An ozone pollution hierarchical cause analysis method and system based on double-index joint threshold division
By combining dual-indicator threshold classification and multiple linear regression models with WRF-Chem and HYSPLIT models, the problems of misjudgment of causes and waste of governance resources in ozone pollution analysis were solved, and accurate subdivision and real-time early warning of ozone pollution types were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA NAT ENVIRONMENTAL MONITORING CENT
- Filing Date
- 2026-03-03
- Publication Date
- 2026-06-05
AI Technical Summary
Existing ozone pollution analysis technologies rely on a single indicator for judgment, leading to misjudgment of causes, one-size-fits-all treatment measures, waste of resources, and inability to meet the needs of real-time early warning.
A dual-indicator joint threshold method is adopted to obtain core indicators and auxiliary data. After data preprocessing, critical, low and high value exceedance thresholds are set, and multiple linear regression models and high value exceedance models are constructed. Combined with WRF-Chem and HYSPLIT models, hierarchical causal analysis is carried out to achieve full-process automated early warning.
It enables precise segmentation of ozone pollution types, improves decision-making efficiency, lowers decision-making threshold, optimizes the allocation of governance resources, reduces the number of polluted days, and meets the needs of real-time early warning.
Smart Images
Figure CN122157876A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing, and more specifically, to a method and system for stratifying the causes of ozone pollution based on a dual-index joint threshold division. Background Technology
[0002] Current ozone pollution analysis technologies mainly rely on the single-indicator judgment logic in the "Technical Regulations for Ambient Air Quality Index (AQI) (Trial)" and the "Ambient Air Quality Standard" (GB3095-2012), with monitoring indicators only... or The system directly determines whether a pollution level is exceeded by comparing the indicator value with the secondary standard threshold. If it exceeds the standard, it only outputs "ozone is the primary pollutant," without considering the combined state of the indicators for stratified analysis. It relies on manual experience to judge precursors or meteorological influences. This limitation of using a single indicator makes it impossible to distinguish different types of pollution, easily leading to misjudgments of causes. The lack of scientific stratification results in a "one-size-fits-all" approach to pollution control, causing resource waste. Furthermore, it leads to fragmented analysis of causes, making it impossible to quantify the contribution of each factor, resulting in a lack of data support for decision-making. Moreover, manual analysis is time-consuming and cannot meet the needs of real-time early warning.
[0003] Therefore, how to provide a method that can accurately classify the types of exceeding the standard and improve decision-making efficiency has become an urgent problem to be solved in this field. Summary of the Invention
[0004] To address the aforementioned issues, this application proposes a method for stratified causal analysis of ozone pollution based on dual-indicator joint threshold division. The method involves data acquisition and preprocessing; dual-indicator joint threshold division based on the preprocessed data; establishment of a stratified causal analysis model based on the results of the dual-indicator joint threshold division; and early warning mapping based on the output of the stratified causal analysis model.
[0005] The above-described method for stratifying ozone pollution causes based on a dual-indicator joint threshold involves data acquisition, including acquiring core indicators and auxiliary data. The core indicators include obtaining the daily maximum 8-hour moving average concentration of ozone. The daily maximum hourly average concentration of ozone The auxiliary data includes meteorological elements and precursors.
[0006] The above-described method for stratifying the causes of ozone pollution based on dual-indicator joint thresholds includes, according to the preprocessed data, setting dual-indicator joint thresholds as follows: setting critical exceedances; setting low exceedances; and setting high exceedances.
[0007] The above-described method for stratifying the causes of ozone pollution based on dual-indicator joint threshold classification includes the following sub-steps for establishing a stratified cause analysis model based on the results of dual-indicator joint threshold classification: for critical exceedances, construct a critical exceedance model; for low exceedances, construct a low exceedance model; and for high exceedances, construct a high exceedance model.
[0008] The above-described method for stratifying ozone pollution causes based on a dual-indicator joint threshold classification includes the following sub-steps for constructing a critical exceedance model: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] The concentrations of olefin VOCs, temperature, and wind speed (WS) are used as input data; the expression for the multiple linear regression model is determined; and the contribution of each factor is output by the multiple linear regression model.
[0009] A system for stratified causal analysis of ozone pollution based on dual-indicator joint threshold division includes: an acquisition and processing unit, a dual-indicator joint threshold division unit, a stratified causal analysis model establishment unit, and an early warning mapping unit. The acquisition and processing unit is used for data acquisition and preprocessing; the dual-indicator joint threshold division unit is used for performing dual-indicator joint threshold division based on the preprocessed data; the stratified causal analysis model establishment unit is used for establishing a stratified causal analysis model based on the results of the dual-indicator joint threshold division; and the early warning mapping unit is used for performing early warning mapping based on the output results of the stratified causal analysis model.
[0010] The ozone pollution stratification and causal analysis system based on dual-indicator joint threshold classification, as described above, includes a data acquisition and processing unit that acquires core indicators and auxiliary data. The core indicators include the acquisition of the daily maximum 8-hour moving average concentration of ozone. The daily maximum hourly average concentration of ozone The auxiliary data includes meteorological elements and precursors.
[0011] The ozone pollution stratification and causal analysis system based on dual-indicator joint threshold division, as described above, includes a dual-indicator joint threshold division unit that performs dual-indicator joint threshold division based on preprocessed data, including: setting a critical exceedance; setting a low exceedance; and setting a high exceedance.
[0012] The ozone pollution stratification and causal analysis system based on dual-indicator joint threshold division, as described above, includes the following sub-steps in which the stratification and causal analysis model building unit establishes a stratification and causal analysis model based on the results of dual-indicator joint threshold division: for critical exceedances, a critical exceedance model is constructed; for low exceedances, a low exceedance model is constructed; and for high exceedances, a high exceedance model is constructed.
[0013] The ozone pollution stratification and causal analysis system based on dual-index joint threshold classification, as described above, includes the following sub-steps in the stratification and causal analysis model building unit for constructing a critical exceedance model: [The following text appears to be a separate, unrelated section and is not translated: "to..."] The concentrations of olefin VOCs, temperature, and wind speed (WS) are used as input data; the expression for the multiple linear regression model is determined; and the contribution of each factor is output by the multiple linear regression model.
[0014] This application has the following beneficial effects: This application overcomes the limitations of single-indicator judgment by using a dual-indicator joint threshold classification, achieving precise segmentation of exceedance types. Furthermore, the entire process is automated, with short response times, significantly improving decision-making efficiency. This application lowers the decision-making threshold through visualization design and effectively reduces the number of ozone pollution days and optimizes the allocation of governance resources through early warning mapping, demonstrating significant practical value and promotional significance. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.
[0016] Figure 1 This is a flowchart illustrating the ozone pollution stratification and causal analysis method based on dual-index joint threshold division provided in the embodiments of this application. Figure 2 This is a schematic diagram of the internal structure of an ozone pollution stratification and cause analysis system based on a dual-index joint threshold division, provided in an embodiment of this application. Detailed Implementation
[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.
[0018] This invention provides a method based on and A monitoring, early warning, and analysis method based on jointly defined exceedance ranges and stratified cause analysis. and The combined threshold is used to categorize ozone exceedances into three types: "critical," "low," and "high," covering all possible exceedance scenarios. For each exceedance type, a differentiated quantitative analysis model is designed to analyze precursors (…). The contribution of VOCs, meteorological conditions, and regional transmission is quantified; the entire process from data collection to early warning output is automated, with an early warning time of ≤15 minutes, providing environmental management departments with precise solutions of "one measure for each type of exceedance"; the causes are intuitively displayed through visual graphs (contribution bar chart, transmission path diagram), reducing the decision-making threshold.
[0019] Example 1 like Figure 1 As shown, this embodiment provides a method for stratifying the causes of ozone pollution based on a dual-index joint threshold, specifically including the following steps: Step S1: Data acquisition and preprocessing.
[0020] Data acquisition includes acquiring core metrics and auxiliary data.
[0021] The core indicators include the daily maximum 8-hour moving average concentration of ozone obtained in real time from the National Air Quality Monitoring Network API. ), the daily maximum hourly average concentration of ozone ( It is understandable that the core metrics obtained above are at the hourly level resolution.
[0022] The supporting data includes meteorological elements and precursors. Meteorological elements include temperature (T, °C), relative humidity (RH, %), wind speed (WS, m / s), wind direction (WD, °), and sunshine duration (H, h) (minute-level resolution, sourced from weather stations or satellite data); precursors include... (μg / m³), VOCs components (olefins, aromatics, alkanes, μg / m³) (hourly resolution, sourced from online monitoring equipment).
[0023] Preprocessing includes removing outliers and filling in missing values in the acquired data.
[0024] Outlier removal includes filtering data that exceeds a reasonable range using the 3σ principle (e.g., ...). >400μg / m³); Missing value imputation includes using linear interpolation to fill in missing data ≤2 hours in time, and marking data older than 2 hours as invalid samples.
[0025] Step S2: Based on the preprocessed data, perform a joint threshold division using two indicators.
[0026] Based on the GB3095-2012 Class II standard and statistical data from 31 provincial monitoring stations across the country during the high-temperature season (June-August) from 2018 to 2022 (a total of 1.2 million samples), the dual-indicator joint threshold division was performed, including setting the following thresholds: 1. 160 < ≤165 and <200 is defined as the critical exceedance. Statistical data shows that the precursor concentration of 80% of the samples in this range is close to the threshold, and the meteorological conditions are slightly unfavorable (T≥30℃, WS≤2m / s).
[0027] 2. >165 and <200 is considered a low value exceeding the limit. The cumulative effect of precursors is significant within this range. (>40μg / m³ or VOCs>60μg / m³), but instantaneous emissions did not trigger a peak.
[0028] 3. >165 and x≥200 is defined as a high value exceeding the standard. Within this range, 90% of the samples are accompanied by strong temperature inversion (vertical temperature lapse rate <-0.5℃ / 100m) or cross-regional transport (external contribution ratio >30%).
[0029] In another embodiment of the present invention, instead of statically limiting the threshold, a dynamic threshold is determined, and real-time data is used. and The threshold is determined by comparing it with a dynamic threshold.
[0030] This includes the following sub-steps: Step S21: Determine the seasonal correction factor .
[0031] ; in For months (1-12). This is the seasonal influence coefficient, which is generally taken as 0.05. The factor value can be adjusted based on this coefficient.
[0032] Step S22: Determine the regional correction factor .
[0033] ; This represents the historical average ozone concentration of the area to be assessed over the past several years (e.g., 5 years) within the current time period. The average ozone concentration in the area to be assessed nationwide during the current time period.
[0034] Step S23: Determine the dynamic threshold based on the seasonal correction factor, the regional correction factor, and the static threshold standard.
[0035] The static thresholds include: Baseline threshold 1: TH_8h_base = 160μg / m³ (corresponding to...) National standard); Reference threshold 2: TH_1h_base = 200 μg / m³ (corresponding to National standard); Sub - threshold: TH_8h_low = 165 μg / m³ (used to distinguish "critical" from "low / high values").
[0036] The dynamic thresholds include: Dynamic reference threshold 1: ; Dynamic reference threshold 2: ; Dynamic sub - threshold: .
[0037] Step S24: Perform threshold division according to the dynamic thresholds.
[0038] If ≤TH_8h_effective and ≤TH_1h_effective, it is judged as meeting the standard; otherwise, continue to judge the type of exceeding the standard.
[0039] The judgment of the type of exceeding the standard includes: if TH_8h_effective < ≤TH_8h_low_effective and ≤TH_1h_effective, it is judged as a critical over - standard.
[0040] If >TH_8h_low_effective and <TH_1h_effective, it is set as a low - value over - standard.
[0041] If >TH_8h_low_effective and ≥TH_1h_effective, it is set as a high - value over - standard.
[0042] Furthermore, if there is a significant deviation (for example, the meteorological transmission contribution in the low - value over - standard cases continues to be high), the technical staff will be automatically reminded to review and calibrate the threshold parameters such as TH_8h_low and the dynamic adjustment factors, forming a closed - loop optimization.
[0043] Due to the national unified threshold (such as The threshold of ≤160 μg / m³ does not consider the concentration differences caused by natural background levels and long-term emission accumulation in different regions. This can easily lead to frequent false alarms in areas with high background levels (such as eastern urban clusters), while potentially causing missed alarms in areas with low background levels (such as western clean zones). Ozone formation has a strong photochemical dependence, with summer concentrations generally much higher than winter concentrations. Fixed thresholds cannot reflect this natural pattern, potentially resulting in almost no warnings in winter and excessively frequent warnings in summer that are difficult to differentiate in severity. Therefore, this embodiment improves the accuracy of warnings by determining a dynamic threshold.
[0044] Step S3: Based on the results of the dual-index joint threshold classification, establish a hierarchical causal analysis model.
[0045] Specifically, for scenarios involving critical, low-value, and high-value exceedances, dedicated models and computational logic are matched. The hierarchical causal analysis models include critical exceedance models, low-value exceedance models, and high-value exceedance models, which can visualize and output the contribution of each single factor. Time series decomposition plot (trend / seasonal / residual), VOCs component OFP comparison plot, pollution transport path map, potential source area contribution heat map, and local / external source contribution split pie chart. Step S3 includes the following sub-steps: Step S31: For critical exceedance, construct a critical exceedance model.
[0046] The critical exceedance model uses a multiple linear regression (MLR) model, which is constructed through the following sub-steps: Step S311: ... The concentrations of olefin VOCs, temperature, and wind speed (WS) are used as input data.
[0047] The coefficients were obtained by fitting historical samples (50,000 critical exceedance data points from 2018 to 2022) using the least squares method. ( Inhibition effect) (VOCs promoting effect) (Temperature promotes) (Wind speed suppression).
[0048] Step S312: Determine the expression for the multiple linear regression model.
[0049] The model formula is: = Concentration + (Olefin) concentration + .
[0050] ε represents the daily maximum 8-hour moving average concentration of ozone, and ε represents the random error term.
[0051] Step S313: Output the contribution of each factor from the multiple linear regression model.
[0052] The contribution of each factor is as follows: VOCs contribute 45%, and temperature contributes 30%.
[0053] After outputting the contribution of each factor, this embodiment can also determine the overall contribution and quantify the overall contribution. The combined contribution of the inhibitory effect and the promoting effect of VOCs to the critical exceedance was determined, addressing the issue of incomplete decomposition of the contribution of a single factor, and laying the foundation for subsequent yellow alerts. Differentiated management provides a direct basis (for example, if the overall contribution of precursors reaches 60%, then the management of precursors should be strengthened first).
[0054] ; in express The weighting coefficients, This represents the weighting coefficient of VOCs. Indicating the MLR model The regression coefficients, Indicates the critical period of exceeding the standard The measured concentration, This represents the regression coefficients of VOCs in the MLR model. This represents the regression coefficient for temperature (T) in the MLR model, where T represents the measured temperature during the critical exceedance period. This represents the regression coefficient of wind speed (WS) in the MLR model. This represents the measured wind speed during the critical period of exceeding the limit. Understandably, this is based on... and Fitting through historical data ( )get, , .
[0055] Step S32: For low-value exceedances, construct a low-value exceedance model.
[0056] The low-value exceedance model is generated through Seasonal Decomposition (STL) and Optical Photochemical Generation Potential (OFP) calculations. It separates the superimposed effects of precursor accumulation and meteorological cycles, and precisely distinguishes the contribution of continuous precursor emissions and meteorological conditions to ozone accumulation by quantifying VOCs photochemical activity. It includes the following sub-steps: Step S321: For periods when low values exceed the standard... Time series data, concentrations of VOCs components (e.g., olefins, aromatics, alkanes), and sunshine duration are used as input data.
[0057] Step S321: Perform time series decomposition based on the input data.
[0058] Decomposition by season The time series consists of a trend term (accumulated precursors), a seasonal term (solar cycle), and a residual term (random fluctuations).
[0059] Step S322: Calculate the photochemical generation potential.
[0060] The OFP of VOCs components is calculated as follows: OFP = concentration of each VOC component × reactivity coefficient. The reactivity coefficient is 10 for olefins and 5 for aromatics.
[0061] Step S323: Perform correlation analysis between the decomposed time series and photochemical generation potential, and output the results.
[0062] The decomposed trend term is correlated with OFP, and the seasonal term is correlated with sunshine duration. The cumulative contribution percentage of precursors and the meteorological contribution percentage are output.
[0063] Step S33: Construct a high-value exceeding-standard model for high-value exceeding-standard.
[0064] Coupled with local ozone formation simulation and regional transport source tracing, a model was constructed using the WRF-Chem model (10km×10km resolution), PSCF potential source contribution factor, and HYSPLIT trajectory model to accurately decompose the contribution ratio of local emissions and external transport, including the following sub-steps: Step S331: Using a 10km×10km resolution WRF-Chem model, input local emission inventory and meteorological field data to simulate... Generation and transmission.
[0065] The local emissions inventory includes industry, transportation, and agriculture.
[0066] Step S332: Track the ozone transport path using the HYSPLIT trajectory model.
[0067] The meteorological field data (air pressure, temperature, humidity, wind speed, and wind direction, which must match the resolution of the WRF-Chem model, such as 10km×10km, updated hourly) during periods of high exceedance values are input into the HYSPLIT trajectory model. Simultaneously, the starting point, simulation duration, and output frequency are set. The latitude and longitude coordinates of the ozone monitoring station are used as the starting point, with the starting altitude set at 10-50m (close to the near-surface monitoring layer to avoid interference from upper-air airflow). The simulation duration includes tracing back 24-72 hours before the pollution event (sufficient to cover the transport time of the air mass from the external source area to the local area); the output frequency is one trajectory node (latitude, longitude, and altitude) recorded every 6 hours to ensure complete trajectory details.
[0068] The HYSPLIT trajectory model integrates dynamic factors such as pressure gradient force and Coriolis force to extrapolate the movement trajectory of air masses hourly, generating a continuous trajectory line (including timestamps marking the arrival time of the air mass at each node) from the potential source area to the pollution monitoring point. Only periods of high-value exceedance are retained. >165μg / m³ and The trajectories of air masses with concentrations ≥200 μg / m³ were analyzed, and invalid trajectories that did not cause pollution were discarded. The starting point distribution of multiple valid trajectories was statistically analyzed, and densely clustered areas were identified as potential external sources of ozone pollution (such as industrial clusters and high-emission areas). Using an electronic map as the base map, the movement path of the air mass from the source area to the local area was displayed, and the time, altitude, and movement speed of key nodes were marked.
[0069] Step S333: Quantify the contribution ratio of the exogenous region using the PSCF model.
[0070] For example, transmission from North China accounts for 35% of the total.
[0071] The PSCF (Potential Source Contribution Factor) model quantifies the contribution of external source regions through trajectory-grid-concentration correlation statistics, binding air mass trajectories with pollution concentrations to calculate the contribution weight of potential source regions.
[0072] The high-value exceeding period output by the HYSPLIT model ( >165μg / m³ and Air mass trajectory data (≥200 μg / m³) (including the latitude and longitude nodes of each trajectory and the corresponding ozone concentration) are input into the PSCF model; the target area (including the pollution occurrence site and surrounding potential source areas) is divided into several uniform grids, and each grid is assigned a unique identifier; each valid trajectory is traversed, all grids passed through by the trajectory line are recorded, and the number of trajectory landing points in each grid is counted (denoted as ). , i.e., the number of times the air mass passes through grid (i,j). The high value exceeds the critical concentration ( Using ≥200μg / m³ as a benchmark, events whose trajectories passed through and whose corresponding ozone concentrations exceeded the standard were selected; the number of times the trajectory landed at the ozone-exceeding point in each grid was recorded (denoted as ). (i.e., the number of times air mass transport causes pollution to exceed the standard when grid (i,j) is the source region).
[0073] Calculate PSCF value : ; in , which is a weighting coefficient used to correct for low-sample grid errors. The less, The smaller the value, the less likely it is to cause misjudgment due to a small number of trajectories.
[0074] Calculate the PSCF of all potential source region grids (PSCFij ≥ 0.3, empirical threshold). sum ; The sum of PSCF values in a certain external source region (such as the North China grid group) and The ratio of the two is the proportion of external contribution to the region.
[0075] The sum of the PSCF values of all potential source region meshes and The ratio of external transmission to local generation is combined with the local generation contribution simulated by WRF-Chem to finally separate the proportion of external transmission to local generation (e.g., external contribution 35%, local generation 65%).
[0076] Step S334: Output the contribution ratio of ozone transport pathways and external source regions.
[0077] Step S4: Perform early warning mapping based on the output results of the hierarchical causal analysis model.
[0078] The hierarchical model constructed in step S3 has quantified the core influencing factors and contribution ratios of each type of exceedance. For example, critical exceedances are due to VOCs and temperature, while high exceedances are due to regional transport and strong temperature inversions. Therefore, this embodiment designs differentiated solutions for the core factors identified by the model. For example, if VOCs are the dominant factor, VOCs control is strengthened; if regional transport is not the dominant factor, joint prevention and control are initiated. Simultaneously, early warning mapping is performed, with the pollution intensity revealed by the model (e.g., critical / low / high value) directly mapping to the early warning level (yellow / orange / red), thereby ensuring that the early warning matches the actual pollution risk.
[0079] The critical value corresponds to a yellow alert, the low value corresponds to an orange alert, and the high value corresponds to a red alert.
[0080] Furthermore, in a yellow alert, if the overall contribution of precursors to V reaches 60%, then the control of precursors will be strengthened as a priority.
[0081] This embodiment can also provide decision-making suggestions, such as adjusting traffic restriction periods (e.g., extending the morning rush hour by 30 minutes) and strengthening the control of fugitive VOC emissions in the coating industry if the threshold is exceeded.
[0082] If the low value exceeds the standard, a photochemical warning will be issued, and petrochemical companies will be reminded to reduce raw material loading and unloading (9-11 am).
[0083] If the value exceeds the standard, cross-provincial joint prevention and control will be initiated (such as reducing industrial VOCs by 30% in sync with neighboring provinces) and open-air spraying operations will be suspended.
[0084] Furthermore, it provides visual output by displaying contribution bar charts, transmission path heatmaps, and early warning SMS / APP push notifications through the system interface (completed within 15 minutes).
[0085] Example 2 like Figure 2 As shown in the figure, an ozone pollution stratification and cause analysis system based on dual-index joint threshold division is provided in an embodiment of this application. Specifically, it includes: an acquisition and processing unit 210, a dual-index joint threshold division unit 220, a stratification cause analysis model establishment unit 230, and an early warning mapping unit 240.
[0086] The acquisition and processing unit 210 is used for data acquisition and preprocessing.
[0087] Data acquisition includes acquiring core metrics and auxiliary data.
[0088] The core indicators include the daily maximum 8-hour moving average concentration of ozone obtained in real time from the National Air Quality Monitoring Network API. ), the daily maximum hourly average concentration of ozone ( It is understandable that the core metrics obtained above are at the hourly level resolution.
[0089] The supporting data includes meteorological elements and precursors. Meteorological elements include temperature (T, °C), relative humidity (RH, %), wind speed (WS, m / s), wind direction (WD, °), and sunshine duration (H, h) (minute-level resolution, sourced from weather stations or satellite data); precursors include... (μg / m³), VOCs components (olefins, aromatics, alkanes, μg / m³) (hourly resolution, from online monitoring equipment); Preprocessing includes removing outliers and filling in missing values in the acquired data.
[0090] Outlier removal includes filtering data that exceeds a reasonable range using the 3σ principle (e.g., ...). >400μg / m³); Missing value imputation includes using linear interpolation to fill in missing data ≤2 hours in time, and marking data older than 2 hours as invalid samples.
[0091] The dual-indicator joint threshold division unit 220 is used to perform dual-indicator joint threshold division based on the preprocessed data.
[0092] Based on the GB3095-2012 Class II standard and statistical data from 31 provincial monitoring stations across the country during the high-temperature season (June-August) from 2018 to 2022 (a total of 1.2 million samples), the dual-indicator joint threshold division was performed, including setting the following thresholds: 1. 160 < ≤165 and <200 is defined as the critical exceedance. Statistical data shows that the precursor concentration of 80% of the samples in this range is close to the threshold, and the meteorological conditions are slightly unfavorable (T≥30℃, WS≤2m / s).
[0093] 2. >165 and <200 is considered a low value exceeding the limit. The cumulative effect of precursors is significant within this range. (>40μg / m³ or VOCs>60μg / m³), but instantaneous emissions did not trigger a peak.
[0094] 3. >165 and ≥200 is considered a high value exceeding the limit. Within this range, 90% of the samples are accompanied by strong temperature inversion (vertical temperature lapse rate < -0.5℃ / 100m) or cross-regional transport (external contribution rate > 30%).
[0095] In another embodiment of the present invention, instead of statically limiting the threshold, a dynamic threshold is determined, and real-time data is used. and The threshold is determined by comparing it with a dynamic threshold.
[0096] This includes the following sub-steps: Step Q1: Determine the seasonal correction factor .
[0097] ; in For months (1-12). This is the seasonal influence coefficient, which is generally taken as 0.05. The factor value can be adjusted based on this coefficient.
[0098] Step Q2: Determine the regional correction factor .
[0099] ; This represents the historical average ozone concentration of the area to be assessed over the past several years (e.g., 5 years) within the current time period. The average ozone concentration in the area to be assessed nationwide during the current time period.
[0100] Step Q3: Determine the dynamic threshold based on the seasonal correction factor, the regional correction factor, and the static threshold standard.
[0101] The static thresholds include: Baseline threshold 1: TH_8h_base = 160μg / m³ (corresponding to...) (National Standards) Benchmark threshold 2: TH_1h_base = 200 μg / m³ (corresponding to national standards); Sub - threshold: TH_8h_low = 165 μg / m³ (used to distinguish "critical" from "low - value / high - value").
[0102] The dynamic thresholds include: Dynamic benchmark threshold 1: ; Dynamic benchmark threshold 2: ; Dynamic sub - threshold: .
[0103] Step Q4: Perform threshold division according to the dynamic thresholds.
[0104] If ≤TH_8h_effective and ≤TH_1h_effective, it is judged as meeting the standard; otherwise, continue to judge the type of exceeding the standard.
[0105] The judgment of the type of exceeding the standard includes: if TH_8h_effective < ≤TH_8h_low_effective and ≤TH_1h_effective, it is judged as a critical over - standard.
[0106] If >TH_8h_low_effective and <TH_1h_effective, it is set as a low - value over - standard.
[0107] If >TH_8h_low_effective and ≥TH_1h_effective, it is set as a high - value over - standard.
[0108] Furthermore, if there is a significant deviation (for example, the meteorological transmission contribution in the case of low - value over - standard continues to be high), the technicians will be automatically reminded to review and calibrate the threshold parameters such as TH_8h_low and the dynamic adjustment factors, forming a closed - loop optimization.
[0109] Due to the national unified threshold (such as The threshold of ≤160 μg / m³ does not consider the concentration differences caused by natural background levels and long-term emission accumulation in different regions. This can easily lead to frequent false alarms in areas with high background levels (such as eastern urban clusters), while potentially causing missed alarms in areas with low background levels (such as western clean zones). Ozone formation has a strong photochemical dependence, with summer concentrations generally much higher than winter concentrations. Fixed thresholds cannot reflect this natural pattern, potentially resulting in almost no warnings in winter and excessively frequent warnings in summer that are difficult to differentiate in severity. Therefore, this embodiment improves the accuracy of warnings by determining a dynamic threshold.
[0110] The hierarchical causal analysis model establishment unit 230 is used to establish a hierarchical causal analysis model based on the results of the dual-index joint threshold division.
[0111] Specifically, for scenarios involving critical, low-value, and high-value exceedances, dedicated models and computational logic are matched. The hierarchical causal analysis models include critical exceedance models, low-value exceedance models, and high-value exceedance models, which can visualize and output the contribution of each single factor. Time series decomposition plot (trend / seasonal / residual), VOCs component OFP comparison plot, pollution transport path map, potential source area contribution heat map, and local / external source contribution split pie chart. The stratified causal analysis model building unit 230 performs the following sub-steps: Step E1: For critical exceedance, construct a critical exceedance model.
[0112] The critical exceedance model uses a multiple linear regression (MLR) model, which is constructed through the following sub-steps: Step E11: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.] The concentrations of olefin VOCs, temperature, and wind speed (WS) are used as input data.
[0113] The coefficients were obtained by fitting historical samples (50,000 critical exceedance data points from 2018 to 2022) using the least squares method. ( Inhibition effect) (VOCs promoting effect) (Temperature promotes) (Wind speed suppression).
[0114] Step E12: Determine the expression for the multiple linear regression model.
[0115] The model formula is: = Concentration + (Olefin) concentration + .
[0116] ε represents the daily maximum 8-hour moving average concentration of ozone, and ε represents the random error term.
[0117] Step E13: The multiple linear regression model outputs the contribution of each factor.
[0118] The contribution of each factor is as follows: VOCs contribute 45%, and temperature contributes 30%.
[0119] After outputting the contribution of each factor, this embodiment can also determine the comprehensive contribution V, thus quantifying the overall contribution. The combined contribution of the inhibitory effect and the promoting effect of VOCs to the critical exceedance was determined, addressing the issue of incomplete decomposition of the contribution of a single factor, and laying the foundation for subsequent yellow alerts. Differentiated management provides a direct basis (for example, if the overall contribution of precursors reaches 60%, then the management of precursors should be strengthened first).
[0120] ; in express The weighting coefficients, This represents the weighting coefficient of VOCs. Indicating the MLR model The regression coefficients, Indicates the critical period of exceeding the standard The measured concentration, This represents the regression coefficients of VOCs in the MLR model. This represents the regression coefficient for temperature (T) in the MLR model, where T represents the measured temperature during the critical exceedance period. This represents the regression coefficient of wind speed (WS) in the MLR model. This represents the measured wind speed during the critical period of exceeding the limit. Understandably, this is based on... and Fitting through historical data ( )get, , .
[0121] Step E2: For low-value exceedances, construct a low-value exceedance model.
[0122] The low-value exceedance model is generated through Seasonal Decomposition (STL) and Optical Photochemical Generation Potential (OFP) calculations. It separates the superimposed effects of precursor accumulation and meteorological cycles, and precisely distinguishes the contribution of continuous precursor emissions and meteorological conditions to ozone accumulation by quantifying VOCs photochemical activity. It includes the following sub-steps:
[0123] Step E21: For periods when low values exceed the standard... Time series data, concentrations of VOCs components (e.g., olefins, aromatics, alkanes), and sunshine duration are used as input data.
[0124] Step E21: Perform time series decomposition based on the input data.
[0125] Decomposition by season The time series consists of a trend term (accumulated precursors), a seasonal term (solar cycle), and a residual term (random fluctuations).
[0126] Step E22: Calculate the photochemical generation potential.
[0127] The OFP of VOCs components is calculated as follows: OFP = concentration of each VOC component × reactivity coefficient. The reactivity coefficient is 10 for olefins and 5 for aromatics.
[0128] Step E23: Perform correlation analysis between the decomposed time series and photochemical generation potential, and output the results.
[0129] The decomposed trend term is correlated with OFP, and the seasonal term is correlated with sunshine duration. The cumulative contribution percentage of precursors and the meteorological contribution percentage are output.
[0130] Step E3: Construct a high-value exceeding-standard model for high-value exceeding-standard.
[0131] Coupled with local ozone formation simulation and regional transport source tracing, a model was constructed using the WRF-Chem model (10km×10km resolution), PSCF potential source contribution factor, and HYSPLIT trajectory model to accurately decompose the contribution ratio of local emissions and external transport, including the following sub-steps: Step E31: Using a 10km×10km resolution WRF-Chem model, input local emission inventory and meteorological field data to simulate... Generation and transmission.
[0132] The local emissions inventory includes industry, transportation, and agriculture.
[0133] Step E32: Track the ozone transport path using the HYSPLIT trajectory model.
[0134] The meteorological field data (air pressure, temperature, humidity, wind speed, and wind direction, which must match the resolution of the WRF-Chem model, such as 10km×10km, updated hourly) during periods of high pollution levels are input into the HYSPLIT trajectory model. The starting point, simulation duration, and output frequency are also set. The latitude and longitude coordinates of the ozone monitoring station are used as the starting point, with the starting altitude set at 10-50m (close to the near-surface monitoring layer to avoid interference from upper-air airflow). The simulation duration includes tracing back 24-72 hours before the pollution event (sufficient to cover the transport time of the air mass from the external source region to the local area); the output frequency is one trajectory node (latitude, longitude, and altitude) recorded every 6 hours to ensure complete trajectory details.
[0135] The HYSPLIT trajectory model integrates dynamic factors such as pressure gradient force and Coriolis force to extrapolate the movement trajectory of air masses hourly, generating a continuous trajectory line (including timestamps marking the arrival time of the air mass at each node) from the potential source area to the pollution monitoring point. Only periods of high-value exceedance are retained. >165μg / m³ and The trajectories of air masses with concentrations ≥200 μg / m³ were analyzed, and invalid trajectories that did not cause pollution were discarded. The starting point distribution of multiple valid trajectories was statistically analyzed, and densely clustered areas were identified as potential external sources of ozone pollution (such as industrial clusters and high-emission areas). Using an electronic map as the base map, the movement path of the air mass from the source area to the local area was displayed, and the time, altitude, and movement speed of key nodes were marked.
[0136] Step E33: Quantify the contribution ratio of the exogenous region using the PSCF model.
[0137] For example, transmission from North China accounts for 35% of the total.
[0138] The PSCF (Potential Source Contribution Factor) model quantifies the contribution of external source regions through trajectory-grid-concentration correlation statistics, binding air mass trajectories with pollution concentrations to calculate the contribution weight of potential source regions.
[0139] The high-value exceeding period output by the HYSPLIT model ( >165μg / m³ and Air mass trajectory data (≥200 μg / m³) (including the latitude and longitude nodes of each trajectory and the corresponding ozone concentration) are input into the PSCF model; the target area (including the pollution occurrence site and surrounding potential source areas) is divided into several uniform grids, and each grid is assigned a unique identifier; each valid trajectory is traversed, all grids passed through by the trajectory line are recorded, and the number of trajectory landing points in each grid is counted (denoted as ). , i.e., the number of times the air mass passes through grid (i,j). The high value exceeds the critical concentration ( Using ≥200μg / m³ as a benchmark, events whose trajectories passed through and whose corresponding ozone concentrations exceeded the standard were selected; the number of times the trajectory landed at the ozone-exceeding point in each grid was recorded (denoted as ). (i.e., the number of times air mass transport causes pollution to exceed the standard when grid (i,j) is the source region).
[0140] Calculate PSCF value : ; in , which is a weighting coefficient used to correct for low-sample grid errors. The less, The smaller the value, the less likely it is to cause misjudgment due to a small number of trajectories.
[0141] Calculate the PSCF of all potential source region grids (PSCFij ≥ 0.3, empirical threshold). sum ; The sum of PSCF values in a certain external source region (such as the North China grid group) and The ratio of the two is the proportion of external contribution to the region.
[0142] The sum of the PSCF values of all potential source region meshes and The ratio of external transmission to local generation is combined with the local generation contribution from WRF-Chem simulation to finally separate the proportion of external transmission to local generation (e.g., external contribution 35%, local generation 65%).
[0143] Step E34: Output the contribution percentage of the ozone transport path and the external source region.
[0144] The early warning mapping unit 240 is used to perform early warning mapping based on the output results of the hierarchical causal analysis model.
[0145] The hierarchical model constructed by the hierarchical causal analysis model building unit 230 has quantified the core influencing factors and contribution ratios of each type of exceedance. For example, critical exceedances are due to VOCs and temperature, while high exceedances are due to regional transport and strong temperature inversions. Therefore, this embodiment designs differentiated solutions for the core factors identified by the model. For example, if VOCs are the dominant factor, VOCs control is strengthened; if regional transport is not the dominant factor, joint prevention and control are initiated. At the same time, early warning mapping is performed, and the pollution intensity revealed by the model (such as critical / low / high value) is directly mapped to the early warning level (yellow / orange / red), thereby ensuring that the early warning matches the actual pollution risk.
[0146] The critical value corresponds to a yellow alert, the low value corresponds to an orange alert, and the high value corresponds to a red alert.
[0147] Furthermore, in a yellow alert, if the overall contribution of precursors to V reaches 60%, then the control of precursors will be strengthened as a priority.
[0148] This embodiment can also provide decision-making suggestions, such as adjusting traffic restriction periods (e.g., extending the morning rush hour by 30 minutes) and strengthening the control of fugitive VOC emissions in the coating industry if the threshold is exceeded.
[0149] If the low value exceeds the standard, a photochemical warning will be issued, and petrochemical companies will be reminded to reduce raw material loading and unloading (9-11 am).
[0150] If the value exceeds the standard, cross-provincial joint prevention and control will be initiated (such as reducing industrial VOCs by 30% in sync with neighboring provinces) and open-air spraying operations will be suspended.
[0151] Furthermore, it provides visual output by displaying contribution bar charts, transmission path heatmaps, and early warning SMS / APP push notifications through the system interface (completed within 15 minutes).
[0152] This application also provides a computer storage medium storing computer instructions, which, when invoked, are used to execute the ozone pollution stratification and causal analysis method based on dual-index joint threshold division.
[0153] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions. When the computer program instructions are executed on a computer, the computer executes the above-described method for analyzing the stratification of ozone pollution causes based on a dual-index joint threshold.
[0154] This invention provides a processor for processing the above-described method for analyzing the stratification of ozone pollution causes based on a dual-index joint threshold.
[0155] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0156] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.
[0157] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.
[0158] Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).
[0159] This application has the following beneficial effects: This application overcomes the limitations of single-indicator judgment by using a dual-indicator joint threshold classification, achieving precise segmentation of exceedance types. Furthermore, the entire process is automated, with short response times, significantly improving decision-making efficiency. This application lowers the decision-making threshold through visualization design and effectively reduces the number of ozone pollution days and optimizes the allocation of governance resources through early warning mapping, demonstrating significant practical value and promotional significance.
[0160] Although the examples referenced in this application are described for illustrative purposes only and not for limiting the scope of this application, changes, additions and / or deletions to the implementation may be made without departing from the scope of this application.
[0161] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for stratifying the causes of ozone pollution based on a dual-index joint threshold classification, characterized in that, Includes the following steps: Data acquisition and preprocessing are performed. Based on the preprocessed data, a dual-indicator joint threshold division is performed; Based on the results of the dual-index joint threshold classification, a hierarchical causal analysis model was established. Early warning mapping is performed based on the output of the hierarchical causal analysis model.
2. The ozone pollution stratification and causal analysis method based on dual-index joint threshold division as described in claim 1, characterized in that, Data acquisition includes acquiring core metrics and auxiliary data; Key indicators include obtaining the daily maximum 8-hour moving average concentration of ozone. The daily maximum hourly average concentration of ozone The auxiliary data includes meteorological elements and precursors.
3. The ozone pollution stratification and causal analysis method based on dual-index joint threshold division as described in claim 2, characterized in that, Based on the preprocessed data, the joint threshold division of the two indicators includes: Setting a critical threshold for exceeding the limit; Setting a low value that exceeds the limit; The set value exceeded the limit.
4. The ozone pollution stratification and causal analysis method based on dual-index joint threshold division as described in claim 3, characterized in that, Based on the results of the dual-index joint threshold classification, the hierarchical causal analysis model is established, including the following sub-steps: For critical exceedances, construct a critical exceedance model; For low-value exceedances, construct a low-value exceedance model; For high values exceeding the standard, a high value exceeding the standard model is constructed.
5. The ozone pollution stratification and causal analysis method based on dual-index joint threshold division as described in claim 4, characterized in that, For critical exceedance, constructing a critical exceedance model includes the following sub-steps: Will Concentration, olefin VOCs concentration, temperature, and wind speed data (WS) are used as input data. Determine the expression for the multiple linear regression model; The multiple linear regression model outputs the contribution of each factor.
6. A system for analyzing the stratified causes of ozone pollution based on a dual-index joint threshold classification, characterized in that, include: The system includes an acquisition and processing unit, a dual-indicator joint threshold division unit, a hierarchical causal analysis model establishment unit, and an early warning mapping unit. The data acquisition and processing unit is used for data acquisition and preprocessing. The dual-indicator joint threshold division unit is used to perform dual-indicator joint threshold division based on the preprocessed data. The hierarchical causal analysis model building unit is used to build a hierarchical causal analysis model based on the results of the dual-index joint threshold division. The early warning mapping unit is used to perform early warning mapping based on the output results of the hierarchical causal analysis model.
7. The ozone pollution stratification and causal analysis system based on dual-index joint threshold division as described in claim 6, characterized in that, The data acquisition and processing unit performs data acquisition, including acquiring core indicators and auxiliary data; Key indicators include obtaining the daily maximum 8-hour moving average concentration of ozone. The daily maximum hourly average concentration of ozone The auxiliary data includes meteorological elements and precursors.
8. The ozone pollution stratification and causal analysis system based on dual-index joint threshold division as described in claim 6, characterized in that, The dual-indicator joint threshold division unit performs dual-indicator joint threshold division based on the preprocessed data, including: Setting a critical threshold for exceeding the limit; Setting a low value that exceeds the limit; The set value exceeded the limit.
9. The ozone pollution stratification and causal analysis system based on dual-index joint threshold division as described in claim 8, characterized in that, The hierarchical causal analysis model establishment unit, based on the results of the dual-index joint threshold classification, establishes the hierarchical causal analysis model, including the following sub-steps: For critical exceedances, construct a critical exceedance model; For low-value exceedances, construct a low-value exceedance model; For high values exceeding the standard, a high value exceeding the standard model is constructed.
10. The ozone pollution stratification and causal analysis system based on dual-index joint threshold division as described in claim 9, characterized in that, For the critical exceedance analysis model building unit, the construction of the critical exceedance model includes the following sub-steps: Use NO2 concentration, olefin VOCs concentration, temperature, and wind speed data (WS) as input data. Determine the expression for the multiple linear regression model; The multiple linear regression model outputs the contribution of each factor.