A water resource system resilience evolution monitoring and early warning method based on GTWR regression

Abstract

This invention discloses a method for monitoring and early warning of the resilience evolution of water resources systems based on GTWR regression, belonging to the field of water resources and environmental management technology. The method includes: constructing a resilience evaluation system for a water resources complex system in a specific watershed; acquiring multi-source data for nodes, determining the objective weights of indicators, and quantifying resilience scores; extracting spatial differentiation characteristics of resilience levels through spatial autocorrelation analysis, and identifying the set of driving factors using a geographic detector; introducing the GTWR model to calculate the spatiotemporal regression coefficients of each driving factor, dynamically identifying the heterogeneity of the driving mechanisms for resilience evolution; and establishing a graded early warning mechanism by combining resilience level standards and driving patterns. This invention solves the technical problem of traditional assessment methods neglecting spatiotemporal nonstationarity, and can accurately identify resilience bottlenecks in different regions and at different times, possessing high monitoring accuracy and practical value.

Description

Technical Field

[0001] This invention relates to the field of water resources management and resource environment assessment technology, and specifically to a method for monitoring and early warning of the resilience evolution of water resources systems based on GTWR regression. Background Technology

[0002] Against the backdrop of global climate change and increased human activity, water resource systems face heightened uncertainties and risks. Resilience, as an assessment of a system's ability to withstand shocks and restore its function, is a core indicator for the sustainable use of water resources. However, existing water resource assessment models are mostly static assessments or GWR models that only consider spatial heterogeneity. Due to the significant spatiotemporal nonstationarity of watershed systems—that is, the intensity of the same factor varies greatly in different years and different cities—traditional models struggle to capture these dynamic fluctuations, resulting in inaccurate early warning signals and an inability to provide timely and location-specific guidance for complex scheduling decisions. Summary of the Invention

[0003] (1) Technical problems to be solved

[0004] This invention aims to overcome the shortcomings of existing evaluation methods that ignore the non-stationarity of time and space coupling, and to provide a water resource system resilience monitoring method that can dynamically identify resilience bottlenecks and provide high-precision early warning.

[0005] (2) Technical solution

[0006] The technical solution adopted in this invention is as follows: S1. Constructing an indicator system and data preprocessing: Constructing an indicator system for evaluating the resilience of a water resource complex system from three dimensions: resistance, resilience, and adaptability; acquiring multi-source indicator data of each monitoring node within the monitoring area during the preset study period, and standardizing the raw data; S2. Comprehensive measurement of resilience level: Determining the weight of each indicator using an objective weighting method, calculating the comprehensive system resilience score of each monitoring node, and analyzing its temporal evolution trend using the Mann-Kendall test; S3. Extracting spatial differentiation features: Combining geographic spatial coordinates, calculating the Moran index using a spatial autocorrelation model to identify the spatial clustering pattern and heterogeneity characteristics of system resilience; 4. Intelligent screening of driving factors: The explanatory power of candidate driving factors is measured using the optimal parameter geographic detector model, and the core driving factor set that has a significant impact on the evolution of system resilience is screened out; S5. Construction of GTWR spatiotemporal regression model: The comprehensive score of system resilience is used as the dependent variable and the core driving factor set is used as the independent variable. A three-dimensional spatiotemporal coordinate system is constructed by introducing space and time; the optimal spatiotemporal bandwidth is determined by spatiotemporal kernel function and cross-validation method, and the regression coefficients of each monitoring node at different spatiotemporal points are calculated; S6. Identification of driving modes and establishment of early warning: Based on the positive and negative characteristics and intensity fluctuations of the regression coefficients, the spatiotemporal driving modes of system resilience are identified, and graded early warning instructions are output in combination with the resilience level classification standard.

[0007] (3) Beneficial effects

[0008] High monitoring accuracy: The introduction of the GTWR algorithm solves the problem of spatiotemporal nonstationarity in long-sequence monitoring, and can more realistically restore the system evolution trajectory than traditional models.

[0009] Precise bottleneck identification: It can quantify the core constraints for each year and each city, and directly pinpoint the "risk source" through the positive and negative changes of the regression coefficient.

[0010] Strong decision support: The early warning results include not only "risk level" but also "driving mechanism classification", providing digital support for the differentiated implementation of the "four waters and four fixed points" policy in different regions. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the overall process of a water resource system resilience evolution monitoring and early warning method based on GTWR regression according to the present invention;

[0012] Figure 2 This is a flowchart of the hierarchical early warning logic in an embodiment of the present invention. Detailed Implementation

[0013] Example: This study focuses on 44 cities at the prefecture-level administrative unit level in the middle and lower reaches of the Yellow River. The monitoring period was from 2008 to 2022, and the sampling frequency was annual. The specific steps are as follows:

[0014] Step 1: Indicator Construction and Data Standardization

[0015] A multi-dimensional indicator system was constructed, encompassing water resources, socio-economic, and ecological environment subsystems. Thirty technical indicators were set from three process dimensions: resistance, resilience, and adaptation.

[0016] Resistance indicators include those reflecting the system's ability to withstand external shocks, such as water consumption per 10,000 yuan of GDP and total water resources; recovery indicators include those reflecting the system's ability to recover to its original state after damage, such as water supply and water consumption in various industries.

[0017] Adaptability indicators include those reflecting the system's ability to self-adjust and evolve, such as the proportion of science and technology expenditure and the green coverage rate of built-up areas.

[0018] Raw spatiotemporal data were obtained by retrieving statistical yearbooks, water resources bulletins, and various raster maps from each monitoring node. The raw data were processed using the range standardization method to eliminate the influence of dimensions and obtain a standard value matrix.

[0019] Step 2: Quantification of Resilience Overall Score

[0020] Calculate the objective weights of each indicator using the entropy weight method. Using standardized data, the TOPSIS method was employed to calculate the attachment progress at each observation point, which was then used as the comprehensive resilience score of the water resources system for that node in that year. The calculation formula is as follows:

[0021]

[0022] in, The larger the value, the higher the system's resilience.

[0023] Step 3: Spatiotemporal Feature Extraction and Core Factor Screening

[0024] Spatial differentiation analysis: The global Moran index is calculated using a spatial autocorrelation model to determine the positive and negative correlation and clustering patterns of resilience levels in geographic space.

[0025] Core Factor Selection: An optimal geographic detector was introduced, with the overall resilience score as the dependent variable and each driving factor as the independent variable. Through iterative optimization to calculate the q-value, factors with an explanatory power greater than 0.3 and a significance level of P < 0.05 were selected to form the core driving factor set X. core This includes water supply, energy conservation and environmental protection expenditures, science and technology expenditures, and education expenditures (based on this embodiment).

[0026] Step 4: Construct the GTWR model for dynamic recognition

[0027] Introducing a spatiotemporal extended version of the geographic weighted regression model, three-dimensional spatiotemporal coordinates are established for each city. .

[0028] Establish a spatiotemporal nonstationary regression equation:

[0029]

[0030] Spatiotemporal bandwidth adaptive optimization: The Gaussian kernel function is used as the weight function, and the optimal spatiotemporal bandwidth is automatically optimized by minimizing the Akaike information criterion.

[0031] Driving coefficient calculation: The system uses the local least squares method to calculate the regression coefficients of each core driving factor at each spatiotemporal point. The sign of this coefficient represents the driving nature, and its magnitude represents the intensity of the influence.

[0032] Step 5: Resilience Evolution Monitoring and Early Warning Output

[0033] Based on the score calculated in step two and the driving coefficients calculated in step four Establish an early warning mechanism:

[0034] Grade classification: The toughness level is divided into three grades: high toughness (≥0.110), medium toughness (0.085~0.110), and low toughness (≤0.085).

[0035] Tiered early warning:

[0036] (1) If a city's resilience score If the trend shows a continuous decline for three consecutive years (according to this embodiment), and the GTWR model identifies a core factor (such as ecological input) whose positive driving coefficient continues to decrease, a yellow alert is triggered.

[0037] (2) A red alert is triggered if the resilience score is lower than the threshold of 0.085 (according to this embodiment) or if the resistance subsystem score shows an abnormal fluctuation of more than 30% in a single year.

[0038] Decision command generation and output: Based on the spatial heterogeneity of the regression coefficients output by the GTWR model, the system provides targeted governance suggestions. For constraint factors with large absolute values ​​of regression coefficients that are continuously decreasing, the system identifies them as "critical bottlenecks" and prompts the management to prioritize precise resource allocation in this area.

Claims

1. A method for monitoring and early warning of the resilience evolution of water resources systems based on GTWR regression, characterized in that, Includes the following steps: S1. Construction of indicator system and data preprocessing: Construct an indicator system for evaluating the resilience of water resource complex system from three dimensions: resistance, resilience and adaptability; obtain multi-source indicator data of each monitoring node in the monitoring area during the preset study period, and standardize the raw data; S2. Comprehensive measurement of resilience level: The weight of each indicator is determined by the objective weighting method, the comprehensive system resilience score of each monitoring node is calculated, and the time evolution trend is analyzed by combining the Mann-Kendall test. S3. Spatial Differentiation Feature Extraction: Combining geographic spatial coordinates, Moran's index is calculated using a spatial autocorrelation model to identify the spatial clustering patterns and heterogeneity characteristics of system resilience. S4. Intelligent screening of driving factors: The explanatory power of candidate driving factors is calculated by using the optimal parameter geographic detector model, and the core driving factor set that has a significant impact on the resilience evolution of the system is screened out. S5. Constructing the GTWR spatiotemporal regression model: Using the comprehensive system resilience score as the dependent variable and the core driving factor set as the independent variable, a three-dimensional spatiotemporal coordinate system is constructed by introducing space and time; the optimal spatiotemporal bandwidth is determined by the spatiotemporal kernel function and cross-validation method, and the regression coefficients of each monitoring node at different spatiotemporal points are calculated. S6. Identify driving modes and establish early warnings: Based on the positive and negative characteristics and intensity fluctuations of the regression coefficients, identify the spatiotemporal driving modes of system resilience, and output graded early warning instructions in conjunction with the resilience level classification standards.

2. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S1, the multi-source indicator data includes water resource quantity data, water quality monitoring data, socio-economic statistics data, and ecological environment data; the standardization process adopts the range standardization method.

3. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S2, the objective weighting method adopts the entropy weight method, which calculates the entropy weight based on the dispersion of the original data sequence of each indicator, and determines the contribution of each indicator in the resilience evaluation through information entropy.

4. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S2, the Mann-Kendall test determines the significance of the trend by calculating the standardized test statistic |Z|: when |Z|>1.96, it is determined that the resilience of the water resource system shows a significant upward or downward evolution trend in the time dimension.

5. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S3, the Moran index includes a global Moran index and a local Moran index, which are used to identify high-high clustering, low-low clustering, and spatial outliers of system resilience in space.

6. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S4, the optimal geographic detector model selects the factor with the largest explanatory power q value and that passes the significance test as the core driving factor by comparing the q values ​​under different spatial stratification schemes.

7. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S5, the regression coefficients of the GTWR spatiotemporal regression model are calculated using the following mathematical model: in, For the first i The spatiotemporal coordinates of each monitoring node. This represents the regression coefficient of the node at a specific point in time and space.

8. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S5, the spatiotemporal kernel function is constructed using a Gaussian kernel function or a double-square kernel function to construct the spatiotemporal weight matrix, and the optimal spatiotemporal bandwidth is determined by minimizing the modified Akaike information criterion.

9. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression as described in claim 1, characterized in that: In step S6, the spatiotemporal driving mode for identifying system resilience refers to: determining the promoting or inhibiting effect of factors based on the positive or negative attributes of regression coefficients, and identifying the evolution type of the influence of each factor based on the changing trend of the absolute value of the coefficient over time.

10. The method for monitoring and early warning of water resource system resilience evolution based on GTWR regression according to claim 1, characterized in that: In step S6, the graded early warning instruction maps the resilience score of real-time monitoring points to three preset early warning level intervals, namely, safety, yellow warning and red warning, and generates a spatiotemporal dynamic early warning map covering the entire monitoring area in real time.

Images