Intelligent analysis method for traceability proportion of mixed limestone water based on chlorine isotope

Abstract

The invention discloses a chlorine isotope-based mixed limestone water traceability proportion intelligent analysis method, which comprises the following steps of 1, acquiring multi-dimensional data, and constructing a spatial database containing geologic structure characteristics; 2, analyzing the mixed water sample; 3, constructing a hybrid model, and calculating a mixing ratio through a Bayesian neural network; step 4, performing uncertainty quantification, generating probability distribution of the mixing proportion, and outputting 95% confidence interval of the mixing proportion; according to the method, the contribution proportion of each end member water source in mixed water is quantitatively analyzed through the natural difference of the chlorine isotope in different water sources in combination with a mass conservation principle and an intelligent algorithm, the characteristics of small fractionation and remarkable end member characteristics of the chlorine isotope are fully utilized, and the chlorine isotope serves as a core tracing index; the problems that a traditional hydrochemical analysis method depends on the concentration ratio of ions such as Cl <-> and Na < + > during tracing, is easily interfered by evaporation concentration, mineral dissolution or man-made pollution, and easily causes misjudgment of water source contribution are solved.

Description

Technical Field

[0001] The present invention relates to the technical fields of hydrogeology and isotope geochemistry, and in particular to an intelligent analysis method for the tracing proportion of mixed limestone water based on chlorine isotopes. Background Art

[0002] Limestone aquifers usually receive multiple water source replenishments, such as precipitation, surface water, seawater (coastal areas), deep tectonic water, etc. The mixing process is complex, and it is difficult for traditional hydrochemical methods to distinguish the contributions of each water source. The dissolution-precipitation of groundwater and carbonate rocks (such as calcite and dolomite) in limestone areas is significant, resulting in the masking of water source characteristics by the changes in conventional ion concentrations. It is necessary to rely on conservative indicators to eliminate interference. Industrial or agricultural pollutants (such as chlorinated organic compounds and pesticides) infiltrate into the aquifer through different paths. Tracing analysis can locate the main pollution sources and quantify their contributions, supporting accountability and remediation. Mine drainage or tunnel engineering may disturb the groundwater system. Analyzing the sources of mixed water can warn of water inrush risks (such as determining whether the water inrush comes from cave water or fault water), guiding engineering protection design. Therefore, the analysis of the tracing proportion of mixed limestone water is a key tool for cracking the "invisible groundwater black box", and its necessity stems from the deepening of scientific understanding, the refined demand for resource management, and the driving force of technological method innovation;

[0003] When using traditional hydrochemical analysis methods for tracing, it depends on the concentration ratios of ions such as Cl - , Na + etc., which are easily interfered by evaporation concentration, mineral dissolution or human pollution, and are prone to misjudging the contribution of water sources. In limestone aquifers, the sensitivity may be insufficient due to evaporation fractionation or insignificant end-member differences (such as the isotopic convergence of deep-shallow water). At the same time, traditional hydrochemical analysis methods rely on artificial experience or simple linear models, with low efficiency and obvious error transmission, and it is difficult to cope with complex mixing scenarios. Therefore, the present invention proposes an intelligent analysis method for the tracing proportion of mixed limestone water based on chlorine isotopes to solve the problems existing in the prior art. Summary of the Invention

[0004] Aiming at the above problems, the purpose of the present invention is to propose an intelligent analysis method for the tracing proportion of mixed limestone water based on chlorine isotopes. This intelligent analysis method for the tracing proportion of mixed limestone water quantitatively analyzes the contribution ratios of each end-member water source in the mixed water by using the natural differences of chlorine isotopes in different water sources, combining the principle of mass conservation and intelligent algorithms, and makes full use of the characteristics of chlorine isotopes with small fractionation and significant end-member characteristics to become the core tracer index, solving the problem that traditional hydrochemical analysis methods rely on the concentration ratios of ions such as Cl - , Na + etc., which are easily interfered by evaporation concentration, mineral dissolution or human pollution and are prone to misjudging the contribution of water sources;

[0005] Through the closed-loop process of "feature library construction → intelligent modeling → dynamic optimization → uncertainty quantification", the geochemical characteristics of chlorine isotopes are combined with artificial intelligence technology to achieve high-precision analysis of the complex limestone water mixing process, providing a scientific tool for water resource management and environmental protection.

[0006] To achieve the purpose of the present invention, the present invention is realized through the following technical solutions: An intelligent analysis method for the mixing limestone water tracing ratio based on chlorine isotopes, comprising the following steps:

[0007] Step 1: Multi-dimensional data collection. Set multiple sampling points according to the hydrogeological conditions of the study area to capture spatial geological differences, and sample respectively during the wet season and the dry season to capture seasonal differences, and construct a spatial database containing geological structure characteristics;

[0008] Step 2: Analysis of mixed water samples. Use a high-precision mass spectrometer to measure the δ 37 Cl value of the mixed water sample, use ion exchange chromatography to separate the eluent from the water sample and then detect the Cl - purity in the eluent, and use a conductivity meter to measure the conductivity and salinity in the water sample;

[0009] Step 3: Construction of a mixing model. Establish an equation through the conservation of chlorine isotopes and chloride ion concentration to obtain a mass balance equation, use the Bayesian-neural network coupling to construct a model, and calculate the mixing ratio through the Bayesian neural network;

[0010] Step 4: Uncertainty quantification. Use Monte Carlo simulation through random sampling and probability statistics to transfer the uncertainty of the input parameters to the calculation result of the mixing ratio, generate the probability distribution of the mixing ratio, and output the 95% confidence interval of the mixing ratio.

[0011] Further improvement lies in: The end-member water sources in Step 1 include seawater, local precipitation, and deep fissure water.

[0012] Further improvement lies in: The preparation method of the eluent in Step 2 includes the following steps:

[0013] S1: After obtaining the mixed water sample, use an anion exchange resin to adsorb Cl - ;

[0014] S2: Add a nitric acid solution to the adsorbed water sample to elute SO4 2- , NO3 - ;

[0015] S3: Pour high-purity water into the water sample obtained in S2 to elute Cl - to obtain the eluent.

[0016] Further improvement lies in: The mass balance equation in Step 3 is:

[0017] ∑f i ·δ i =δ mix ,∑f i =1(f i ≥0)

[0018] The f i is the mixing ratio of each end member, and δ i is the end member isotope value.

[0019] A further improvement lies in that: the Bayesian-neural network coupling model in the step three includes an input layer, a hidden layer, and an output layer. The input layer is used to input the δ 37 Cl value and the Cl - concentration. The hidden layer is used to introduce Dropout regularization to prevent overfitting. The output layer is used to output the contribution probability distribution of each end member.

[0020] A further improvement lies in that: the sources of quantifying uncertainty in the step four include measurement error, natural variability of end member water samples, and limitations of model assumptions.

[0021] A further improvement lies in that: the Monte Carlo output in the step four is a probability distribution, and the output result is combined with the hydrogeological background to judge the rationality.

[0022] The beneficial effects of the present invention are as follows: By using the natural difference of chlorine isotopes in different water sources, combining the principle of mass conservation and intelligent algorithms, the present invention quantitatively analyzes the contribution ratio of each end member water source in the mixed water, and makes full use of the characteristics of chlorine isotopes with small fractionation and significant end member characteristics to become the core tracer index, solving the problem that the traditional hydrochemical analysis method relies on the concentration ratio of ions such as Cl - and Na + , which is easily affected by evaporation and concentration, mineral dissolution or human pollution interference, and is prone to misjudging the water source contribution;

[0023] Through the closed-loop process of "characteristic library construction → intelligent modeling → dynamic optimization → uncertainty quantification", the geochemical characteristics of chlorine isotopes are combined with artificial intelligence technology to achieve high-precision analysis of the complex limestone water mixing process, providing a scientific tool for water resource management and environmental protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is the step flow chart of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0025] In order to deepen the understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. The embodiments are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0026] According toFigure 1 As shown in the figure, this embodiment provides an intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotopes, including the following steps:

[0027] Step 1: Multi-dimensional data collection. According to the hydrogeological conditions of the study area, multiple sampling points are set to capture spatial geological differences. By setting multiple sampling points, the spatial heterogeneity of geological structures is captured, avoiding data deviation of a single sampling point. Based on the distribution characteristics of tectonic fractures in the study area, a sampling point network is arranged along the main runoff direction of groundwater, covering three typical units: the hanging wall of the fault, the karst development area, and the coastal transition zone. Sampling is carried out separately during the wet season and the dry season to capture seasonal differences. An automated sampling device (AquaTroll 600 multi-parameter probe) is used for continuous annual hydrological monitoring. During the wet season (June - September), high-frequency sampling is implemented on a daily scale (three times at 08:00, 14:00, and 20:00 every day). During the dry season (December - March), sampling is carried out on a weekly scale. Sampling in the wet season and the dry season reflects the impact of seasonal hydrological changes on groundwater and enhances the temporal representativeness of the data. A spatial database containing geological structure characteristics is constructed to facilitate the integration and invocation of multi-source data;

[0028] The end-member water sources include seawater, local precipitation, and deep fissure water. Seawater end-member samples of seawater profiles are collected during the spring tides every month. Precipitation end-member event precipitation is collected by setting up gradient altitude rain gauges. Fissure water end-member primary fissure water is obtained by drilling deep wells in the outcropping area of basement rocks.

[0029] Step 2: Analysis of mixed water samples. Use a high-precision mass spectrometer to measure the δ 37 Cl value of the mixed water sample. After separating the eluate from the water sample using ion exchange chromatography, detect the purity of Cl- in the eluate. Use a conductivity meter to measure the conductivity and salinity of the water sample. The combination of the mass spectrometer and ion chromatography ensures the accuracy of the δ 37 Cl and Cl- purity data, reducing cross-ion interference. Utilize the characteristics that δ 37 Cl has a small fractionation effect in natural water bodies (usually <1‰), and different water sources (such as seawater δ 37 Cl≈0‰, precipitation δ 37 Cl is negatively biased, and deep water δ 37 Cl is positively biased) have significant characteristic differences, making δ 37 Cl convenient as an ideal tracer, solving the problem that traditional hydrochemical methods are easily interfered by evaporation, mineral dissolution, or human pollution and cannot accurately identify the mixing ratio of multi-source water;

[0030] The preparation method of the eluate includes the following steps:

[0031] S1: After obtaining the mixed water sample, use an anion exchange resin to adsorb Cl in the water sample -, Dowex 1×8 anion resin (200 - 400 mesh) is used, and the dynamic adsorption capacity is controlled at 2.1 meq / mL;

[0032] S2: Add nitric acid solution to the adsorbed water sample to elute SO4 2- , NO3 - , and the nitric acid eluent (0.5 M) is controlled at 40 °C for circulating leaching, with a flow rate of 0.8 mL / min;

[0033] S3: Pour high-purity water into the water sample obtained in S2 to elute Cl- to obtain the eluent.

[0034] Step 3: Construction of the mixing model. By using the conservation of chlorine isotope and chloride ion concentration to establish equations to obtain the mass balance equation, ensuring that the model conforms to the law of isotope conservation, a Bayesian-neural network coupling model is constructed. The mixing ratio is calculated through the Bayesian neural network, and the complex nonlinear relationship is learned through data to improve the prediction accuracy;

[0035] The mass balance equation is:

[0036] ∑f i ·δ i =δ mix , ∑f i =1 (f i ≥0)

[0037] f i is the mixing ratio of each endmember, and δ i is the endmember isotope value;

[0038] The Bayesian-neural network coupling model includes an input layer, a hidden layer, and an output layer. The input layer is used to input the δ 37 Cl value and the Cl - concentration. The hidden layer is used to introduce Dropout regularization to prevent overfitting. The output layer is used to output the contribution probability distribution of each endmember.

[0039] Step 4: Uncertainty quantification. By using Monte Carlo simulation through random sampling and probability statistics, the uncertainty of the input parameters is transferred to the calculation result of the mixing ratio, generating the probability distribution of the mixing ratio, and outputting the 95% confidence interval of the mixing ratio;

[0040] The sources of quantified uncertainty include measurement error, natural variability of the endmember water samples, and limitations of model assumptions;

[0041] The Monte Carlo output is a probability distribution. The output results are combined with the hydrogeological background to judge the rationality. The probability distribution rather than point estimation is output to help users understand the confidence level of the results (such as the 95% confidence interval). The transfer effect of parameters on the results is quantified through random sampling to reveal the potential deviation range, which is applicable to high-risk scenarios such as resource management or pollution source tracing.

[0042] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotopes, comprising the following steps: Step 1: Multi-dimensional data collection. Set multiple sampling points according to the hydrogeological conditions of the study area to capture spatial geological differences, and sample respectively during the wet season and the dry season to capture seasonal differences, and construct a spatial database containing geological structure characteristics; Step 2: Analyze the mixed water sample. Use a high-precision mass spectrometer to measure the δ 37 Cl value of the mixed water sample. After separating the eluate from the water sample using ion exchange chromatography, detect the Cl - purity in the eluate. Use a conductivity meter to measure the conductivity and salinity in the water sample; Step 3: Mixed model construction. Establish an equation through the conservation of chlorine isotopes and chloride ion concentration to obtain a mass balance equation, use the Bayesian-neural network coupling to construct a model, and calculate the mixing ratio through the Bayesian neural network; Step 4: Uncertainty quantification. Use Monte Carlo simulation through random sampling and probability statistics to transfer the uncertainty of the input parameters to the calculation result of the mixing ratio, generate the probability distribution of the mixing ratio, and output the 95% confidence interval of the mixing ratio.

2. The intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotopes according to claim 1, wherein: The end-member water sources in Step 1 include seawater, local precipitation, and deep fissure water.

3. The intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotopes according to claim 1, characterized in that: The preparation method of the eluent in Step 2 includes the following steps: S1: After obtaining the mixed water sample, use an anion exchange resin to adsorb Cl in the water sample - ; S2: Add a nitric acid solution to the adsorbed water sample to elute SO4 2- , NO3 - ; S3: Pour high-purity water into the water sample obtained in S2 to elute Cl - An eluate is obtained.

4. The intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotopes according to claim 1, characterized in that: The mass balance equation in Step 3 is: ∑f i ·δ i =δ mix ,∑f i =1(f i ≥0) The f i is the mixing ratio of each endmember, and δ i is the endmember isotope value.

5. The intelligent analysis method for the tracing ratio of mixed limestone water based on chlorine isotope according to claim 1, characterized in that: The Bayesian-neural network coupling model in the third step includes an input layer, a hidden layer, and an output layer. The input layer is used to input the δ 37 Cl value and Cl- concentration. The hidden layer is used to introduce Dropout regularization to prevent overfitting. The output layer is used to output the contribution probability distribution of each endmember.

6. The intelligent analysis method for tracing the proportion of mixed limestone water based on chlorine isotopes according to claim 1, wherein: The sources of uncertainty quantification in Step 4 include measurement errors, natural variability of end-member water samples, and limitations of model assumptions.

7. An intelligent analysis method for tracing the proportion of mixed limestone water based on chlorine isotopes according to claim 1, characterized in that: The Monte Carlo output in Step 4 is a probability distribution, and the rationality of the output result is judged in combination with the hydrogeological background.

Images