A method for tracing the pollution source of groundwater in the sea-land ecotone based on multi-pollutant fingerprints

Abstract

The application discloses a kind of based on multi-pollutant fingerprint sea-land interaction zone groundwater pollution source tracing method, by constructing the fingerprint characteristic library including 8 key pollutants of cadmium, arsenic, nitrate nitrogen, ammonium nitrogen, phosphate, total petroleum hydrocarbon, sulfamethoxazole, pH value etc., establishes the statistical characteristic model of four types of pollution sources of industrial emission, agricultural non-point source, oil exploitation, aquaculture, adopts Z-Score standardization and similarity calculation method based on standard normal distribution probability density function, combined with risk weight distribution mechanism carries out comprehensive similarity evaluation, sets 0.7 similarity threshold value to carry out pollution source identification and ordering, consider sea-land interaction zone tide, season, geological etc. Environmental factors are modified.The application can quickly and accurately identify sea-land interaction zone groundwater pollution source, and the tracing accuracy is more than 80%, to provide scientific and technical support for sea-land interaction zone groundwater pollution prevention and control and environmental protection policy making.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and pollution source tracing technology, and in particular to a method for tracing groundwater pollution sources in the marine-terrestrial interaction zone based on multi-pollutant fingerprints. Background Technology

[0002] The land-sea transition zone, as a transitional area between land and sea, possesses unique hydrogeological conditions and complex pollution source distribution characteristics. This region bears dual environmental pressures from both land and sea, including the combined effects of multiple pollution sources such as industrial wastewater discharge, agricultural non-point source pollution, oil extraction activities, and aquaculture wastewater, leading to increasingly serious groundwater pollution problems. Traditional groundwater pollution source identification methods mainly rely on single pollutant concentration analysis, isotope tracing, or simple multivariate statistical analysis. These methods have significant limitations in the complex environment of the land-sea transition zone: first, the accuracy of single-indicator identification is insufficient, making it difficult to accurately distinguish different pollution sources; second, there is a lack of adaptive algorithms for the special environment of the land-sea transition zone; third, the ability to identify multi-source pollution is limited; and fourth, the real-time response capability is poor. Existing fingerprinting methods are mostly designed for general environmental conditions and lack consideration for the special environmental factors of the land-sea transition zone, such as the impact of tidal forces, seawater intrusion, and seasonal changes on pollutant concentration and distribution. Therefore, there is a need to develop high-precision, rapid-response groundwater pollution source tracing technologies specifically suitable for the land-sea transition zone environment. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for tracing groundwater pollution sources in the marine-terrestrial interaction zone based on multi-pollutant fingerprints. This method can quickly and accurately identify groundwater pollution sources in complex marine-terrestrial interaction zones, providing a scientific basis for pollution prevention and control and environmental management.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A method for tracing groundwater pollution sources in the land-sea interaction zone based on multi-pollutant fingerprints is proposed. This method constructs a multi-dimensional pollutant fingerprint feature database, adopts an improved similarity calculation algorithm, and combines environmental feature correction of the land-sea interaction zone to achieve accurate identification and tracing of pollution sources.

[0006] The method includes the following steps:

[0007] Step S1: Construct a multi-pollutant fingerprint feature database. Based on the environmental characteristics of the land-sea transition zone and the main pollution source types, eight representative pollutants are selected as fingerprint features, including heavy metals cadmium and arsenic, nutrients nitrate nitrogen, ammonium nitrogen and phosphate, organic pollutants total petroleum hydrocarbons, antibiotics sulfamethoxazole, and the physical parameter pH value. Based on the main human activities in the land-sea transition zone, a classification system for four types of pollution sources is established: industrial emissions, agricultural non-point sources, oil extraction, and aquaculture. A normal distribution statistical model is established for each pollutant in each type of pollution source, and the mean μ is extracted. ij and standard deviation σ ij As fingerprint feature parameters, i represents the pollution source type and j represents the pollutant type.

[0008] Step S2: Pollutant Concentration Standardization. An improved Z-Score standardization method was used to process the measured pollutant concentrations, introducing a correction coefficient for environmental factors in the land-sea transition zone to eliminate dimensional and numerical range differences between different pollutants, providing standardized input for subsequent similarity calculations.

[0009] The calculation formula is:

[0010] Z ij =(C j -μ ij ) / σ ij

[0011] Among them, Z ij C is the standardized value of pollutant j relative to pollution source i. j : Measured concentration of pollutant j at monitoring point, in μ ij σ is the theoretical mean value of pollutant j in pollution source i. ij Theoretical standard deviation of pollutant j in pollution source i;

[0012] Introduce an environmental correction mechanism for the land-sea interaction zone.

[0013] Step S3: Multidimensional similarity calculation. Based on the Gaussian probability density function, the similarity between each pollutant and the characteristics of different pollution sources is calculated. Considering the environmental risk and toxicity characteristics of pollutants, a risk weight allocation mechanism is established, and a weighted average method is used to calculate the comprehensive similarity, achieving effective fusion of information from multiple pollutants.

[0014] Step S3 includes:

[0015] S3.1 Calculates single-pollutant similarity based on the Gaussian probability density function, using the formula: S ij =exp(-Z ij 2 / 2);

[0016] S3.2 Establish a risk weight allocation mechanism: Determine the risk weight based on the degree of pollutant exceedance and toxicity coefficient, using the following formula:

[0017] W j =log(R) j +1) / log(11), where R j The overall risk score for pollutant j;

[0018] S3.3 uses a weighted average method to calculate the overall similarity, with the formula: S i =Σ(S ij ×W j ) / Σ(W j ).

[0019] Step S4: Pollution Source Identification and Ranking. A dynamic similarity threshold setting mechanism is established, adjusting the judgment threshold based on environmental background conditions and pollution levels to accurately identify potential pollution sources. When multiple highly similar pollution sources exist, they are sorted by similarity and identified as multi-source composite pollution, outputting a list of major pollution sources, a list of potential pollution sources, and contribution assessment results.

[0020] Step S4 includes:

[0021] S4.1 sets a dynamic similarity threshold, with a baseline threshold of 0.7, which is reduced to 0.6 when the contamination is severe and increased to 0.8 when the background is clean;

[0022] S4.2 Pollution source identification and sorting: When the similarity between a monitoring point and a certain type of pollution source exceeds a set threshold, it is determined to be a possible pollution source; when there are multiple highly similar pollution sources, they are sorted according to the similarity, with the one with the highest similarity being the primary pollution source and the others being secondary pollution sources.

[0023] S4.3 outputs the source tracing results, including the main pollution sources, a list of possible pollution sources, the contribution of each pollution source, and a confidence level assessment.

[0024] The aforementioned environmental correction for the land-sea transition zone. In response to the unique environmental conditions of the land-sea transition zone, mechanisms for tidal influence correction, seasonal correction, and geological condition correction are established to improve the accuracy and reliability of pollution source identification.

[0025] First, tidal effect correction is performed: using formula C' j =C j ×(1+α j ×f(tide)), where f(tide) is the tidal correction factor;

[0026] Next, seasonal adjustments are made: using the adjustment formula C” j =C' j ×β j (season), βj (season) is the seasonality correction factor;

[0027] Finally, geological conditions are corrected using the correction formula C”'. j =C” j ×γ j (geology), γ j (geology) is the geological correction factor;

[0028] The corrected concentrations were standardized and similarity was calculated, resulting in a significant improvement in similarity.

[0029] Furthermore, the method also includes an uncertainty assessment step, which uses Monte Carlo simulation to analyze the confidence interval of the pollution source identification results and comprehensively assess the reliability of the source tracing results.

[0030] The beneficial effects of this invention are as follows:

[0031] 1. Significantly improved identification accuracy: By constructing a multi-dimensional fingerprint feature library containing eight key pollutants and combining it with an improved similarity calculation algorithm, the accuracy of pollution source identification can reach more than 80%, which is more than 30% higher than the traditional single-indicator method.

[0032] 2. Strong environmental adaptability: The correction mechanism, specifically designed for the marine-land interaction zone environment, effectively solves the interference of special environmental factors such as tides and seasonal changes on the source tracing results, and significantly improves the environmental adaptability of the method.

[0033] 3. Outstanding ability to identify multi-source pollution: The innovative dynamic threshold setting and weighted similarity calculation method can effectively identify and distinguish multi-source compound pollution, providing a scientific basis for the treatment of complex pollution situations.

[0034] 4. High computational efficiency: The algorithm has a complexity of O(n×m) and can achieve a response time in seconds, meeting the needs of real-time monitoring and rapid emergency response.

[0035] 5. Good scalability: The fingerprint feature library and algorithm framework have good scalability, and new pollutant indicators or pollution source types can be added according to actual needs to adapt to the ever-changing environmental monitoring requirements. Attached Figure Description

[0036] Figure 1 This is a flowchart of the groundwater pollution source tracing method based on multi-pollutant fingerprinting in the marine-terrestrial interaction zone according to the present invention;

[0037] Figure 2 This is a schematic diagram of the multi-pollutant fingerprint feature library construction method of the present invention;

[0038] Figure 3This is a schematic diagram illustrating the principle of multidimensional similarity calculation in this invention.

[0039] Figure 4 This is a schematic diagram of the marine-terrestrial interaction zone environmental correction mechanism of the present invention;

[0040] Figure 5 This is an example diagram of the output format of the pollution source identification results of the present invention;

[0041] Figure 6 This is a comparison chart of the recognition accuracy between the method of this invention and the traditional method;

[0042] Figure 7 This is a flowchart of the overall process of the method of the present invention. Detailed Implementation

[0043] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] This invention provides a method for tracing groundwater pollution sources in the land-sea transition zone based on multi-pollutant fingerprints. This method achieves rapid and accurate identification of groundwater pollution sources in the land-sea transition zone by establishing a multi-pollutant fingerprint feature database and an innovative similarity calculation algorithm.

[0045] like Figure 1 , 7 As shown, the present invention includes the following specific implementation steps:

[0046] Step S1: Construct a multi-contaminant fingerprint feature library, such as... Figure 2 As shown

[0047] S1.1 Based on the environmental characteristics and pollution source types of the land-sea transition zone, eight representative pollutants were selected as fingerprint characteristics. For heavy metal pollutants, cadmium and arsenic were chosen; these two pollutants mainly originate from industrial activities and have strong environmental persistence and biotoxicity. For nutrient pollutants, nitrate nitrogen, ammonium nitrogen, and phosphate were chosen; these mainly originate from agricultural activities and domestic sewage and are key factors leading to eutrophication. For organic pollutants, total petroleum hydrocarbons were chosen, mainly from leaks during oil extraction, transportation, and storage. For antibiotic pollutants, sulfamethoxazole was chosen, mainly from drug use in aquaculture. pH value was selected as the physical parameter, reflecting the impact of different pollution sources on the acidity and alkalinity of the water body.

[0048] S1.2 Establish a classification system for four major pollution sources. The characteristic pollutant combination of industrial emission sources mainly includes heavy metals and certain organic pollutants, usually accompanied by abnormal pH changes. Agricultural non-point source pollution is characterized by a significant increase in nutrient concentrations, especially nitrogen and phosphorus pollutants. The fingerprint characteristic of oil extraction activities is mainly a substantial increase in total petroleum hydrocarbon concentrations. The pollution characteristics of aquaculture activities mainly include combined pollution from nutrients and antibiotics.

[0049] S1.3 Establishes a normal distribution statistical model for each type of pollution source. By collecting historical monitoring data of typical pollution sources in the land-sea transition zone, the maximum likelihood estimation method is used to determine the theoretical distribution parameters of each pollutant in each pollution source, including the mean μ. ij and standard deviation σ ij .

[0050] Step S2: Standardization of pollutant concentration, such as... Figure 3 As shown

[0051] An improved Z-Score normalization method was used to process the measured pollutant concentrations. The normalization formula is:

[0052] Z ij =(C j -μ ij ) / σ ij

[0053] Among them, Z ij C is the standardized value of pollutant j relative to pollution source i. j : Measured concentration of pollutant j at monitoring point, in μ ij σ is the theoretical mean value of pollutant j in pollution source i. ij : The theoretical standard deviation of pollutant j in pollution source i. The improvement lies in introducing a correction coefficient for environmental factors in the land-sea interaction zone, taking into account the influence of factors such as tidal action and seawater dilution on pollutant concentration.

[0054] Step S3: Multidimensional Similarity Calculation

[0055] S3.1 Calculates the similarity of a single pollutant based on the Gaussian probability density function. The calculation formula is as follows:

[0056] S ij =exp(-Z ij 2 / 2)

[0057] This formula reflects the degree of deviation between the measured concentration and the theoretical distribution center; the smaller the deviation, the higher the similarity.

[0058] S3.2 Establish a risk weight allocation mechanism: Determine the weights based on the environmental risk level and toxicity coefficient of the pollutants. The calculation formula is as follows:

[0059] W j =log(R) j +1) / log(11)

[0060] Where R j This represents the overall risk score for pollutant j, ranging from 0 to 10. The risk score comprehensively considers factors such as the pollutant's toxicity, persistence, bioaccumulation, and environmental concentration.

[0061] S3.3 Calculation of comprehensive similarity: The weighted average method is used, and the formula is as follows:

[0062] S i =Σ(S ij ×W j ) / Σ(W j )

[0063] To achieve effective integration of information on multiple pollutants.

[0064] Step S4: Pollution Source Identification and Sequencing

[0065] S4.1 Establish a dynamic similarity threshold setting mechanism: The baseline threshold is set at 0.7, which is lowered to 0.6 when environmental pollution is severe to improve identification sensitivity, and raised to 0.8 when the background environment is clean to reduce false positives. The threshold adjustment is based on the comprehensive environmental quality evaluation index and statistical analysis of historical monitoring data.

[0066] S4.2 Pollution source identification and ranking: When the similarity between a monitoring point and a certain type of pollution source exceeds a set threshold, it is determined to be a possible pollution source. When there are multiple highly similar pollution sources, they are ranked according to the degree of similarity, with the one with the highest similarity being the primary pollution source and the others being secondary pollution sources.

[0067] S4.3 Output source tracing results: The results include information such as the main pollution source types, similarity values, confidence assessment, list of possible pollution sources, and contribution analysis of each pollution source.

[0068] Step S5: Environmental correction of the land-sea interaction zone, such as... Figure 4 As shown

[0069] S5.1 Implement tidal impact correction: Based on the geographical location of the monitoring point, tidal cycle, and degree of seawater intrusion, correct the concentration of salt-sensitive pollutants. The correction factor is related to tidal range, distance from the sea, and groundwater salinity concentration.

[0070] S5.2 Seasonal Correction: Based on years of historical monitoring data, establish seasonal variation patterns for the concentrations of various pollutants, taking into account the influence of seasonal factors such as rainfall, temperature, and biological activity, and make corresponding adjustments to the identification results.

[0071] S5.3 Consideration of geological condition correction: Based on the unique geological structure of the marine-continental interaction zone, including multi-layered aquifer systems, groundwater flow direction, and permeability changes, analyze the migration and transformation patterns of pollutants and make geological condition-related corrections to the source tracing results.

[0072] Uncertainty assessment

[0073] Monte Carlo simulation was used to assess the uncertainty of the source tracing results. Considering factors such as sampling error, experimental analysis error, and model parameter uncertainty, confidence intervals were calculated through multiple random simulations to provide risk assessment information for decision-making.

[0074] Example 1: Identification of Industrial Emission Sources

[0075] The following pollutant concentrations were detected at monitoring point A in the land-sea transition zone: cadmium concentration was 4.2 μg / L, arsenic concentration was 85 μg / L, total petroleum hydrocarbon concentration was 750 μg / L, pH value was 6.8, and the concentrations of other pollutants were all within the background range.

[0076] First, standardization is performed: taking industrial emission sources as an example, the theoretical mean of cadmium is 4.0 μg / L, and the standard deviation is 1.0 μg / L. Z is calculated as follows: Cd = (4.2-4.0) / 1.0 = 0.2. The theoretical mean of arsenic is 80.0 μg / L, and the standard deviation is 20.0 μg / L. Z is calculated as follows. As = (85-80) / 20 = 0.25. The theoretical mean of total petroleum hydrocarbons is 800.0 μg / L, and the standard deviation is 200.0 μg / L. Z is calculated as follows: TPH = (750-800) / 200 = -0.25.

[0077] Then calculate the similarity: the similarity S of cadmium. Cd =exp(-0.2) 2 / 2)=0.98, the similarity S of arsenic As =exp(-0.25) 2 / 2)=0.97, the similarity S of total petroleum hydrocarbons TPH =exp(-(-0.25) 2 / 2)=0.97.

[0078] Next, the weights are determined: based on the environmental risk assessment, the weight W for cadmium is... Cd =0.85, weight W of arsenic As =0.78, the weight W of total petroleum hydrocarbons TPH =0.72.

[0079] Finally, calculate the overall similarity:

[0080] S industrial = (0.98×0.85+0.97×0.78+0.97×0.72) / (0.85+0.78+0.72)=0.974.

[0081] Since the overall similarity of 0.91 is greater than the threshold of 0.7, the monitoring point is determined to be polluted by industrial emission sources with a confidence level of 97.4%.

[0082] Example 2: Identification of Multi-Source Complex Pollution

[0083] At a monitoring point B in the land-sea transition zone, pollutants such as ammonium nitrogen (2.1 mg / L), total petroleum hydrocarbons (680 μg / L), and sulfamethoxazole (320 ng / L) were detected simultaneously.

[0084] Similarity calculations showed that the monitoring point had a similarity of 0.76 with agricultural non-point source pollution, 0.73 with oil extraction activities, and 0.71 with aquaculture activities, all of which exceeded the threshold of 0.7.

[0085] Therefore, the pollution was determined to be multi-source and complex, with agricultural non-point source pollution as the main source (similarity 0.76), and secondary sources including oil extraction activities (similarity 0.73) and aquaculture activities (similarity 0.71). The contribution rates of each pollution source were 34.5%, 33.2%, and 32.3%, respectively.

[0086] Example 3: Application of Environmental Correction in the Land-Sea Interchange Zone

[0087] Monitoring point C is located in the tidal influence area, 2.5 kilometers from the coastline. During spring tides, the pollutant concentrations detected were: cadmium 1.8 μg / L, arsenic 45 μg / L, nitrate nitrogen 12.5 mg / L, ammonium nitrogen 3.2 mg / L, phosphate 0.8 mg / L, total petroleum hydrocarbons 320 μg / L, sulfamethoxazole 85 ng / L, and pH 7.2. Due to the dilution effect of seawater, the pollutant concentrations were generally low.

[0088] according to Figure 4 The environmental correction mechanism for the land-sea interaction zone shown first corrects for tidal effects. The current tidal range is 3.2 meters, indicating a moderate degree of seawater intrusion; therefore, the correction formula C' is used. j =C j ×(1+α j ×f(tide)), where f(tide) = 0.164. Correction factor α for salt-sensitive heavy metal pollutants. j =0.6, nutrient α j =0.8, organic α j =0.4, pH value α j =0.2. The corrected pollutant concentrations are as follows: cadmium 1.98 μg / L, arsenic 49.4 μg / L, nitrate nitrogen 14.1 mg / L, ammonium nitrogen 3.6 mg / L, phosphate 0.9 mg / L, total petroleum hydrocarbons 341 μg / L, sulfamethoxazole 91 ng / L, and pH 7.24.

[0089] Next, seasonal adjustments are made. The monitoring period is July, with a rainfall of 180 mm and a monthly average temperature of 28℃. The adjustment formula C” is used. j =C'j ×β j (season), nutrient β was calculated j =1.066, β of heavy metals j =1.022, β of organic compounds j =1.044, pH value β j =1.011. The corrected pollutant concentrations are as follows: cadmium 2.02 μg / L, arsenic 50.5 μg / L, nitrate nitrogen 15.0 mg / L, ammonium nitrogen 3.8 mg / L, phosphate 0.96 mg / L, total petroleum hydrocarbons 356 μg / L, sulfamethoxazole 95 ng / L, and pH 7.32.

[0090] Finally, geological conditions were corrected. The monitoring point is located in a multi-layered confined aquifer region with a permeability coefficient K = 5 × 10⁻⁶. - 4 cm / s, groundwater flows northeast. Using the corrected formula C”' j =C” j ×γ j (geology), determining heavy metal gammas based on pollutant migration characteristics. j =1.15, γ-nutrients j =0.95, organic γ j =1.05, pH value γ j =1.02. The final corrected pollutant concentrations are: cadmium 2.32 μg / L, arsenic 58.1 μg / L, nitrate nitrogen 14.3 mg / L, ammonium nitrogen 3.6 mg / L, phosphate 0.91 mg / L, total petroleum hydrocarbons 374 μg / L, sulfamethoxazole 100 ng / L, and pH 7.47.

[0091] After standardizing the concentrations and calculating similarity, the similarity of each pollution source before correction was below the threshold of 0.7: industrial emissions 0.64, agricultural non-point sources 0.58, and oil extraction 0.62, making it impossible to accurately identify the pollution sources. After correction, the similarity of pollution sources was significantly improved: industrial emissions 0.78, agricultural non-point sources 0.72, and oil extraction 0.69.

[0092] like Figure 7 The source tracing results show that the monitoring point has multiple sources of pollution. The main source is industrial emissions (similarity 0.78, contribution 35.7%), the secondary source is agricultural non-point source pollution (similarity 0.72, contribution 32.9%), and the possible source is oil extraction activities (similarity 0.69, contribution 31.4%). Uncertainty assessment using Monte Carlo simulation yielded identification confidence levels of 78%, 72%, and 69% for each pollution source, respectively.

[0093] Figure 5This is an example diagram of the output format of the pollution source identification results of the present invention, showing the pollution source feature fingerprint comparison analysis and pollution source similarity.

[0094] Figure 6 The recognition accuracy of the method of the present invention is shown to be more than 30% higher than that of the traditional single index method.

[0095] In summary, the source tracing method for groundwater pollution in the marine-terrestrial interaction zone based on multi-pollutant fingerprints provided by this invention has the advantages of high identification accuracy, strong environmental adaptability, and high computational efficiency, providing important technical support for the prevention and control of groundwater pollution in the marine-terrestrial interaction zone.

[0096] This invention is not limited to the above embodiments. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the protection scope of this invention.

Claims

1. A method for tracing groundwater pollution sources in the land-sea transition zone based on multi-pollutant fingerprinting, characterized in that, Includes the following steps: S1 constructs a multi-pollutant fingerprint feature library, selecting cadmium, arsenic, nitrate nitrogen, ammonium nitrogen, phosphate, total petroleum hydrocarbons, sulfamethoxazole, and pH value as pollutant fingerprint features, and establishes a statistical feature model of pollution sources; S2 standardizes the measured pollutant concentrations; S3 calculates multidimensional similarity; S4 is used for pollution source identification and sorting; S5 performs environmental correction for the land-sea interaction zone; In step S2, an improved Z-Score standardization method is used, and the calculation formula is as follows: Z ij =(C j -m ij ) / s ij Among them, Z ij C is the standardized value of pollutant j relative to pollution source i. j : Measured concentration of pollutant j at monitoring point, μ ij σ is the theoretical mean value of pollutant j in pollution source i. ij : The theoretical standard deviation of pollutant j in pollution source i, where i represents the type of pollution source and j represents the type of pollutant; Introducing an environmental correction mechanism for the land-sea interaction zone; Step S3 includes: S3.1 Calculates single-pollutant similarity based on the Gaussian probability density function, using the formula: S ij =exp(-Z ij 2 / 2); S3.2 Establish a risk weight allocation mechanism: Determine the risk weight based on the degree of pollutant exceedance and toxicity coefficient, using the formula: W j =log(R j +1) / log(11), where R j The overall risk score for pollutant j; S3.3 uses a weighted average method to calculate the overall similarity, with the formula: S i =Σ(S ij ×W j ) / Σ(W j ); The proposed environmental correction mechanism for the land-sea interaction zone: S5.1 Tidal Impact Correction: The concentration of salt-sensitive pollutants is corrected based on the geographical location of the monitoring point, the tidal cycle, and the degree of seawater intrusion. S5.2 Seasonal Correction: A seasonal correction factor is established based on historical monitoring data, taking into account the effects of seasonal variations in rainfall, temperature, and biological activity. S5.3 Geological conditions correction, taking into account the influence of the multi-layered aquifer structure in the marine-continental transition zone, groundwater flow direction and pollutant migration and transformation patterns; First, perform tidal effect correction: Formula C ' j =C j ×(1+α j ×f(tide)), where f(tide) is the tidal correction factor, and α j Correction factor for heavy metal pollutants sensitive to salt content; Next, seasonal adjustments will be made: Using the corrected formula C '' j =C ' j ×β j (season), β j (season) is the seasonality correction factor; Finally, geological conditions were corrected: Using the corrected formula C ''' j =C '' j ×γ j (geology), γ j (geology) is the geological correction factor; The corrected concentrations were standardized and similarity was calculated, resulting in a significant improvement in similarity.

2. The method according to claim 1, characterized in that, Step S1 specifically includes: S1.1 Pollutant indicators are selected based on the environmental characteristics of the land-sea transition zone. Heavy metal pollutants include cadmium and arsenic, nutrient pollutants include nitrate nitrogen, ammonium nitrogen and phosphate, organic pollutants include total petroleum hydrocarbons, antibiotic pollutants include sulfamethoxazole, and physical parameters include pH value. S1.2 defines the classification of pollution sources, including four categories of pollution sources: industrial emissions, agricultural non-point sources, oil extraction, and aquaculture. S1.3 Establish a normal distribution statistical model for each pollutant of each type of pollution source and extract the mean μ. ij and standard deviation σ ij As fingerprint feature parameters.

3. The method according to claim 1, characterized in that, Step S4 includes: S4.1 sets a dynamic similarity threshold, with a baseline threshold of 0.7, which is reduced to 0.6 when the contamination is severe and increased to 0.8 when the background is clean; S4.2 Pollution source identification and sorting: When the similarity between a monitoring point and a certain type of pollution source exceeds a set threshold, it is determined to be a possible pollution source; when there are multiple highly similar pollution sources, they are sorted according to the similarity, with the one with the highest similarity being the primary pollution source and the others being secondary pollution sources. S4.3 outputs the source tracing results, including the main pollution sources, a list of possible pollution sources, the contribution of each pollution source, and a confidence level assessment.

4. The method according to claim 1, characterized in that, The method also includes an uncertainty assessment step, which uses Monte Carlo simulation to analyze the confidence interval of the pollution source identification results, taking into account the impact of sampling error, analysis error and model uncertainty on the source tracing results.

Images