Methods and systems for street-scale atmospheric pollution simulation and source apportionment

Abstract

The application provides a street block scale atmospheric pollution simulation and tracing method and system, the method comprising: obtaining a street block scale dynamic field based on mesoscale meteorological data by using a comprehensive simulation strategy of machine learning model and parameterized model fusion; analyzing street block scale emission data into pollution source characteristic information; based on simulation grid data, mesoscale pollution distribution and departmental contribution data, precipitation data, street block scale dynamic field and pollution source characteristic information, performing collaborative processing to obtain street block scale pollution distribution data; based on the street block scale pollution distribution data and the interaction mechanism of pollutants and environmental elements to generate a correction signal to dynamically adjust the collaborative processing process; outputting street block scale pollution distribution and tracing result data. The application can reduce the consumption of computing resources and simulation time while ensuring high-precision simulation of wide-area coverage, and has the ability to simulate the interaction of pollutants and environmental factors, which can improve the authenticity of street block scale pollution simulation.

Description

Technical Field

[0001] This application relates to the field of atmospheric pollution simulation technology, and in particular to a method and system for simulating and tracing atmospheric pollution at the street scale. Background Technology

[0002] Air pollution poses a serious threat to human health. In the field of air pollution simulation and assessment, existing methods mainly fall into two categories: one directly utilizes concentration data from environmental monitoring stations, or performs spatial interpolation based on this data, to estimate the pollution distribution within a region; the other is simulation methods based on numerical air quality models. Commonly used mesoscale air quality models include the Community Multiscale Air Quality (CMAQ) model and the Comprehensive Air Quality Model with Extensions (CAMx) model. These models typically require the output of meteorological field models (such as Weather Research and Forecasting, WRF) as input to provide dynamic support for the transport and diffusion of pollutants. Mesoscale models can comprehensively consider pollution source emissions, atmospheric physical processes (convection, diffusion, deposition, etc.), and atmospheric chemical processes (photochemical reactions, secondary pollution generation, etc.), and have been widely used in urban and regional-scale air pollution simulation and source tracing studies. Their typical spatial resolution is 3–9 km, which can be improved to approximately 1 km when using high-resolution configurations. While this resolution is sufficient for regional-scale pollution studies, at the block scale, a 1 km grid often includes multiple roads, building complexes, and complex underlying surfaces, making it difficult to depict the high-concentration zonal distribution of road traffic emissions and the pollution accumulation effect within street valleys. Therefore, mesoscale models are insufficient to meet the needs for refined assessment of pollution distribution and population exposure at the block scale.

[0003] To address the need for street-scale pollution simulation, downscaling methods centered on mechanistic models have gradually developed in recent years, mainly including two categories: parametric models and computational fluid dynamics (CFD) models. Parametric models are typically based on Gaussian plume or smoke cloud theory, abstractly modeling pollution source emission characteristics, urban underlying surface, and local meteorological conditions, and incorporating street geometry to simplify the calculation of pollutant turbulent diffusion. Their main functional modules include: pollution source emission representation, meteorological element parameterization (wind speed, wind direction, turbulence characteristics), pollutant diffusion and deposition calculation, and street geometric effects. These models have high computational efficiency and lower requirements for input data accuracy, thus they are widely used for near-road ambient air quality assessment and urban environmental monitoring. In contrast, CFD models use high-resolution gridded modeling of urban geometry and street valley airflow, iteratively solving for pollutant transport processes based on fluid dynamics equations. CFD models can accurately reproduce the effects of building flow, local turbulence characteristics, and pollutant transport and reaction processes. They can not only simulate the diffusion process of primary pollutants, but also couple atmospheric chemical mechanisms to study the reaction characteristics of reactive pollutants in street canyons.

[0004] Although existing street-scale air pollution models have made some progress and have been applied to some research and management scenarios, current research still has many shortcomings, mainly in the following aspects: (1) Insufficient description of complex microenvironments and dynamic mechanisms. Parametric models, based on Gaussian plume or smoke cloud theory, simplify the local dynamic field by abstracting the emission from pollution sources and street geometry, and can quickly obtain relatively accurate concentration estimates. However, this simplification leads to a lack of accurate description of the local eddies around buildings and the resulting differences in pollution transport. In contrast, although CFD models can simulate the dynamic field of street valley microenvironments and pollutant transport processes with high accuracy, they have extremely high computational resource requirements, making them difficult to apply in large-scale urban areas, thus limiting their practicality and generalizability.

[0005] (2) The systematic and mechanistic descriptions of atmospheric processes are incomplete. Existing street-scale downscaling models still have significant shortcomings in the systematic description of atmospheric processes. Most models mainly focus on the emission, diffusion, and partial chemical transformation of primary gaseous pollutants, failing to systematically integrate various important atmospheric processes affecting the distribution of pollution in street blocks, such as the transport, secondary formation, dry and wet deposition, and resuspension of pollutants in street valleys. This results in an incomplete description of the overall mechanism of pollutant concentration evolution. In particular, in terms of particulate matter deposition, existing parameterization schemes generally have defects, often ignoring the interception effect of underlying vegetation on particulate matter. This may lead to an underestimation of the deposition efficiency of particles around 1 μm by about an order of magnitude, while this mode is less accurate in PM2.5. 2.5The proportion of particulate matter in the mass is as high as 40% or more. In addition, even though the model has considered the dry deposition process, it has not fully incorporated the impact of the dynamic characteristics of the urban environment, which are significantly different from those of the free atmosphere, on the deposition efficiency, so that the simulation of particulate matter distribution at the urban scale still has biases.

[0006] (3) Lack of feedback mechanisms between pollutants and environmental factors. Most downscaling models still focus on single or gaseous pollutants, lacking the ability to systematically simulate the interaction between pollutants and environmental factors. For example, PM2.5... 2.5 The optical properties of pollutants can affect surface photochemical flux, thereby altering the photochemical reaction rate of gaseous pollutants in the valley. However, existing models generally do not incorporate this feedback mechanism, resulting in an incomplete description of the pollutant generation and transformation process.

[0007] (4) Limited computational efficiency and application scope. Although high-precision CFD models can accurately simulate local pollution transport, each calculation is time-consuming and requires huge computational resources. For example, simulating urban airflow in a 33×33 km² area of ​​a large city requires 300 supercomputer nodes to perform 16 hours of calculation, which limits its feasibility in simulating large-scale urban street networks. Parametric models are more efficient in large-scale applications, but their accuracy is limited; and research on the coupling of the two is still in its early stages, lacking mature and generalizable methods.

[0008] (5) Lack of source tracing capability for pollution in urban core areas. In complex street environments, there are many types of potential pollution sources, including traffic sources, regional transport and other area and point sources. Moreover, the interaction between pollution at multiple scales has not been fully considered (such as the contribution of background atmosphere to the canyon of the street), making it difficult for existing models to distinguish the contributions of various pollution sources and limiting their ability to support the identification and contribution assessment of pollution sources at the street scale. Summary of the Invention

[0009] In view of this, this application provides a method and system for simulating and tracing atmospheric pollution at the street scale to solve at least one of the aforementioned problems.

[0010] To achieve the above objectives, this application adopts the following approach: According to a first aspect of this application, a method for simulating street-scale air pollution is provided, the method comprising: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is used to obtain the street-scale dynamic field. Interpret street-scale emission data into pollution source characteristic information; Based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and pollution source characteristic information, collaborative processing is performed to obtain street-scale pollution distribution data. Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process. Output the pollution distribution data at the street scale.

[0011] In some embodiments of this application, the above method obtains the street-scale dynamic field based on mesoscale meteorological data using a comprehensive simulation strategy that fuses machine learning models and parameterized models, including: Background meteorological parameters were obtained based on the aforementioned mesoscale meteorological data; Based on the background meteorological parameters and the corresponding street morphology parameters, the dynamic field parameters are calculated using a pre-trained machine learning model. When the background meteorological parameters and corresponding street morphology parameters exceed the applicable range of the machine learning model, the dynamic field parameters are calculated using a parameterized multivariate adaptive regression splines (MARS) model. The block-scale dynamic field is obtained based on the dynamic field parameters.

[0012] In some embodiments of this application, the above method further includes the following steps in obtaining the block-scale dynamic field based on the dynamic field parameters: The local surface roughness length of the urban underlying surface is corrected by combining the urban background roughness and the increments caused by local street valleys and building morphology; and In light of the urban heat island effect, the local sensible heat flux and the height of the mixing layer formed by thermal turbulence on the urban underlying surface are modified.

[0013] In some embodiments of this application, the above method performs collaborative processing based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and the pollution source characteristic information to obtain street-scale pollution distribution data, including: In simulating a single diffusion of pollutants, based on the simulated grid data and the pollution source characteristic information, the concentration contribution of each individual emission source to a given receptor point is calculated using the Gaussian diffusion formula. In simulating the secondary generation of pollutants, NO and NO2 are treated as inert gases and their diffusion is simulated to obtain the primary concentration. After assuming that light steady state is reached, the concentrations of NO, NO2, and O3 are calculated. In the simulated background concentration mixing, the background concentration in the mesoscale pollution distribution and sectoral contribution data is deducted from the contribution of street-scale pollution sources, and then multiplied by the background concentration mixing ratio to obtain the background concentration at a specified height near the ground. In simulating dry particulate matter deposition, based on the aforementioned street-scale dynamic field, the gravitational settling velocity of the particulate matter is utilized. 、 The dry settling rate of particulate matter is calculated from aerodynamic drag and surface layer drag. In simulated wet particulate matter deposition, the cloud clearance coefficient, which characterizes wet deposition, is calculated using a semi-empirical formula based on the mass median diameter and the precipitation data. Street-scale pollution distribution data are generated based on the results of simulating primary pollutant diffusion, secondary pollutant generation, background concentration mixing, dry particulate matter deposition, and wet particulate matter deposition.

[0014] In some embodiments of this application, the above method generates a correction signal based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements to dynamically adjust the collaborative processing process, including: Using PM 2.5 The concentration C is used to correct the total solar radiation within the valley area, thereby further correcting the change in photochemical reaction rate. The total solar radiation correction term is: ; In the above formula C is the total solar radiation correction term, where C represents PM2.5. 2.5 The concentration of RH is the humidity.

[0015] According to a second aspect of this application, a method for tracing the source of air pollution at the street scale, the method comprising: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is used to obtain the street-scale dynamic field. Interpret street-scale emission data into pollution source characteristic information; Based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and pollution source characteristic information, collaborative processing is performed to obtain street-scale pollution distribution data. Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process. Output the pollution distribution data at the street scale; A multi-source quantitative source tracing analysis was performed on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations within the street.

[0016] In some embodiments of this application, the above method performs multi-source quantitative source tracing analysis on the street-scale pollution distribution data, including: A traversal algorithm is used to sum the individual pollutants affecting each receptor site, and the pollution contribution of each individual pollutant to the receptor site is obtained in sequence. By specifying the grouping field in the input file, the pollution contribution of each group of pollutants to the specified receptor site is quantified, thereby realizing source tracing analysis.

[0017] According to a third aspect of this application, a street-scale air pollution simulation system is provided, the system comprising: an input data acquisition module, a dynamic field module, a pollution source module, an atmospheric physicochemical process module, a pollutant-environment interaction feedback module, and an output result module, wherein: The input data acquisition module is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data. The dynamic field module is used to receive mesoscale meteorological data from the input data acquisition module and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data. The pollution source module is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module. The atmospheric physical and chemical process module is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data. The pollutant-environment interaction feedback module is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module. The output results module is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module.

[0018] According to a fourth aspect of this application, a street-scale air pollution source tracing system is provided, the system comprising: an input data acquisition module, a dynamic field module, a pollution source module, an atmospheric physicochemical process module, a pollutant-environment interaction feedback module, a source tracing module, and an output result module, wherein: The input data acquisition module is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data. The dynamic field module is used to receive mesoscale meteorological data from the input data acquisition module and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data. The pollution source module is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module. The atmospheric physical and chemical process module is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data. The pollutant-environment interaction feedback module is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module. The source tracing module is used to perform multi-source quantitative source tracing analysis on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations in the street. The output results module is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module and the source tracing analysis results generated by the source tracing module.

[0019] According to a fifth aspect of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0020] According to a sixth aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the above-described method.

[0021] As can be seen from the above technical solutions, the street-scale air pollution simulation and source tracing method and system provided in this application adopts a comprehensive simulation strategy that integrates machine learning and parametric models. While ensuring high-precision simulation and wide-area coverage, it can reduce computational resource consumption and single simulation time, enabling operable and scalable applications across large street networks. Furthermore, this application also constructs a model framework that reflects the interaction between pollutants and the environment, possessing the ability to simulate the interaction between pollutants and environmental factors, thereby improving the realism of street-scale pollution simulation. Attached Figure Description

[0022] 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 of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating a street-scale air pollution simulation method provided in an embodiment of this application; Figure 2 This embodiment provides a schematic diagram of the process for obtaining the street-scale dynamic field using a comprehensive simulation strategy that integrates machine learning models and parameterized models. Figure 3 This is a schematic diagram of the process for obtaining street-scale pollution distribution data provided in an embodiment of this application; Figure 4 The PM provided in the embodiments of this application 2.5 Figure 1 shows the simulation results of average hourly concentrations of NO2 and O3. Figure 5 This is a flowchart illustrating a method for tracing the source of air pollution at the street scale, as provided in an embodiment of this application. Figure 6 This is a schematic diagram of the structure of a street-scale air pollution simulation system provided in an embodiment of this application; Figure 7 This is a schematic diagram of the module interaction of a street-scale air pollution simulation and source tracing system provided in an embodiment of this application; Figure 8 This is a schematic block diagram of the system configuration of the electronic device provided in the embodiments of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments and descriptions of this application are used to explain this application, but are not intended to limit this application.

[0024] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0025] Furthermore, it should be noted that the use of terms such as "first" and "second" to define related concepts is merely for the purpose of distinguishing the corresponding concepts. Unless otherwise stated, these terms have no special meaning and therefore should not be construed as limiting the scope of protection of this application. In addition, although the terminology used in this application is selected from commonly known and used terms, some terms mentioned in this application's specification may have been chosen by the applicant according to his or her judgment, and their detailed meanings are explained in the relevant sections of this description. Moreover, this application should be understood not only through the actual terms used, but also through the meaning implied by each term.

[0026] like Figure 1 The diagram shown is a flowchart of a street-scale air pollution simulation method provided in an embodiment of this application. The method includes the following steps: Step S101: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is adopted to obtain the street-scale dynamic field.

[0027] The mesoscale meteorological data here refers to the output of mesoscale meteorological models, which drive the calculation of the street-scale dynamic field. Specifically, it can include background wind, temperature, and humidity fields from the mesoscale meteorological model, as well as parameters such as sensible heat flux output by the model, reflecting the influence of the urban underlying surface and street geometry on local meteorology. This mesoscale meteorological data serves to provide boundary and background conditions for the street-scale dynamic field and, combined with street morphological characteristics, locally corrects the near-surface wind field, thus obtaining the street-scale dynamic field distribution. For example, this scale meteorological data can come from a 1 km resolution simulation result of WRF.

[0028] In this embodiment, in order to balance accuracy and efficiency, a comprehensive simulation strategy that integrates machine learning models and parameterized models is adopted. This strategy first conducts CFD simulation based on typical urban street morphology, and then performs machine learning and multivariate adaptive regression spline (MARS) analysis on the simulation results and the corresponding street morphology parameters and background meteorological parameters, respectively. In other words, the black box algorithm is combined with the MARS model.

[0029] Step S102: Parse the street-scale emission data into pollution source characteristic information.

[0030] This step is mainly to obtain the spatial distribution and emission intensity of key pollution sources (such as traffic sources) within the block area, so as to provide accurate input for the physicochemical simulation in subsequent steps.

[0031] Step S103: Based on the simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and the pollution source characteristic information, perform collaborative processing to obtain street-scale pollution distribution data.

[0032] This step involves co-processing various external data and the data generated in the preceding steps to obtain pollution distribution data at the street scale. This co-processing includes simulating the primary diffusion, secondary generation, background concentration mixing, and dry and wet deposition of pollutants, thereby achieving the physical and chemical coupling of the entire pollution process in the street.

[0033] Step S104: Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process.

[0034] Step S105: Output the pollution distribution data at the street scale.

[0035] The street-scale air pollution simulation method provided in this application employs a comprehensive simulation strategy that integrates machine learning and parametric models. While ensuring high-precision simulation and wide-area coverage, it reduces computational resource consumption and simulation time per run, enabling operable and scalable applications across large street networks. Furthermore, this application constructs a model framework that reflects the interaction between pollutants and the environment, possessing the ability to simulate the interaction between pollutants and environmental factors, thereby improving the realism of street-scale pollution simulations.

[0036] In some embodiments of this application, such as Figure 2 As shown, in step S101 above, the block-scale dynamic field is obtained based on mesoscale meteorological data using a comprehensive simulation strategy that integrates machine learning models and parameterized models, including: Step S1011: Obtain background meteorological parameters based on the mesoscale meteorological data.

[0037] This step involves extracting and organizing background meteorological parameters from the mesoscale meteorological data output by the mesoscale model. These parameters will be used as boundary conditions and background inputs in subsequent street-scale calculations.

[0038] Step S1012: Based on the background meteorological parameters and the corresponding street morphology parameters, calculate the dynamic field parameters using a pre-trained machine learning model.

[0039] The dynamic field parameters in this step mainly refer to the V of the local dynamic field. xWith V y These two core variables. Given background weather and street morphology, the key parameters of the street-scale dynamic field are first directly output using a trained machine learning model, specifically including V. x V y Distribution of isovelocity components or local velocity field in a street-scale grid.

[0040] The machine learning model takes into account background meteorological parameters and street morphology parameters, such as the background wind vector along the X and Y directions at the top H of the street valley, the height-to-width ratio (H / W), length-to-width ratio (L / W), and height symmetry ratio (Hl / Hr) of the street valley, and outputs the local horizontal wind field component V. x V y .

[0041] The training process of the machine learning model in this step is as follows: Based on a pre-constructed typical street valley flow field dataset generated using CFD simulation, the mean values ​​V of the X and Y velocity components at different heights z within the street valley are first established using Random Forest (RF). x and V y The nonlinear relationship between the street canyon geometry and the background wind field. The predictor variables include the background wind vector Vbg at the top H of the street canyon along the X and Y directions. x With Vbg y .

[0042] The construction process of a random forest is as follows: (1) From the original training dataset, approximately two-thirds of the data are randomly sampled with replacement using a bootstrap method to form n. tree A new sample set, and thus form n tree A classification regression tree, where each unselected sample forms an n-fold regression tree. tree One out-of-bag (OOB) dataset is used as the test set for a single decision tree to evaluate the prediction error.

[0043] (2) During the growth of a single decision tree, m features are randomly selected from M features at each node. try Features (m) try ≤M), according to the principle of minimizing node impurity, from this m try One feature is selected from the given features for node splitting and growth. Each tree grows to its maximum potential without any pruning.

[0044] (3) All the generated decision trees are combined into a random forest. The final prediction value of the random forest is obtained by taking the average of the prediction results of each tree (for regression problems) or by taking the majority vote results (for classification problems).

[0045] In the above process, m is selected try =6 and n tree =200 to achieve the minimum out-of-bag prediction error.

[0046] Step S1013: When the background meteorological parameters and the corresponding street morphology parameters exceed the applicable range of the machine learning model, the dynamic field parameters are calculated using the parameterized MARS model.

[0047] When the input features of a machine learning model fall outside the training data, exceed the model's extrapolation capabilities, or when abnormal or missing inputs render the machine learning model's output unusable, this embodiment can utilize a parameterized MARS model to calculate the dynamic field parameters. This embodiment uses a MARS-based parameterized model to estimate the dynamic field parameters as a fallback solution for the machine learning model, ensuring a stable dynamic field output even under extreme or unpredictable conditions.

[0048] The black-box algorithm obtained from machine learning in step S1012 above cannot be specifically described using mathematical language and can only be embedded in the model as a module. The MARS model in this step is based on adaptive piecewise linear combination and basis functions, and its mathematical description is shown in equations (1) and (2) below: (1) (2) in: β 0 For constant parameters, x For predictor variables, h i (x) For the first i basis functions β i For the first i The weight coefficients of each basis function n This represents the total number of basis functions. These basis functions can be composed of spline function products of one or more different predictor variables. For the first i The first basis function k A spline function, The value can be 1 or -1. This represents the threshold of the predictor variable in the spline function. M i For the first i The total number of spline functions in the basis functions.

[0049] The process of establishing the MARS model is as follows: (1) Forward selection: The original training dataset is segmented into regions, and nodes are selected. Each selected node will divide a small region and generate a pair of basis functions to be added to the model. After traversing each predictor variable, a large number of basis function pairs will be generated to improve the accuracy of the model.

[0050] (2) Backward pruning: Since the forward selection process will eventually generate too many basis functions, which can easily lead to overfitting of the model, backward pruning mainly removes redundant basis functions that contribute little to the model, reduces the complexity of the model, and finds the optimal model. This process is mainly based on the Generalized Cross Validation (GCV) criterion, as shown in equation (3): (3) In the formula, MSE is the mean squared error, H is the number of basis functions, N is the total number of training datasets, and d is the penalty coefficient, which is 3.

[0051] When building a MARS model, two hyperparameters need to be set appropriately to improve prediction accuracy: the total number of terms (nprune=n+1) and the maximum number of interactions allowed in a single basis function (degree=max(M). i Therefore, this study tested the impact of perturbations on these two parameters at different levels on simulation bias. The optimal hyperparameter combination was obtained using ten-fold cross-validation (CV). This involved randomly dividing the entire dataset into ten equal groups, using nine groups as the training set for modeling, and the remaining group as the test set to evaluate prediction bias. After iterating through all ten groups, ten validation results were obtained, and the average of these results was taken as the bias of the final prediction result. Regarding V... x and V y In the construction of prediction models, the combination of nprune=23 and degree=3 can be used to keep the RMSE of cross-validation at the lowest level and achieve optimal simulation performance.

[0052] Table 1 below shows V x The MARS model equations and coefficients in the fitting are shown in Table 2. y The equations and coefficients of the fitted MARS model.

[0053] Table 1

[0054] Table 2

[0055] Step S1014: Obtain the block-scale dynamic field based on the dynamic field parameters.

[0056] This step involves pooling the dynamic field parameters obtained from S1012 or S1013 to form the final block-scale dynamic field distribution, which serves as the boundary conditions and inputs for subsequent processes such as pollution transport and chemical reactions.

[0057] In a further preferred embodiment, the method of this application further includes, in the process of obtaining the block-scale dynamic field based on the dynamic field parameters, correcting the local surface roughness length of the urban underlying surface by combining the urban background roughness and the increment caused by the influence of local street valleys and building morphology; and correcting the local sensible heat flux and the height of the mixed layer formed by thermal turbulence of the urban underlying surface by combining the urban heat island effect.

[0058] The following is an explanation of the above parameter corrections: First, this application considers the severe dynamic impact of the complex underlying surface and heat island effect within the city on the near-surface meteorological field, and incorporates urban background roughness ( z 0,bg The increase in local surface roughness length is influenced by local street valleys and building morphology. z 0 has been corrected as shown in (4)-(6) below: (4) (5) (6) in, z 0 represents the local surface roughness length of the road; z 0,bg The surface roughness length for urban areas is taken as 1 m. α rr and bd r These represent the street valley morphology and building density near the road, respectively. z The effect of 0; α r The aspect ratio of the street valley corresponding to the road H / W bh, bhsd, and bd are the average height, standard deviation of height, and planar density of buildings near the road, respectively. C This is an empirical coefficient, with a value of 0.05.

[0059] Secondly, this application also considers the urban heat island effect to measure the local sensible heat flux of the urban underlying surface. H u,UHI (W / m) 2 The height of the mixed layer formed by thermal turbulence and ) Z mix,c (m) was calculated, as shown in equation (7) below: (7) in, α This is an empirical coefficient, with a value of 0.03. ρ air density; c p The isobaric specific heat capacity of air is taken as 1004 J / kg·K; Δ T u-r The temperature difference between urban and suburban areas; Z mix,ref and P ref The boundary layer height and the city population, used as references, are 400 m and 2 million people, respectively. P The total population of urban areas within the study region is given. Based on this, in the process of obtaining the block-scale dynamic field based on the aforementioned dynamic field parameters, this application also recalculates the atmospheric turbulence parameters, including the zero-level displacement height, using methods from AERMOD. d h ), Surface friction speed ( ), convection velocity scale ( ), Mixing layer height ( Z mix ) and Mönning-Obukhov length ( L MO (And other parameters.)

[0060] In some embodiments of this application, such as Figure 3 As shown, in step S103 above, based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and the pollution source characteristic information, collaborative processing is performed to obtain street-scale pollution distribution data, including: Step S1031: In the simulated primary diffusion of pollutants, based on the simulated grid data and the pollution source characteristic information, and using the Gaussian diffusion formula, calculate the concentration contribution of each individual emission source to a given receptor point.

[0061] This step, in simulating a single diffusion event, is based on the Gaussian diffusion formula, calculating the concentration contribution of each individual emission source to a given receiver point. For example, for a line source, the emission source is mathematically divided into several point sources, and the horizontal Gaussian plume is calculated. and meandering The diffusion process is then performed, and the ratios of the two factors are adjusted according to a proportional parameter to calculate the emission diffusion contribution of the point source to the target receptor point. Then, the trapezoidal integral method is used to transform the point source emission diffusion results into the line source emission diffusion results, as shown in equations (8)-(11) below: (8) (9) (10) (11) in, y It is the offset distance of the receptor point in the crosswind direction; It is the lateral diffusion parameter; z It is the receptor point height; H It is the point source release height; It is the vertical diffusion parameter; u The effective downwind wind speed (V) x and V y (synthetic wind speed) It is a proportional parameter; R It is the distance from the point source to the receptor point; Q It is the strength of the emission source.

[0062] Step S1032: In the simulation of secondary generation of pollutants, NO and NO2 are used as inert gases to simulate diffusion and obtain primary concentrations. After assuming that light steady state is reached, the concentrations of NO, NO2 and O3 are calculated.

[0063] This application employs a simplified NO2-NO-O3 two-reaction method to characterize NO. x The photochemical process that generates O3. During operation, NO is first... x As an inert gas, only the concentration after its diffusion process was simulated. Then, assuming that a steady-state equilibrium was reached, the concentrations of NO, NO2, and O3 were calculated using the "two-reaction" method. The basic chemical principle is shown in the following equation (12): (12) Where: square brackets represent the molar concentration of pollutants (mol / m³) 3 ), [NO x ] d For the model, NO x The concentration of [NO] is directly simulated as an inert gas. b [NO2] b and [O3] b The background concentrations of NO, NO2, and O3 are respectively, and the non-traffic source contribution concentrations are provided from the mesoscale model results. NO2 and NO in traffic source emissions x The ratio is 0.2; NO2 photolysis rate constant. k Rate constants of oxidation reactions of 1 and O3 k The calculation of 2 is shown in the following formulas (13)-(14): (13) (14) Where: TSR is the total solar radiation (W / m²) 2 Characterized using shortwave radiation flux reaching the Earth's surface; Z The solar zenith angle (°); T The ambient temperature (K) is the ambient temperature.

[0064] Step S1033: In the simulated background concentration mixing, the background concentration in the mesoscale pollution distribution and sectoral contribution data is deducted from the contribution of the street-scale pollution source, and then multiplied by the background concentration mixing ratio to obtain the background concentration at a specified height near the ground.

[0065] In this application, the background concentration at a specified height near the ground is determined by the background concentration provided by the mesoscale model and the background concentration mixing ratio (fac). bg The results are obtained by multiplying the two. Since specific source emissions are modeled at both the mesoscale and block scales, to avoid secondary calculations, the contribution of the corresponding source needs to be subtracted from the mesoscale simulation results before inputting them. For example, if the emission source at the block scale is traffic emission, the contribution concentration of the traffic sector needs to be subtracted from the mesoscale simulation results before it can be used as the background concentration input into the block model.

[0066] This application uses the ratio of near-ground wind speed to the wind speed at the top of surrounding buildings (WS). sfc / WS bh The fac parameter is used as a substitute parameter to primarily calculate the fac. bg fac bg The calculations are based on sensible heat fluxes estimated from mesoscale meteorological models. H u (W / m) 2 ) distinguished H u >0 and H u The case of <0 is also considered, and the influence of the building density around the receiver point is also taken into account, as shown in (15): (15) in, F = m +abs(0.25-bd), where m This is an empirical parameter with a value of 0.1.

[0067] Step S1034: In the simulated dry sedimentation of particulate matter, based on the street-scale dynamic field, the dry sedimentation rate of particulate matter is calculated using the gravity sedimentation velocity, aerodynamic drag, and surface layer drag of the particulate matter.

[0068] In this application, the dry sedimentation rate of particulate matter is used as... V d It indicates that,V d The modeling divides it into several force-bearing components, including gravity, which provides the main force source for settlement (represented as gravitational settlement velocity). V g ), and aerodynamic drag ( R a ) and surface layer resistance ( R s The interference caused by particulate matter sedimentation. Specifically, the following formulas (16)-(19) are used for calculation: (16) (17) (18) (19) in: ρ p The density of particulate matter is typically specified as 1000 kg / m³. 3 ; D p The aerodynamic equivalent diameter of the particulate matter; g It is the acceleration due to gravity; C Cunningham's correction factor is used to correct for the Stokes law, which is derived under the assumption of no slip but no longer applies at high Knudsen numbers. μ The viscosity coefficient of air can be expressed as a function of temperature. T The function; hn It is the dimensionless temperature distribution constant under neutral conditions, taken as 0.74; z It is to calculate the dry deposition rate. V d Height; z 0 represents the surface roughness length; Ψ H It is a thermal stability correction function used to measure the degree of atmospheric thermodynamic stability; κ It is the von Kármán constant (≈ 0.40); It is the friction speed; E B For Brownian diffusion terms; E IM For inertial impact terms; E IN For the intercepted effect term; ε 0 is an empirical constant, with a value of 3; R 1 represents the resuspension correction factor.

[0069] The non-constant parameters mentioned in the above formula can be calculated using the following formula: (i)Coefficient of air viscosity μ can be calculated using Sutherland's law. This law is based on an idealized intermolecular potential, and its value at temperature T can be calculated according to the following formula (20): (20) Where: μ 0 is the coefficient of air viscosity at the standard temperature T 0 = 273.15 K, taking the value of 1.716×10 -5 N·s / m 2 ; S μ is the effective temperature called the Sutherland constant, with a value of 111 K.

[0070] (ii)Correction factor C can be calculated according to the following formula (21): (21) Where: λ is the mean free path of air molecules, which is usually a function of temperature, pressure, and kinematic viscosity of air.

[0071] (iii)For air molecules λ can be calculated using the following formula (22): (22) Where: k B is the Boltzmann constant; v is the kinematic viscosity; p is the air pressure; R specfic is the specific gas constant, taking the value of 287 J / (kg·K).

[0072] (iv)Kinematic viscosity v can be calculated by taking the ratio of the air viscosity μ to the current air density, using the following formulas (23)-(24): (23) (24) Where: ρ air is the density of air at temperature T ; p is the air pressure; R specfic is the specific gas constant.

[0073] (v)Thermal stability correction function Ψ H It can be expressed as the following formula (25): (25) in: L 0 represents the Monin-Obukhov length, used to describe the effect of buoyancy on turbulence, particularly in the near-surface region, about one-tenth of the way down from the atmospheric boundary layer.

[0074] (vi) L The formulas for calculating 0 are shown in equations (26)-(27): (26) (27) in: c p The specific heat capacity of air at constant pressure is 1004 J / (kg·K); H For atmospheric thermal flux; α The empirical coefficient is 0.03. ρ air density; T u-r This refers to the temperature difference between urban and suburban areas.

[0075] (vii) The Stokes number (St) and the Schmidt number (Sc) can be calculated using the following formulas (28)-(30): (28) (29) (30) (viii) Resuspension correction factor R 1 can be calculated using the following formula (31): (31) (ix) The efficiency term of Brownian diffusion ( E B The following formula (32) is used for calculation: (32) in: C b This is an empirical parameter, set to 1; γ The value of is determined by the properties of the underlying surface; for smooth underlying surfaces, it is 2 / 3, and for rough underlying surfaces, it is 1 / 2.

[0076] (x) Efficiency term of inertial impact ( EIM The following formula (33) is used for calculation: (33) in: β This is an empirical parameter, so we set it to 0.8. C Im , α The values ​​of and are usually determined by the properties of the underlying surface, as shown in Table 3 below: Table 3

[0077] (xi) Efficiency term of the interception effect ( E IN The efficiency caused by interception is expressed as being only related to St: (34) in: C In , β These are empirical parameters. β Take 0.8, C In Take 2.5, A This refers to the radius or equivalent circle radius of the particulate trap on the underlying surface. Typically, in modeling, vegetation leaves are considered as particulate traps, therefore... A The corresponding parameter is the leaf's ability to trap particulate matter. A The value needs to be provided by an additional input file.

[0078] In the above V d Based on the modeling, the particulate matter sedimentation reference adopts the core idea of ​​mass conservation, and the PM before and after sedimentation... 2.5 The total mass concentration remains unchanged, that is: (35) in: H The effective atmospheric height (m) for settlement. Based on the background concentration mixing scheme, this model uses the equivalent settlement height calculation method, and adjusts the background concentration mixing ratio determined by the street valley morphology. H The value of is shown in the following formula (36): (36) Where cyH is the height of the street valley, PBLH is the height of the atmospheric boundary layer, and frapg is the mixing ratio of the background concentration.

[0079] Step S1035: In the simulated wet particulate matter deposition, based on the mass median diameter and the precipitation data, the cloud clearance coefficient characterizing wet deposition is calculated using a semi-empirical formula.

[0080] In this application, the wet deposition of particulate matter is calculated using a semi-empirical formula, with the cloud cover removal coefficient as the basis. λ (s) -1 ) as a representation. λ and D p (m) exhibits a logarithmic high-power polynomial relationship and is related to rainfall. R (mm / h) is related to the following formulas (37)-(38): (37) (38) Where a, b, c, d, e, and f are all empirical constants, a = 274.35, b = 332839, c = 226656, d = 58005, e = 6588, and f = 0.245.

[0081] Regardless of whether it is dry or wet settlement, this application will include PM 2.5 The settling rate was calculated for each of the three particle size ranges. (Background transmission PM) 2.5 The mass percentage and mass median diameter were extracted from the mesoscale simulation results. Local PM emissions 2.5 The mass percentage is specified in the emission document, and the mass median is uniformly specified for the model. The mass median of the Aigen core model, aggregation model, and coarse particle model (i.e., in the formula above) D p The thicknesses are 0.06 μm, 0.5 μm, and 1.8 μm, respectively.

[0082] Step S1036: Generate street-scale pollution distribution data based on the results of simulated primary pollutant diffusion, secondary pollutant generation, background concentration mixing, dry particulate matter deposition, and wet particulate matter deposition.

[0083] This step combines the contributions of gas phase and particulate matter and the background concentration obtained in steps S1031-S1035 to obtain the street-scale pollution distribution data of the target grid at a specific time step. It can also output the sub-contributions and multi-dimensional results required for visualization as needed.

[0084] Since the radiative forcing of particulate matter is mainly manifested in the unidirectional reaction of NO2 to O3 catalyzed by light in the NO2-NO-O3 ternary reaction chain, and this reaction adopts the photolysis rate constant based on total solar radiation (TSR). k 1. Perform steady-state calculations. Therefore, in some embodiments of this application, step S104 above generates a correction signal based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements to dynamically adjust the collaborative processing process, including: Using PM 2.5The concentration C is used to correct the total solar radiation within the valley area, thereby further correcting the change in photochemical reaction rate. The total solar radiation correction term is: (39) In the above formula C is the total solar radiation correction term, where C represents PM2.5. 2.5 The concentration of RH is the humidity.

[0085] This application can support on / off settings, allowing for flexible inclusion or exclusion of secondary feedback effects based on research objectives.

[0086] The above method will be further illustrated below through a specific embodiment: This embodiment uses the area within the Fourth Ring Road of XX City in August 20XX as an example to conduct PM2.5 analysis. 2.5 Case simulation of NO2 and O3 concentrations.

[0087] The mesoscale atmospheric pollution distribution simulation results and sectoral source contributions are derived from the 1 km resolution simulation results of CMAQ-ISAM; the mesoscale meteorological simulation results are derived from the 1 km resolution simulation results of WRF; the street-scale emission data are from the urban motor vehicle road level emission model (SLOVE); the precipitation data are from observations of the National Climate Data Center (NCDC) of Country X; the simulation grid is a uniform grid with a 50 m resolution drawn using ArcGIS. During runtime, all the aforementioned functional modules were enabled, and 24 programs were executed simultaneously using csh scripts, with each program utilizing 6 cores for parallel computation. The simulation time was approximately 72 hours, consuming a total of 10368 cores per hour.

[0088] In August 20XX, the PM2.5 concentration in the core urban area of ​​XX City was... 2.5 The average concentrations of NO2 and O3 were 23 μg / m³. 3 32 μg / m 3 69 μg / m 3 The simulation results are as follows Figure 4 As shown. Source tracing analysis indicates that PM2.5 levels in the core urban area of ​​XX city... 2.5 Traffic emissions contribute an average of 1.7% and 39% to NO2 emissions, respectively, with peak contributions reaching 23% and 75%. For example, for a receptor point B within a 200-meter radius of road segment A, the PM2.5 concentration at road segment A... 2.5 NO2 emissions contributed 7.5% and 52% to the pollution at point B, respectively.

[0089] By comparing the data with observation data from national monitoring stations, this application provides information on PM2.5. 2.5The performance indicators R for NO2 and O3 are 0.89, 0.52 and 0.78, respectively. Compared with the CMAQ model widely used in the industry for pollution simulation in XX City, the performance indicators are improved by 54%, 104% and 114%, respectively, showing better simulation performance.

[0090] As can be seen from the above, the street-scale air pollution simulation method proposed in this application has the following beneficial effects: (1) This application systematically integrates various important atmospheric processes that affect the distribution of pollution in urban areas, such as the transport, secondary formation, dry and wet deposition, and resuspension of pollutants in the street valley, thus solving the obvious shortcomings of existing street-scale downscaling models in terms of the systematic description of atmospheric processes. In particular, in terms of particulate matter deposition, this application can accurately characterize the dry and wet deposition and resuspension mechanisms of particulate matter, improving the accuracy and reliability of urban particulate matter concentration simulation.

[0091] (2) The verification results with observational data show that, compared with traditional models, this application has higher accuracy in simulating pollutants at the street scale, better spatiotemporal representativeness, and robust model performance in terms of overall simulation effect. For modeling some atmospheric process mechanisms, the constructed parameterization scheme can significantly reduce the description error of traditional models and has higher reliability.

[0092] (3) This application adopts a comprehensive simulation strategy that integrates machine learning and parametric models. While ensuring high-precision simulation and wide-area coverage, it significantly reduces the consumption of computing resources and the time for a single simulation, enabling operable and scalable applications within a large urban network. It solves the problems of long single-computation time and huge computing resources required for high-precision CFD models, as well as the limitations of accuracy of parametric models in large-scale applications despite their high efficiency.

[0093] (4) This application constructs a model framework that can reflect the interaction between pollutants and the environment, and has the ability to simulate the interaction between pollutants and environmental factors, so as to improve the authenticity of pollution simulation and source tracing results at the street scale.

[0094] like Figure 5 The diagram shown is a flowchart of a street-scale air pollution source tracing method provided in an embodiment of this application. The method includes the following steps: Step S501: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is adopted to obtain the street-scale dynamic field.

[0095] Step S502: Parse the street-scale emission data into pollution source characteristic information.

[0096] Step S503: Based on the simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and the pollution source characteristic information, perform collaborative processing to obtain street-scale pollution distribution data.

[0097] Step S504: Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process.

[0098] Step S505: Output the pollution distribution data at the street scale.

[0099] Step S506: Perform multi-source quantitative source tracing analysis on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations in the street.

[0100] The above steps S501-S505 can be found in the description of steps S101-S105 above, and will not be repeated here.

[0101] In this embodiment, step S506 can provide a numerical decomposition of the contribution of different pollution sources in a block to a specific location, revealing the relative importance of traffic emissions, regional transport, background pollution, etc. in local pollution, and providing a basis for precise governance and health risk assessment.

[0102] Preferably, step S506 may further include: summing the monomeric pollution sources affecting each receptor site using a traversal algorithm, obtaining the pollution contribution of each monomeric pollution source to the receptor site in sequence, and quantifying the pollution contribution of each group of pollution sources to the specified receptor site by specifying grouping in the grouping field of the input file, thereby realizing source tracing analysis.

[0103] This step employs a traversal algorithm to progressively sum the contributions of individual pollution sources to each receptor site, ensuring that the sum of contributions is consistent with the background transport contribution and contributions from other local sources. This facilitates the identification of key pollution control areas and decision support. The input data for this step includes, but is not limited to: street-scale pollution distribution data, the location and emission intensity of each individual pollution source, grouping information, and the grid coordinates of the receptor sites. Then, the input individual pollution sources are traversed one by one (or traversed according to the grouping structure), and while keeping other source inputs constant, the contribution component of each individual source at each receptor site is calculated.

[0104] This step allows you to specify the grouping granularity (e.g., by traffic source, regional transmission, other sources, etc.) in the input grouping field. If the grouping granularity is a single pollution source, it can output the contribution value of the single source to each receptor site and its assigned group label, thereby achieving source tracing analysis.

[0105] This application, based on the realization of refined pollution simulation at the street scale, can quantitatively assess the contribution of a single pollution source to street air quality, filling the gap in existing ultra-high resolution source tracing models. This application can systematically integrate source tracing analysis methods involving multiple pollution sources, multi-scale transport, and the effects of the street microenvironment to achieve quantitative assessment of pollution contributions at the street scale. It solves the problem that existing models struggle to distinguish the contributions of various pollution sources and have limited support for identifying and assessing pollution sources and their contributions at the street scale.

[0106] like Figure 6 The diagram shown is a schematic representation of a street-scale air pollution simulation system provided in an embodiment of this application. The system includes: an input data acquisition module 610, a dynamic field module 620, a pollution source module 630, an atmospheric physicochemical process module 640, a pollutant-environment interaction feedback module 650, and an output result module 660, wherein: The input data acquisition module 610 is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data.

[0107] The dynamic field module 620 is used to receive mesoscale meteorological data from the input data acquisition module 610 and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data.

[0108] The pollution source module 630 is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module.

[0109] The atmospheric physical and chemical process module 640 is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data.

[0110] The pollutant-environment interaction feedback module 650 is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module.

[0111] The output result module 660 is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module.

[0112] The interactions between the modules in this street-scale air pollution simulation system can be found in [reference needed]. Figure 7 The corresponding descriptions in the aforementioned method embodiments will not be repeated here.

[0113] As described above, the street-scale air pollution simulation method system provided in this application adopts a comprehensive simulation strategy that integrates machine learning and parametric models. While ensuring high-precision simulation and wide-area coverage, it reduces computational resource consumption and simulation time per run, enabling operable and scalable applications across large street networks. Furthermore, this application constructs a model framework that reflects the interaction between pollutants and the environment, possessing the ability to simulate the interaction between pollutants and environmental factors, thereby improving the realism of street-scale pollution simulations.

[0114] This application also provides a street-scale air pollution source tracing system, the system comprising: an input data acquisition module, a dynamic field module, a pollution source module, an atmospheric physicochemical process module, a pollutant-environment interaction feedback module, a source tracing module, and an output result module, wherein: The input data acquisition module is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data. The dynamic field module is used to receive mesoscale meteorological data from the input data acquisition module and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data. The pollution source module is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module. The atmospheric physical and chemical process module is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data. The pollutant-environment interaction feedback module is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module. The source tracing module is used to perform multi-source quantitative source tracing analysis on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations in the street. The output results module is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module and the source tracing analysis results generated by the source tracing module.

[0115] and Figure 6 Compared to the simulation system in this embodiment, the source tracing system in this embodiment adds a source tracing module, which can perform multi-source quantitative source tracing analysis on the street-scale pollution distribution data, thereby improving the authenticity of street-scale pollution source tracing.

[0116] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0117] This application also provides a computer-readable storage medium storing a computer program that performs the above-described methods.

[0118] like Figure 8 As shown, the electronic device 600 may also include: a communication module 110, an input unit 120, an audio processor 130, a display 160, and a power supply 170. It is worth noting that the electronic device 600 does not necessarily need to include these components. Figure 8 All components shown; in addition, the electronic device 600 may also include Figure 8 For components not shown, please refer to existing technologies.

[0119] like Figure 8 As shown, the central processing unit 100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device. The central processing unit 100 receives inputs and controls the operation of various components of the electronic device 600.

[0120] The memory 140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 100 may execute the program stored in the memory 140 to perform information storage or processing, etc.

[0121] Input unit 120 provides input to central processing unit 100. Input unit 120 may be, for example, a keypad or touch input device. Power supply 170 provides power to electronic device 600. Display 160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0122] The memory 140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 140 can also be some other type of device. The memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application / function storage unit 142 for storing application programs and function programs or processes for executing the operation of the electronic device 600 via the central processing unit 100.

[0123] The memory 140 may also include a data storage unit 143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 144 of the memory 140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0124] The communication module 110 is a transmitter / receiver that transmits and receives signals via the antenna 111. The communication module 110 (transmitter / receiver) is coupled to the central processing unit 100 to provide input signals and receive output signals, which can be the same as in a conventional mobile communication terminal.

[0125] Based on different communication technologies, multiple communication modules 110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 110 (transmitter / receiver) is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide audio output via the speaker 131 and receive audio input from the microphone 132, thereby enabling typical telecommunications functions. The audio processor 130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 130 is coupled to a central processing unit 100, enabling on-device recording via the microphone 132 and on-device playback of stored audio via the speaker 131.

[0126] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied 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.

[0127] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0128] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0129] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0130] This application uses specific embodiments to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for simulating air pollution at the street-level, characterized in that, The method includes: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is used to obtain the street-scale dynamic field. Interpret street-scale emission data into pollution source characteristic information; Based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and pollution source characteristic information, collaborative processing is performed to obtain street-scale pollution distribution data. Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process. Output the pollution distribution data at the street scale.

2. The street-scale air pollution simulation method as described in claim 1, characterized in that, The street-scale dynamic field, obtained by using a comprehensive simulation strategy that integrates machine learning and parametric models based on mesoscale meteorological data, includes: Background meteorological parameters were obtained based on the aforementioned mesoscale meteorological data; Based on the background meteorological parameters and the corresponding street morphology parameters, the dynamic field parameters are calculated using a pre-trained machine learning model. When the background meteorological parameters and corresponding street morphology parameters exceed the applicable range of the machine learning model, the dynamic field parameters are calculated using the parameterized MARS model. The block-scale dynamic field is obtained based on the dynamic field parameters.

3. The street-scale air pollution simulation method as described in claim 2, characterized in that, In the process of obtaining the block-scale dynamic field based on the dynamic field parameters, the method further includes: The local surface roughness length of the urban underlying surface is corrected by combining the urban background roughness and the increments caused by local street valleys and building morphology; and In light of the urban heat island effect, the local sensible heat flux and the height of the mixing layer formed by thermal turbulence on the urban underlying surface are modified.

4. The street-scale air pollution simulation method as described in claim 1, characterized in that, The method of collaboratively processing simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and pollution source characteristic information to obtain street-scale pollution distribution data includes: In simulating a single diffusion of pollutants, based on the simulated grid data and the pollution source characteristic information, the concentration contribution of each individual emission source to a given receptor point is calculated using the Gaussian diffusion formula. In simulating the secondary generation of pollutants, NO and NO2 are treated as inert gases and their diffusion is simulated to obtain the primary concentration. After assuming that light steady state is reached, the concentrations of NO, NO2, and O3 are calculated. In the simulated background concentration mixing, the background concentration in the mesoscale pollution distribution and sectoral contribution data is deducted from the contribution of street-scale pollution sources, and then multiplied by the background concentration mixing ratio to obtain the background concentration at a specified height near the ground. In simulating dry particulate matter deposition, based on the aforementioned street-scale dynamic field, the gravitational settling velocity of the particulate matter is utilized. 、 The dry settling rate of particulate matter is calculated from aerodynamic drag and surface layer drag. In simulated wet particulate matter deposition, the cloud clearance coefficient, which characterizes wet deposition, is calculated using a semi-empirical formula based on the mass median diameter and the precipitation data. Street-scale pollution distribution data are generated based on the results of simulating primary pollutant diffusion, secondary pollutant generation, background concentration mixing, dry particulate matter deposition, and wet particulate matter deposition.

5. The street-scale air pollution simulation method as described in claim 1, characterized in that, The process of generating correction signals based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements to dynamically adjust the collaborative processing includes: Using PM 2.5 The concentration C is used to correct the total solar radiation within the valley area, thereby further correcting the change in photochemical reaction rate. The total solar radiation correction term is: ； In the above formula C is the total solar radiation correction term, where C represents PM2.

5. 2.5 The concentration of RH is the humidity.

6. A method for tracing the source of air pollution at the street-level, characterized in that, The method includes: Based on mesoscale meteorological data, a comprehensive simulation strategy that integrates machine learning models and parameterized models is used to obtain the street-scale dynamic field. Interpret street-scale emission data into pollution source characteristic information; Based on simulated grid data, mesoscale pollution distribution and sectoral contribution data, precipitation data, the street-scale dynamic field, and pollution source characteristic information, collaborative processing is performed to obtain street-scale pollution distribution data. Based on the street-scale pollution distribution data and the interaction mechanism between pollutants and environmental elements, a correction signal is generated to dynamically adjust the collaborative processing process. Output the pollution distribution data at the street scale; A multi-source quantitative source tracing analysis was performed on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations within the street.

7. The method for tracing the source of air pollution at the street scale as described in claim 6, characterized in that, The multi-source quantitative source tracing analysis of the street-scale pollution distribution data includes: A traversal algorithm is used to sum the individual pollutants affecting each receptor site, and the pollution contribution of each individual pollutant source to the receptor site is obtained in sequence. By specifying the grouping field in the input file, the pollution contribution of each group of pollutants to the specified receptor site is quantified, thereby realizing source tracing analysis.

8. A street-scale air pollution simulation system, characterized in that, The system includes: an input data acquisition module, a dynamic field module, a pollution source module, an atmospheric physicochemical process module, a pollutant-environment interaction feedback module, and an output result module, wherein: The input data acquisition module is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data. The dynamic field module is used to receive mesoscale meteorological data from the input data acquisition module and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data. The pollution source module is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module. The atmospheric physical and chemical process module is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data. The pollutant-environment interaction feedback module is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module. The output results module is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module.

9. A street-scale air pollution source tracing system, characterized in that, The system includes: an input data acquisition module, a dynamic field module, a pollution source module, an atmospheric physicochemical process module, a pollutant-environment interaction feedback module, a source tracing module, and an output result module, wherein: The input data acquisition module is used to acquire mesoscale meteorological data, street-scale emission data, simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data. The dynamic field module is used to receive mesoscale meteorological data from the input data acquisition module and output the street-scale dynamic field to the atmospheric physical and chemical process module based on the mesoscale meteorological data. The pollution source module is used to receive block-scale emission data from the input data acquisition module, parse the block-scale emission data into pollution source characteristic information, and then transmit it to the atmospheric physical and chemical process module. The atmospheric physical and chemical process module is used to receive simulated grid data, mesoscale pollution distribution and sectoral contribution data, and precipitation data from the input data acquisition module. It also receives the street-scale dynamic field output by the dynamic field module and the pollution source characteristic information transmitted by the pollution source module. Through its included primary diffusion submodule, secondary generation submodule, background concentration mixing scheme submodule, dry deposition submodule, and wet deposition submodule, it performs collaborative processing to generate street-scale pollution distribution data. The pollutant-environment interaction feedback module is used to receive the street-scale pollution distribution data generated by the atmospheric physicochemical process module, generate a correction signal based on the interaction mechanism between pollutants and environmental elements, and transmit the correction signal back to the atmospheric physicochemical process module to dynamically adjust the processing of the atmospheric physicochemical process module. The source tracing module is used to perform multi-source quantitative source tracing analysis on the street-scale pollution distribution data to assess the pollution contribution of each pollution source to specific locations in the street. The output results module is used to acquire and present the street-scale pollution simulation results generated by the atmospheric physicochemical process module and the source tracing analysis results generated by the source tracing module.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

Images