis module for driving factors is used to classify the driving factors according to the magnitude of their influence in each core cycle based on the time scale of the core cycle, and to perform Spearman correlation tests on the water level of each station in each core cycle and each component of its main driving factors, and to calculate the rank correlation coefficient between the water level and each component of the driving factors. The contribution matrix construction module is used to construct a contribution matrix for the region by taking the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor and the contribution weight of the dominant factor with the absolute value of the Spearman rank correlation coefficient. The hierarchical and regional management module performs hierarchical management on a time scale based on the dominant factors of different periods, generates regional difference reports using contribution matrices, and performs regional management based on regional differences.

[0040] The above is a further description of the present invention in conjunction with specific embodiments, and the scope of protection of the present invention is not limited thereto.

Claims

1. A dynamic periodic management method for groundwater monitoring data based on multiple time scales, characterized by: The steps include the following: S1. Select a region covering multiple typical hydrogeological units as the research object, obtain monthly average water level data from multiple groundwater monitoring stations over many years, as well as driving factor data during the same period, and perform data preprocessing. S2. The time series of groundwater data is decomposed into periods of different scales using db4 wavelet decomposition. The identified periods are verified by wavelet power spectrum significance test. The core period of groundwater level data in the area is determined based on the number of stations with strong significance. S3. Based on the time scale of the core cycle, the driving factors of each core cycle are classified according to the magnitude of their impact as: seasonal factors, interannual factors, and interdecadal factors. S4. Perform Spearman correlation test on the water level of each station in each core cycle and each component of its main driving factors, and calculate the rank correlation coefficient between the water level and each component of the driving factors. S5. Take the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor, and use the absolute value of the Spearman rank correlation coefficient as the contribution weight of the dominant factor to construct the contribution matrix of the region by period-dominant factor-contribution weight. S6. Implement time-scale hierarchical management based on the dominant factors of different cycles, generate regional difference reports using contribution matrices, and implement zoning management based on regional differences.

2. The method for dynamic periodic management of groundwater monitoring data based on multiple time scales as described in claim 1, characterized in that, The data preprocessing described in step S1 includes data cleaning, missing value imputation, and outlier imputation.

3. The dynamic periodic management method for groundwater monitoring data based on multiple time scales as described in claim 2, characterized in that, The missing value imputation method uses precipitation-weighted interpolation, and its calculation formula is: missing value = Σ (water level of the three adjacent months × proportion of precipitation in the corresponding month), where the proportion of precipitation in the corresponding month = precipitation in that month / total precipitation in the three adjacent months; The outlier imputation process uses the moving median correction method for adjacent three months, that is, the median of the month before, the month after, and the month of the outlier is calculated as the correction value.

4. The method for dynamic periodic management of groundwater monitoring data based on multiple time scales as described in claim 1, characterized in that, The driving factor data include precipitation, temperature, agricultural output, El Niño index, and the commissioning time of water conservancy projects.

5. The method for dynamic periodic management of groundwater monitoring data based on multiple time scales as described in claim 1, characterized in that, The db4 wavelet decomposition scale described in step S2 is set to 8 layers, ranging from 1 to 256 months, specifically divided as follows: layers 1-2 are used to capture short-term random fluctuations, mainly corresponding to decadal fluctuations; layers 3-4 are used to identify seasonal cycles, corresponding to cycles of 12-24 months, which match the precipitation cycle of the region; layers 5-6 are used to extract interannual cycles, corresponding to cycles of 3-8 years, mainly associated with El Niño climate anomaly cycles; layers 7-8 are used to analyze interdecadal trends, corresponding to cycles of 10-20 years, mainly reflecting the impact of long-term mining on groundwater levels.

6. The method for dynamic periodic management of groundwater monitoring data based on multiple time scales as described in claim 1, characterized in that, The seasonal factors mentioned in step S3 include monthly precipitation and irrigation extraction, the interannual factors include annual average temperature and El Niño index, and the interdecadal factors include cumulative extraction and the commissioning time of water conservancy projects.

7. The method for dynamic periodic management of groundwater monitoring data based on multiple time scales as described in claim 1, characterized in that, The core of the regional disparity report in step S6 is to divide the area into hydrogeological units and conduct comparative analysis of core data, specifically including: (1) Divide the regions according to hydrogeological units, extract the contribution matrix data of monitoring stations in each region, and calculate the average contribution weight ratio of each region in different periods. (2) Compare the core dominant cycles, dominant factor types, and weight values ​​of each region to clarify the core differences between regions; (3) Summarize the water level driving characteristics of each region, analyze the core reasons for the differences, and present them in a concise table and text summary to form a report.

8. A dynamic periodic management system for groundwater monitoring data based on multiple time scales, characterized in that: It includes modules for data acquisition and preprocessing, multi-scale period segmentation, significance calculation, core period screening, Spearman correlation analysis of driving factors, contribution matrix construction, and hierarchical and regional management. The data acquisition and preprocessing module is used to select areas covering multiple typical hydrogeological units as research objects, acquire monthly average water level data from multiple groundwater monitoring stations over many years, as well as driving factor data from the same period, and perform data preprocessing. The multi-scale period partitioning module uses db4 wavelet decomposition to decompose the groundwater data time series into periods of different scales; The significance calculation module verifies the identified periodicity through wavelet power spectrum significance calculation. The core period selection module determines the core period of groundwater level data in the area based on the number of highly significant stations. The Spearman correlation analysis module for driving factors is used to classify the driving factors according to the magnitude of their influence in each core cycle based on the time scale of the core cycle, and to perform Spearman correlation tests on the water level of each station in each core cycle and each component of its main driving factors, and to calculate the rank correlation coefficient between the water level and each component of the driving factors. The contribution matrix construction module is used to construct a contribution matrix for the region by taking the driving factor component with the largest absolute value of the Spearman rank correlation coefficient as the dominant factor and the contribution weight of the dominant factor with the absolute value of the Spearman rank correlation coefficient. The hierarchical and regional management module performs hierarchical management on a time scale based on the dominant factors of different periods, generates regional difference reports using contribution matrices, and performs regional management based on regional differences.

Images