A method for estimating the photolysis rate of ambient NO2 based on multivariate data

Abstract

The invention discloses an environmental NO based on multivariate data. 2 The photolysis rate estimation method uses conventional pollutant concentration detection data from environmental monitoring stations combined with meteorological reanalysis data and the hour and month of the data as covariates to participate in the model calculation. The machine learning model is applied to NO measured at specific urban points. 2 Photolysis rate, calculation and inference of NO in the whole area at the city scale 2 The spatial resolution is 0.25° and the temporal resolution is one hour. The estimation results of the present invention are more comprehensive and accurate, which improves the reliability and stability of the model and improves the prediction of NO in the environment. 2 The real-time estimation capability and accuracy of the photolysis rate provide data support for the reconstruction of the historical photolysis rate of the region and the study of the mechanism of regional photochemical pollution.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence, environmental science and atmospheric science technology, and in particular to a method for estimating the photolysis rate of environmental NO2 based on multivariate data. Background Art

[0002] Solar ultraviolet radiation participates in atmospheric photochemical reactions by driving photodissociation reactions in the troposphere. Photolysis reactions play an important role in atmospheric gas-phase chemical reactions and have an important impact on the transformation of chemical substances in the atmosphere. NO2 is a key chemical component in atmospheric chemistry. The rapid photolysis of NO2 in the troposphere largely controls the formation of tropospheric ozone (O3) and OH radicals (secondary products of O3 photolysis). The photolysis rate is a key parameter that reflects the characteristics of my country's atmospheric complex pollution and is the hub of the interaction between different pollutants. Therefore, the photolysis rate of NO2 is one of the key influencing factors for the formation of tropospheric ozone.

[0003] The current direct observation method for photolysis rate is to directly observe the actinic radiation flux spectrum, and then combine the laboratory data of cross-sectional area and quantum yield to calculate the photolysis rate of the photochemical reaction. Actinic flux is the only radiation quantity suitable for photolysis rate determination, but it is relatively lacking in actual environmental observations, resulting in the current observation data on photolysis rate being relatively scarce. Summary of the invention

[0004] In view of the defects of the prior art, the present invention provides an environmental NO2 photolysis rate estimation method based on multivariate data to obtain the NO2 photolysis rate in the atmospheric environment of an area without direct measurement of the actinic radiation flux spectrum.

[0005] In order to achieve the above invention object, the technical solution adopted by the present invention is as follows:

[0006] A method for estimating the photolysis rate of ambient NO2 based on multivariate data comprises the following steps:

[0007] 1) Data collection and information integration: Collect NO2 photolysis rate measured by spectrometer and atmospheric pollutant concentration data at fixed sites, including O3, NO2, PM 2.5 and PM 10 , as well as meteorological reanalysis data and the hour and month of the data.

[0008] 2) Data cleaning preprocessing: Obtain the point location based on the NO2 photolysis rate, and select the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid. Clean the outliers.

[0009] 3) Machine learning modeling: A machine learning model was established with NO2 photolysis rate as the dependent variable and atmospheric pollutant observation data, meteorological reanalysis data, and the hour and month of the data as covariates.

[0010] 4) Model validation and evaluation: The data were divided into training set and validation set using the ten-fold cross validation method to evaluate the model fitting effect parameters.

[0011] 5) Analysis of the changing characteristics of the NO2 photolysis rate in the study area: Based on the constructed machine learning model, the NO2 photolysis rate constant in the study area is calculated by inputting the covariates of the specific area.

[0012] Preferably, in step 1), the time resolution of all data is 1 hour.

[0013] Preferably, in step 4), the model fitting effect parameters are evaluated, including: the coefficient of determination R 2 , root mean square error RMSE, slope slope.

[0014] Preferably, in step 4), the accuracy of the model is tested based on multiple dimensions including years and quarters.

[0015] Preferably, in step 3) and step 4), different machine learning model solutions are used for comparison, including random forest, XGBoost and LightGBM. The best hyperparameters of each model are obtained through grid search, and the fitting effects of different models in the best state are compared to obtain the optimal model.

[0016] Preferably, in step 6), the regional resolution is 0.25° grid and the time resolution is 1 hour.

[0017] The present invention also discloses a system for estimating the photolysis rate of NO2 in the environment based on multivariate data. The system can be used to implement the above-mentioned method for estimating the photolysis rate of NO2 in the environment, and specifically includes:

[0018] Data collection and integration module: responsible for collecting NO2 photolysis rate data measured by spectrometer, atmospheric pollutant concentration data at fixed sites, meteorological reanalysis data and data hours and months. And integrate data from different data sources into a format that can be used by the system.

[0019] Data preprocessing module: responsible for data cleaning and preprocessing, including obtaining points according to the NO2 photolysis rate, selecting the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid, and cleaning outliers.

[0020] Machine learning model module: Take the NO2 photolysis rate as the dependent variable, and the atmospheric pollutant observation data and meteorological reanalysis data as well as the hour and month of the data as covariates to establish a machine learning model.

[0021] Model validation and evaluation module: responsible for validating and evaluating the model using the ten-fold cross validation method. The data is divided into a training set and a validation set, the performance of the model on the validation set is evaluated, and the model fitting effect parameters are calculated.

[0022] Ambient NO2 photolysis rate change characteristics analysis module: Based on the constructed machine learning model, this module calculates the ambient NO2 photolysis rate constant in the study area by inputting the covariates of a specific area. This module can further analyze and interpret the model results and provide a detailed analysis of the characteristics of the ambient NO2 photolysis rate changes in the study area.

[0023] Preferably, the ambient NO2 photolysis rate estimation system also includes: a user interface module: providing a user-friendly interface so that users can easily use the various functional modules of the system, including data uploading, model training, and result display.

[0024] Preferably, the ambient NO2 photolysis rate estimation system also includes: a report generation module: generating reports or visualization charts based on the results of model validation evaluation and ambient NO2 photolysis rate change characteristic analysis, so that users can understand the performance of the model and research results more intuitively.

[0025] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for estimating the ambient NO2 photolysis rate when executing the program.

[0026] The present invention also discloses a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned method for estimating the ambient NO2 photolysis rate is implemented.

[0027] Compared with the prior art, the advantages of the present invention are:

[0028] Comprehensive data information: Utilizing multivariate data, including NO2 photodecomposition rate measured by spectrometer, atmospheric pollutant observation data, and meteorological reanalysis data, the impact of multiple factors on NO2 photodecomposition rate is comprehensively considered, which can provide more comprehensive and accurate estimation results.

[0029] Data richness: By integrating multi-source data, richer information can be provided, including data on atmospheric pollutant concentrations, meteorological conditions, and other aspects, which helps to more comprehensively analyze the changing patterns of NO2 photolysis rate in the environment.

[0030] Model building flexibility: Machine learning modeling methods are highly flexible and can adapt to different data characteristics and patterns. They can also automatically learn and adjust models, making them more adaptable and predictive.

[0031] The model validation method is reliable: the model is validated and evaluated using the ten-fold cross-validation method, which can effectively evaluate the generalization ability and prediction accuracy of the model and improve the reliability and stability of the model.

[0032] Real-time and accuracy: Through the comprehensive use of multivariate data, the real-time estimation capability and accuracy of NO2 photolysis rate in the environment can be improved, providing timely and accurate data support for environmental monitoring and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of a method for estimating the ambient NO2 photolysis rate based on multivariate data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples.

[0035] like Figure 1 As shown, the present invention provides a method for estimating the photolysis rate of ambient NO2 based on multivariate data. The specific steps are as follows:

[0036] 1) Data collection and information integration: Obtain NO2 photolysis rate calculated by photochemical flux measured by spectrometer, and atmospheric pollutant concentration data (O3, NO2, PM 2.5 , PM 10 ), meteorological reanalysis data such as ERA5 data produced by the European Centre for Medium-Range Weather Forecasts. All data have a time resolution of 1 hour.

[0037] 2) Data cleaning preprocessing: The point position is obtained according to the NO2 photolysis rate, and the atmospheric pollutant observation data of the nearest fixed station and the meteorological reanalysis data within the grid are matched and selected to clean the abnormal values ​​(continuous identical values).

[0038] 3) Machine learning modeling: The NO2 photolysis rate was used as the dependent variable, and the atmospheric pollutant observation data, meteorological reanalysis data, and the hour and month of the data were used as covariates to construct a machine learning model.

[0039] 4) Model validation evaluation: The 10-fold cross validation method was used to divide the training set and the validation set, and the model fitting effect parameters, such as the determination coefficient R 2, RMSE, slope, etc.; the accuracy of model calculation is also tested from multiple dimensions such as year and quarter. The specific steps are as follows:

[0040] Data preparation: Group the raw data according to different time dimensions (year, quarter, etc.) to ensure the integrity and accuracy of the data in each time period.

[0041] Model application: Use the established machine learning model to predict or estimate the data for each time period.

[0042] Performance evaluation: For each time period, the error between the model's predictions and the true observations is calculated.

[0043] Result summary: Summarize the model performance evaluation results for each time period to obtain the performance of the model in different time dimensions.

[0044] Analyze and explain: Analyze the performance of the model over time, compare the model predictions for different time periods, check whether the model performs worse in some time periods, and try to explain the reasons for this difference.

[0045] 5) Model optimization: In the step of machine learning modeling and model verification, different machine learning model solutions, such as random forest, XGBoost, and LightGBM, are selected for comparison. Grid search is used to obtain the best hyperparameters of each model, and the fitting results of different models under the best conditions are compared to obtain the model with the best fitting effect.

[0046] 6) When studying the changing characteristics of the NO2 photolysis rate in the regional environment, the NO2 photolysis rate constant in the study area is calculated based on the constructed model by inputting the covariates in the specific region. The regional resolution is 0.25° grid and the time resolution is 1 hour.

[0047] Embodiment 1

[0048] A certain city is one of the megacities in China with serious complex pollution. In recent years, the pollution characteristics have changed from being dominated by particulate matter pollution to being dominated by O3 pollution. It is of great significance to study the laws of urban atmospheric O3 pollution. The change of O3 concentration depends on the calculation of atmospheric photochemical models, among which the photolysis rate, especially the photolysis rate of NO2 (jNO2), is one of the key factors affecting the formation of tropospheric ozone. The reconstruction of the spatiotemporal distribution of jNO2 at the urban scale is an important basis for the refined control of air quality and is of great significance to the prevention and control of atmospheric pollution in the region.

[0049] This embodiment uses the LightGBM machine learning calculation module to integrate the jNO2 data monitored by the ultrafast CCD detector spectrometer (UF-CCD, MetCon, Germany), the ERA5 data of the fifth-generation reanalysis online database of global climate and weather conducted by the European Center for Medium-Range Weather Forecasts (ECMWF), and the atmospheric pollutant concentration data monitored by fixed stations to achieve the goal of accurately reconstructing the spatiotemporal distribution of jNO2 in urban areas.

[0050] 1. Data Collection and Processing

[0051] (1) jNO2 monitoring data

[0052] The jNO2 monitoring data were obtained from an ultrafast CCD detector spectrometer (UF-CCD, MetCon, Germany).

[0053] Since the photochemical reaction at night is not strong, the data has little reference value for model construction and may even affect normal construction. Therefore, the study removed the night data. The division of the daytime and nighttime data should consider both retaining as much data as possible and removing data points with low reference value. In a large number of verifications, the study determined that the dividing point is that jNO2 less than 0.00026 is night data and is deleted. Abnormal values ​​that appear repeatedly at the same time for multiple days may be caused by the shielding of nearby objects. The abnormal values ​​are identified and manually removed.

[0054] (2) ERA5 data

[0055] ERA5 is the fifth generation of ECMWF reanalysis of global climate and weather over the past 80 years. The reanalysis uses the laws of physics to combine model data with observations from around the world into a globally complete and consistent data set. This principle is called data assimilation and is based on the methods used by the Numerical Weather Prediction Center. The reanalysis produces data from decades ago, providing an accurate description of past climate.

[0056] This example extracts variables from ERA5, including temperature, irradiation, heat exchange, evaporation rate, cloud cover, precipitation, pressure, etc., with a resolution of 0.25° for 6 species parameters and 0.1° for 8 species parameters. The latter is generated by replaying the land component ERA5-land of the ECMWFERA5 climate reanalysis, resulting in a higher resolution result.

[0057] (3) Air pollution index data

[0058] Air pollution index data comes from fixed stations (national control stations), and variables include NO2, O3, PM 2.5 、PM 10 .

[0059] (4) Other data

[0060] Extract the hour and month of each data item.

[0061] 2. Model Construction and Optimization

[0062] LightGBM is a tree-based machine learning algorithm specifically designed to solve classification and regression problems. It can perform efficient and high-performance parallel computing and has excellent performance in controlling model overfitting.

[0063] In this embodiment, the data preprocessing stage first uses the data resampling method in the R language foxat package to match the jNO2 data with the data of covariates such as temperature, irradiation, heat exchange, evaporation rate, cloud cover, precipitation, pressure, forecast albedo and four air quality indicators in ERA5. Next, the LightGBM algorithm is used for machine learning training and prediction, with jNO2 as the dependent variable, other covariates and the extracted hours and months as independent variables to train the model, and the rows with all or not all null values ​​in the data set are deleted to reconstruct the high-resolution hourly jNO2 concentration of the city from 2021 to 2023.

[0064] Hyperparameters refer to parameters that need to be determined before model training, such as learning rate, maximum depth of decision tree, number of layers of neural network, etc. In machine learning and deep learning, hyperparameter tuning is an important part of the model training process. Appropriate hyperparameter settings can significantly improve the performance of the model. The best hyperparameter search method used in the study is grid search. As an exhaustive search method, it searches for the optimal hyperparameter by traversing all possible combinations of hyperparameters. Grid search first sets a set of candidate values ​​for each hyperparameter, and then generates the Cartesian product of these candidate values ​​to form a combination grid of hyperparameters. Then, grid search trains and evaluates the model for each hyperparameter combination to find the hyperparameter combination with the best performance.

[0065] In summary, the preparation of jNO2 data and its covariates in the study area, the construction and optimization of the model can calculate and infer the high-resolution spatiotemporal distribution of jNO2 concentration. The results of the ten-fold cross validation based on the samples show that the model R 2 The value is 0.845, the RMSE is 8.400E-04, and the slope of the fitting line is 0.847.

[0066] Table 1 Variable information table

[0067]

[0068]

[0069] By applying the training model, combined with the pollutant concentrations at the observation sites in the study area and the ERA5 reanalysis data, the spatiotemporal distribution of the regional NO2 photolysis rate can be calculated and inferred.

[0070] In another embodiment of the present invention, a system for estimating the photolysis rate of NO2 in an environment based on multivariate data is provided. The system can be used to implement the above-mentioned method for estimating the photolysis rate of NO2 in an environment based on multivariate data, and specifically includes:

[0071] Data collection and integration module: responsible for collecting NO2 photolysis rate data measured by spectrometer, atmospheric pollutant concentration data at fixed sites, meteorological reanalysis data and data hours and months. And integrate data from different data sources into a format that can be used by the system.

[0072] Data preprocessing module: responsible for data cleaning and preprocessing, including obtaining points according to the NO2 photolysis rate, selecting the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid, and cleaning outliers.

[0073] Machine learning model module: Take the NO2 photolysis rate as the dependent variable, and the atmospheric pollutant observation data and meteorological reanalysis data as well as the hour and month of the data as covariates to establish a machine learning model.

[0074] Model validation and evaluation module: responsible for validating and evaluating the model using the ten-fold cross validation method. The data is divided into a training set and a validation set, the performance of the model on the validation set is evaluated, and the model fitting effect parameters are calculated.

[0075] Ambient NO2 photolysis rate change characteristics analysis module: Based on the constructed machine learning model, this module calculates the ambient NO2 photolysis rate constant in the study area by inputting the covariates of a specific area. This module can further analyze and interpret the model results and provide a detailed analysis of the characteristics of the ambient NO2 photolysis rate changes in the study area.

[0076] User interface module: provides a user-friendly interface that enables users to easily use the various functional modules of the system, including data uploading, model training, result display, etc.

[0077] Report generation module: Generate reports or visualization charts based on the results of model validation evaluation and environmental NO2 photolysis rate change characteristics analysis, so that users can understand the model performance and research results more intuitively.

[0078] In another embodiment of the present invention, a terminal device is provided, the terminal device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding functions; the processor described in the embodiment of the present invention can be used for the operation of a method for estimating the photolysis rate of environmental NO2 based on multivariate data, including the following steps:

[0079] 1) Data collection and information integration: Collect NO2 photolysis rate measured by spectrometer and atmospheric pollutant concentration data at fixed sites, including O3, NO2, PM 2.5 、PM 10 , as well as meteorological reanalysis data and the hour and month of the data.

[0080] 2) Data cleaning preprocessing: Obtain the point location based on the NO2 photolysis rate, and select the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid. Clean the outliers.

[0081] 3) Machine learning modeling: A machine learning model was established with NO2 photolysis rate as the dependent variable and atmospheric pollutant observation data, meteorological reanalysis data, and the hour and month of the data as covariates.

[0082] 4) Model validation and evaluation: The data were divided into training set and validation set using the ten-fold cross validation method to evaluate the model fitting effect parameters.

[0083] 5) Analysis of the changing characteristics of the NO2 photolysis rate in the study area: Based on the constructed machine learning model, the NO2 photolysis rate constant in the study area is calculated by inputting the covariates of the specific area.

[0084] In another embodiment of the present invention, the present invention further provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one disk storage.

[0085] One or more instructions stored in a computer-readable storage medium may be loaded and executed by a processor to implement the corresponding steps of a method for estimating the photolysis rate of ambient NO2 based on multivariate data in the above embodiment; one or more instructions in the computer-readable storage medium may be loaded and executed by a processor as follows:

[0086] 1) Data collection and information integration: Collect NO2 photolysis rate measured by spectrometer and atmospheric pollutant concentration data at fixed sites, including O3, NO2, PM 2.5 、PM 10 , as well as meteorological reanalysis data and the hour and month of the data.

[0087] 2) Data cleaning preprocessing: Obtain the point location based on the NO2 photolysis rate, and select the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid. Clean the outliers.

[0088] 3) Machine learning modeling: A machine learning model was established with NO2 photolysis rate as the dependent variable and atmospheric pollutant observation data, meteorological reanalysis data, and the hour and month of the data as covariates.

[0089] 4) Model validation and evaluation: The data were divided into training set and validation set using the ten-fold cross validation method to evaluate the model fitting effect parameters.

[0090] 5) Analysis of the changing characteristics of the NO2 photolysis rate in the study area: Based on the constructed machine learning model, the NO2 photolysis rate constant in the study area is calculated by inputting the covariates of the specific area.

[0091] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0095] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the implementation methods of the present invention, and should be understood that the protection scope of the present invention is not limited to such special statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for estimating the photolysis rate of ambient NO2 based on multivariate data, characterized in that: The following steps are involved: 1) Data collection and information integration: Collect NO2 photolysis rate measured by spectrometer and atmospheric pollutant concentration data at fixed sites, including O3, NO2, PM 2.5 and PM 10 , meteorological reanalysis data and the hour and month of the data; 2) Data cleaning and preprocessing: obtain the point location according to the NO2 photolysis rate, select the atmospheric pollutant observation data of the nearest fixed station and the meteorological reanalysis data within the grid; clean the outliers; 3) Machine learning modeling: Taking NO2 photolysis rate as the dependent variable, and taking atmospheric pollutant observation data and meteorological reanalysis data as well as the hour and month of the data as covariates, a machine learning model was established; 4) Model validation and evaluation: Using the ten-fold cross-validation method, the data is divided into a training set and a validation set to evaluate the model fitting effect parameters; the accuracy of the model is tested from multiple dimensions including years and quarters; The specific steps are as follows: Data preparation: Group the raw data according to different time dimensions; Model application: Use the established machine learning model to predict or estimate the data for each time period; Performance evaluation: For each time period, the error between the model's predictions and the true observations is calculated; Result summary: Summarize the model performance evaluation results of each time period to obtain the performance of the model in different time dimensions; Analysis and interpretation: Analyze the performance of the model in different time dimensions and compare the model prediction results in different time periods; 5) Analysis of the variation characteristics of NO2 photolysis rate in the study area: Based on the constructed machine learning model, the NO2 photolysis rate constant in the study area is calculated by inputting the covariates of the specific area; In step 3) and step 4), different machine learning model solutions are used for comparison, including random forest, XGBoost, and LightGBM; the best hyperparameters of each model are obtained through grid search, and the fitting effects of different models in the best state are compared to obtain the optimal model.

2. The method for estimating the ambient NO2 photolysis rate according to claim 1, characterized in that: In step 1), the time resolution of all data is 1 hour.

3. The method for estimating the ambient NO2 photolysis rate according to claim 1, characterized in that: In step 4), the model fitting parameters are evaluated, including the coefficient of determination R 2 , root mean square error RMSE, slope slope.

4. The method for estimating the ambient NO2 photolysis rate according to claim 1, characterized in that: In step 6), the regional resolution is 0.25° grid and the temporal resolution is 1 hour.

5. A system for estimating the ambient NO2 photolysis rate based on multivariate data, characterized in that: The system can be used to implement the method for estimating the ambient NO2 photolysis rate according to any one of claims 1 to 4, comprising: Data collection and integration module: responsible for collecting NO2 photolysis rate data measured by the spectrometer, atmospheric pollutant concentration data at fixed sites, and meteorological reanalysis data; and integrating data from different data sources into a format that can be used by the system; Data preprocessing module: responsible for data cleaning and preprocessing, including obtaining points according to the NO2 photolysis rate, selecting the nearest fixed-site atmospheric pollutant observation data and the meteorological reanalysis data within the grid, and cleaning abnormal values; Machine learning model module: Take the NO2 photolysis rate as the dependent variable, and the atmospheric pollutant observation data and meteorological reanalysis data as well as the hour and month of the data as covariates to establish a machine learning model; Model validation and evaluation module: responsible for validating and evaluating the model using the ten-fold cross-validation method; dividing the data into a training set and a validation set, evaluating the performance of the model on the validation set, and calculating the model fitting effect parameters; Ambient NO2 photolysis rate change characteristics analysis module: Based on the constructed machine learning model, this module calculates the ambient NO2 photolysis rate constant in the study area by inputting covariates of a specific area; this module can further analyze and interpret the model results, and provide a detailed analysis of the characteristics of the change in ambient NO2 photolysis rate in the study area.

6. The system for estimating the ambient NO2 photolysis rate according to claim 5, characterized in that: The ambient NO2 photolysis rate estimation system also includes: User interface module: provides a user-friendly interface that enables users to easily use the various functional modules of the system, including data uploading, model training, and result display.

7. The system for estimating the ambient NO2 photolysis rate according to claim 6, characterized in that: The ambient NO2 photolysis rate estimation system also includes: a report generation module: based on the results of the model validation evaluation and the analysis of the characteristics of the ambient NO2 photolysis rate changes, a report or a visualization chart is generated so that users can understand the performance of the model and the research results more intuitively.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for estimating the ambient NO2 photolysis rate as described in one of claims 1 to 4 is implemented.

Images