High-precision atmospheric pollution prediction method

Abstract

The invention belongs to the technical field of environmental protection monitoring, and particularly relates to a high-precision atmospheric pollution prediction method, which comprises the following steps: S1, dividing a monitoring area into a plurality of monitoring units; s2, marking a potential atmospheric pollutant emission source as a fixed monitoring point; s3, determining random monitoring points in each monitoring unit; s4, constructing an atmospheric pollutant diffusion model of the monitoring area; s5, acquiring real-time atmospheric pollutant data of the fixed monitoring points and the random monitoring points, and inputting the real-time atmospheric pollutant data into the atmospheric pollutant diffusion model to obtain an atmospheric pollution condition prediction result of the monitoring area; and S6, determining the prediction deviation of the atmospheric pollutant diffusion model based on the current monitoring data and the historical atmospheric pollution condition prediction result, and correcting the prediction result according to the prediction deviation. The atmospheric pollution prediction method has the advantages of simple and reasonable model structure, high prediction precision, strong timeliness, strong migration ability and wide application range.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of environmental protection monitoring, and particularly relates to a high-precision atmospheric pollution prediction method. BACKGROUND

[0002] Atmospheric pollutants refer to those substances discharged into the atmosphere due to human activities or natural processes, and which have harmful effects on humans and the environment. Although the composition of clean air is relatively stable, once trace harmful substances that do not originally exist or have a significantly increased concentration appear in a certain area, and the quantity and duration reach a certain degree, they can cause direct or potential harm to humans, animals, plants and materials. When the concentration of pollutants in the atmosphere reaches or exceeds the threshold of harm, thereby destroying the stability of the ecological system and threatening the normal survival and development conditions of humans, it constitutes atmospheric pollution.

[0003] Atmospheric pollution not only affects the respiratory system, cardiovascular system and other physiological health of humans, but also can induce psychological health problems, cause social problems such as reduced labor efficiency, weakened creativity and decreased life happiness, and therefore receives high attention from people. The migration and diffusion process of pollutants in the atmosphere has significant spatial heterogeneity and temporal dynamics, and the concentration change thereof is jointly driven by multiple factors, including the accumulation and dilution process of local pollutants, regional meteorological conditions, geographical features and pollutant transport from surrounding areas. Therefore, constructing a high-precision and high-time-efficiency pollutant monitoring and diffusion prediction technology system to timely regulate and control atmospheric pollution is an important means to cope with the problem of atmospheric pollution.

[0004] At present, the monitoring and prediction of atmospheric pollutants mainly rely on two types of technical systems: one is the conventional monitoring technology represented by fixed environmental air quality monitoring stations, which is mainly used for long-term collection of basic data such as pollutant concentration and meteorological parameters at a specific geographic location; the other is the pollutant migration and diffusion prediction technology based on meteorological driving and geographical factor modeling, which combines pollution sources, wind speed and direction, topography and other factors to build a simulation model of the spatial-temporal distribution of pollutants, and can assist in realizing regional pollution situation judgment and trend prediction. The traditional fixed monitoring method has the advantages of high monitoring accuracy and good data continuity, but due to its limited number of points and insufficient spatial coverage, it is difficult to fully reflect the dynamic migration process of pollutants in a large-scale atmosphere, and generally lacks the coupling ability with topography and real-time weather conditions, resulting in a lag or too large deviation in reflecting the diffusion trend of pollutants. The prediction modeling method developed on this basis uses grid modeling and remote sensing data fusion to combine meteorological factors and topographic data to build a diffusion path simulation model of pollutants in the space-time dimension, representing the modern development direction of pollution prediction technology. However, this type of model still has problems such as complex data processing, incomplete consideration of prediction factors, limited model accuracy, and bloated model structure, which restrict its timeliness and applicability in actual environmental monitoring and emergency response. Specifically, the current prediction modeling method mainly has the following problems:

[0005] First, the data processing amount is large and the complexity is high: as known, the precision and accuracy of the prediction model are directly related to the sample size and the number of monitoring stations. Multiple stations can provide more regional pollution data, enabling the model to learn spatial heterogeneity and improve prediction accuracy and accuracy. However, the number of existing air quality monitoring stations is limited, which cannot provide enough sample data for the prediction model. Therefore, existing high-precision prediction models rely on remote sensing images to uniformly and continuously cover a large monitoring area, and then divide the image into multiple grids for one-by-one operation and analysis to improve sample size and sample coverage. However, this approach has the disadvantage of large data processing amount and high complexity. This approach not only requires a large number of images, but also requires image stitching, coordinate correction, spatial inversion, radiation correction, cloud detection and cloud removal, image enhancement and other high-complexity processing after the image is transmitted back to the ground. This approach not only results in a large amount of data and a tedious processing process, but also requires high performance and data communication bandwidth of computing devices, thereby limiting the processing efficiency and real-time prediction capability of the model, making it difficult to meet the application requirements of large-scale regional pollutant migration rapid response.

[0006] Second, the prediction factors are not comprehensive, the model accuracy is limited, and the prediction accuracy is poor: Currently, most models mainly focus on the influence of terrain, wind speed and direction, precipitation and other conventional meteorological geographical factors on pollutant dispersion, but in fact, the dispersion of atmospheric pollutants is also affected by many factors, such as atmospheric stability, atmospheric temperature and temperature inversion layer, atmospheric relative humidity, solar radiation, sea-land breeze, urban heat island effect, pollution source emission intensity and height, urban building layout, pollutant molecular mass and density, water solubility and reactivity of pollutants, relative content of primary and secondary pollutants, and surface cover type, etc. For example, existing prediction models generally ignore the regulatory effect of surface cover type on pollutant dispersion, especially the adsorption, retardation and transformation capacity of vegetation systems such as mountain forests on pollutants. Simplification and neglect of the factors affecting pollutant dispersion in the prediction model will lead to deviation in the simulation of pollutant dispersion, reducing the accuracy and adaptability of the prediction model.

[0007] Third, the model structure is redundant and the calculation efficiency is low: In order to improve the stability and adaptability of the model in complex scenarios, existing atmospheric pollution prediction models often rely on the introduction of a large number of feature parameters and redundant modules to achieve this, but this approach will result in a large model size, complex calculation, and a decline in running efficiency, making it prone to overfitting problems, poor generalization ability, and not conducive to efficient deployment and application in actual scenarios.

[0008] On this basis, in order to improve the prediction accuracy and timeliness of modern modeling techniques for the migration and dispersion process of pollutants, researchers in the field have made many technical explorations, such as building multi-objective prediction models, using multi-pollutant collaborative prediction mechanisms to identify the non-linear coupling relationship between pollutants, using online learning or incremental learning mechanisms to enable the model to have self-adaptive updating capability, using transfer learning and graph neural networks to improve the model's transferability in different cities or regions and improve the model's cross-regional generalization ability. Among them, the most representative research achievement is the WRF-Chem+LSTM method, which is a hybrid prediction method combining physical mechanism models (WRF-Chem) and data-driven models (LSTM). It first uses WRF-Chem for preliminary prediction to obtain the predicted value of air pollutant concentration in the future time period, then introduces the LSTM model for error learning and correction, trains LSTM based on historical observation data and WRF-Chem prediction results to identify the system bias of WRF-Chem, and generates a more accurate correction prediction. The essence is to post-process or bias correct the output of WRF-Chem using LSTM. Although LSTM can significantly improve prediction accuracy, it also has some technical limitations and potential drawbacks:

[0009] (1) LSTM model needs to be frequently retrained, poor timeliness: atmospheric pollution is significantly affected by factors such as season, emission source, policy intervention, LSTM model is sensitive to training data distribution, if external conditions change, the model may fail, so it needs to be continuously updated and retrained to maintain accuracy, high maintenance cost;

[0010] (2) LSTM model is difficult to migrate across regions: a LSTM model trained in one city or region is difficult to directly apply to another region, and needs to collect data and train again for different geographical environments and emission structures, the transfer learning of LSTM model is still in the initial stage of research;

[0011] (3) LSTM model has the risk of "error amplification": if WRF-Chem prediction itself is abnormal, LSTM may "learn" this error trend and overfit some type of bias of WRF-Chem, further amplifying the error.

[0012] In summary, the existing atmospheric pollution monitoring and prediction technology still has many shortcomings in data processing efficiency, comprehensiveness of prediction factors and rationality of model structure, and there are many limitations for the post-correction technology of the prediction model, it is difficult to realize efficient, low-cost and fast pollutant diffusion prediction while ensuring prediction accuracy. Therefore, how to improve the prediction accuracy of the model while effectively reducing the complexity of data processing, improving the timeliness and practicality of the system has become a key technical problem to be solved in the field. SUMMARY

[0013] The purpose of the present application is to solve the above-mentioned technical problems, provide a high-precision atmospheric pollution prediction method, to improve the prediction accuracy of the model while effectively reducing the complexity of data processing, reducing feature parameters and redundant modules, improving the timeliness and practicality of the system.

[0014] Therefore, the present application provides a high-precision atmospheric pollution prediction method, comprising the steps of:

[0015] S1, dividing the monitoring area into several monitoring units based on land use, surface cover type and / or vegetation distribution data;

[0016] S2, determining the potential atmospheric pollutant emission source of the monitoring area, and labeling it as a fixed monitoring point;

[0017] S3, determining the number of random monitoring points in each monitoring unit according to the atmospheric pollutant prediction accuracy requirement and the area of each monitoring unit and the gas diffusion anomaly factor, and coupling the random monitoring points and the fixed monitoring points to establish a high-precision monitoring network;

[0018] S4, constructing an atmospheric pollutant diffusion model of the monitoring area based on geographical information, historical meteorological information and historical pollutant information of the fixed monitoring points and the random monitoring points;

[0019] S5, obtaining real-time atmospheric pollutant data of the fixed monitoring points and the random monitoring points based on remote sensing images of the fixed monitoring points and the random monitoring points, and inputting the real-time atmospheric pollutant data into the atmospheric pollutant diffusion model to predict future atmospheric pollution, and obtaining a prediction result of atmospheric pollution of the monitoring area in a future period of time;

[0020] S6, determining a prediction deviation of the atmospheric pollutant diffusion model based on current monitoring data and the prediction result of the historical atmospheric pollution, and correcting the prediction result of atmospheric pollution of the monitoring area in the future period of time according to the determined prediction deviation to obtain a corrected prediction result of atmospheric pollution.

[0021] Further, the step S2 specifically comprises:

[0022] S21, obtaining or constructing geographical information of the monitoring area;

[0023] S22, identifying and classifying potential atmospheric pollution sources;

[0024] S23, marking the potential atmospheric pollutant emission sources as fixed monitoring point candidates;

[0025] S24, merging and screening the fixed monitoring point candidates to determine the final fixed monitoring points;

[0026] S25, determining specific positions of the fixed monitoring points.

[0027] Further, in the step S3, the process of determining the arrangement number of the random monitoring points in each monitoring unit is as follows:

[0028] (1) establishing a dynamic evaluation mechanism of gas diffusion anomaly factors, wherein the gas diffusion anomaly factor of each monitoring unit = historical pollutant content standard deviation / historical pollutant content average value;

[0029] (2) determining the basic number of random monitoring points of each monitoring unit according to the size of the gas diffusion anomaly factor;

[0030] (3) adjusting the basic number of random monitoring points in combination with the area of each monitoring unit.

[0031] Further, in the step S5, after obtaining the prediction result of atmospheric pollution of the monitoring area in a future period of time, firstly, the prediction result of atmospheric pollution is analyzed and processed, and the specific process comprises steps of:

[0032] S51, obtain the pollutant distribution prediction result, determine the pollutant distribution boundary and center;

[0033] S52, draw the pollutant distribution vector diagram C corresponding to the future atmospheric pollution prediction result, taking the pollutant distribution center as the starting point and the pollutant distribution boundary line as the end point.

[0034] Further, the step S6 specifically comprises:

[0035] S61, obtain the atmospheric pollution prediction result at the current time output by the atmospheric pollutant diffusion model;

[0036] S62, obtain the atmospheric pollution measurement result at the current time based on the remote sensing image monitoring;

[0037] S63, analyze the atmospheric pollution prediction result and the atmospheric pollution measurement result at the current time respectively, determine the pollutant distribution boundary and center, and then draw the pollutant distribution vector diagram A corresponding to the current atmospheric pollution prediction result and the pollutant distribution vector diagram B corresponding to the current atmospheric pollution measurement result, taking the pollutant distribution center as the starting point and the pollutant distribution boundary line as the end point.

[0038] S64, determine the prediction deviation of the atmospheric pollutant diffusion model based on the pollutant distribution vector diagram A and B results;

[0039] S65, correct the atmospheric pollution prediction result of the monitoring area in the future period of time according to the determined prediction deviation, and obtain the corrected atmospheric pollution prediction result.

[0040] Further, in the steps S52 and S63, the pollutant distribution vector diagram drawn comprises a plurality of one-to-one corresponding and same angle sub-vectors, the sub-vector taking the pollutant distribution center as the starting point and the pollutant distribution boundary line as the end point, and being drawn every set angle interval.

[0041] Further, in the step S64, the process of determining the prediction deviation of the atmospheric pollutant diffusion model based on the pollutant distribution vector diagram A and B results is as follows:

[0042] S641, calculate the deviation coefficient α of the pollutant distribution area in the pollutant distribution vector diagram A and B, wherein the deviation coefficient α=(1-K)*100%, the K is the coincidence ratio, and the coincidence ratio K=the overlapping area of the vector diagram A and B / the total area of the pollutant distribution area in the diagram B;

[0043] S642, compare the relative sizes of the deviation coefficient α and the preset threshold α 阈 .

[0044] If α<α阈 If the atmospheric pollution situation prediction result of the next period output by the atmospheric pollution diffusion model needs to be corrected, the step S65 is continued to directly output the atmospheric pollution situation prediction result of the atmospheric pollution diffusion model;

[0045] If α ≥ α 阈 , the atmospheric pollution situation prediction result of the next period output by the atmospheric pollution diffusion model needs to be corrected, and the step S643 is continued to be executed;

[0046] S643, taking the center point of the pollutant distribution area in the pollutant distribution vector diagram A as the starting point and the center point of the pollutant distribution area in the pollutant distribution vector diagram B as the ending point, a prediction deviation vector between the center points of the pollutant distribution areas in the pollutant distribution vector diagrams A and B is drawn ;

[0047] Then, the pollutant distribution vector diagram A is translated along the starting point to the ending point of the vector to obtain a translated pollutant distribution vector diagram A';

[0048] The difference between each component vector in the pollutant distribution vector diagram B and the pollutant distribution vector diagram A' is calculated respectively to obtain a set of component vector deviation vectors corresponding to each angle

[0049] And the prediction deviation vector between the center points and the component vector deviation vector represent the prediction deviation of the atmospheric pollution diffusion model.

[0050] Further, the step S65 comprises:

[0051] When α < α 阈 , the atmospheric pollution situation prediction result of the atmospheric pollution diffusion model is directly output;

[0052] When α ≥ α 阈 , first, the center point of the pollutant distribution vector diagram C is translated along the vector , then the component vectors in the pollutant distribution vector diagram C are added to the corresponding component vector deviation vectors to obtain corrected component vectors, and the boundaries of the pollutant distribution vector diagram C are adjusted according to the corrected component vectors to obtain a pollutant distribution vector diagram C', and the pollutant distribution vector diagram C' is output as the final atmospheric pollution situation prediction result.

[0053] Further, the step S65 further comprises: when α ≥ α 阈 , judging whether deep correction of the prediction result is needed:

[0054] Firstly, the pollutant distribution vector diagram A is translated along the predicted deviation vector to obtain a translated pollutant distribution vector diagram A', so that the center point of the translated pollutant distribution vector diagram A' coincides with the center point in the pollutant distribution vector diagram B, and then a deviation coefficient a' of the pollutant distribution area in the pollutant distribution vector diagrams A' and B is calculated, wherein the deviation coefficient a' = (1-K') * 100%, and the K' is a coincidence ratio, and the coincidence ratio K = a coincident area of the vector diagrams A' and B / a total area of the pollutant distribution area in the diagram B;

[0055] Then, the relative size of the deviation coefficient a' and a preset threshold a 阈 is compared.

[0056] If a 阈 '< a, no deep correction is needed for the pollutant distribution vector diagram C.

[0057] If a 阈 '> a, deep correction is needed for the pollutant distribution vector diagram C.

[0058] Further, the deep correction process is as follows:

[0059] Firstly, a boundary line of the pollutant distribution vector diagram B is analyzed to obtain extreme points on the boundary line, and then an extreme value vector of the pollutant distribution vector diagram B is drawn with the center point of the pollutant distribution vector diagram B as a starting point and each extreme point on the boundary line as an ending point, and a vector set composed of the extreme value vectors is referred to as an extreme value vector set.

[0060] Then, an extreme value vector corresponding to the angle of each extreme value vector of the pollutant distribution vector diagram B is drawn on the pollutant distribution vector diagram A.

[0061] After that, a difference between each extreme value vector of the pollutant distribution vector diagram B and the pollutant distribution vector diagram A is calculated to obtain a set of extreme value deviation vectors corresponding to each extreme value

[0062] When a 阈 '> a, firstly, an extreme value vector corresponding to the angle of each extreme value vector of the pollutant distribution vector diagram B is drawn on the pollutant distribution vector diagram C, and then the center point of the pollutant distribution vector diagram C is translated along the vector , and then a component vector in the pollutant distribution vector diagram C is added to the corresponding component vector deviation vector to obtain a corrected component vector, and an extreme value vector in the pollutant distribution vector diagram C is added to the corresponding extreme value deviation vector to obtain a corrected extreme value vector.

[0063] And according to the corrected subvector and the extreme value vector, the boundary of the pollutant distribution vector graph C is adjusted to obtain the pollutant distribution vector graph C', and the pollutant distribution vector graph C' is output as the final atmospheric pollution prediction result.

[0064] The high-precision atmospheric pollution prediction method has the advantages of simple and reasonable model structure, high prediction accuracy, strong timeliness, strong migration ability and wide application range. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 is a flow chart of the high-precision atmospheric pollution prediction method of the present application;

[0066] Figure 2 is a schematic diagram of the pollutant distribution vector graphs A and B of the present application;

[0067] Figure 3 is a comparison chart of the prediction result, the measured value and the prediction result of the prior art of the embodiment of the present application. DETAILED DESCRIPTION

[0068] The technical solutions of the present application will be described in detail below with specific embodiments. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application.

[0069] In the description of the present application, it should be noted that the terms used herein are only for the purpose of describing the specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. For the convenience of description, the known technologies, methods and devices for those skilled in the related art may not be discussed in detail, but should be regarded as part of the authorized description.

[0070] It should be noted that in this application, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article, or apparatus that comprises a list of elements does not only include those elements, but can also include other elements that are not expressly listed or inherent to such process, method, article, or apparatus. Without more limitations, an element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element. In addition, it should be pointed out that the scope of the methods and apparatus in the embodiments of the present application is not limited to performing functions in the order shown or discussed, but can also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method can be performed in an order different from that described, and various steps can also be added, omitted, or combined. In addition, the features described with reference to certain examples can be combined in other examples.

[0071] A high-precision atmospheric pollution prediction method, as shown in Figure 1 includes the steps of:

[0072] S1, dividing the monitoring area into a plurality of monitoring units based on land use, surface cover type and / or vegetation distribution data;

[0073] S2, determining potential atmospheric pollutant emission sources of the monitoring area, and marking them as fixed monitoring points;

[0074] S3, determining the number of random monitoring points in each monitoring unit according to the atmospheric pollutant prediction accuracy requirement, based on the area of each monitoring unit and the gas diffusion anomaly factor, and coupling the random monitoring points and the fixed monitoring points to establish a high-precision monitoring network;

[0075] S4, constructing an atmospheric pollutant diffusion model of the monitoring area based on the geographic information, historical weather information and historical pollutant information of the fixed monitoring points and the random monitoring points;

[0076] S5, obtaining real-time atmospheric pollutant data of the fixed monitoring points and the random monitoring points based on remote sensing images of the fixed monitoring points and the random monitoring points, and inputting the real-time atmospheric pollutant data into the atmospheric pollutant diffusion model to predict future atmospheric pollution, and obtaining a prediction result of the atmospheric pollution of the monitoring area in a future period of time;

[0077] S6, determining the prediction deviation of the atmospheric pollutant diffusion model based on the current monitoring data and the prediction result of the historical atmospheric pollution, and correcting the prediction result of the atmospheric pollution of the monitoring area in the future period of time according to the determined prediction deviation to obtain a corrected prediction result of the atmospheric pollution.

[0078] Further, in the step S1, the monitoring area can be divided according to land use, ground cover type and / or vegetation distribution data, respectively. For example, the land use can be divided into construction land, forest land, water land, farmland, etc.; the ground cover type can be divided into artificial surface, farmland, forest land, grassland, wetland, water body, bare land, snow land / glacier, etc.; and the vegetation distribution data can be divided into forest, shrub land, grassland, farmland, wetland vegetation, bare land, glacier, etc.

[0079] Preferably, in the step S1, the difference in topography should also be considered when dividing the monitoring units. Each monitoring unit should have the same or similar topography.

[0080] The diffusion of atmospheric pollutants has strong spatial heterogeneity, and different landforms and types directly affect the diffusion rate, migration path and deposition mode. In the present application, the monitoring units are classified and divided, forming the smallest unit of the monitoring process. The division of the smallest unit is not only based on topography, but also takes into account the actual properties of the ground, so that each monitoring unit has ecological and geographical significance, and can more truly reflect the differences in the diffusion behavior of pollutants on different underlying surfaces, such as water bodies, forests, urban roads, and bare land. This can improve the geographical adaptability and physical fit of the atmospheric pollutant diffusion model constructed therefrom.

[0081] In particular, the monitoring unit has spatial and topographical consistency, and has a high consistency in the influence on the diffusion process of atmospheric pollutants. This high spatial and topographical consistency reduces the uncertainty of the diffusion process, so that the kinetic mechanism of atmospheric pollutant diffusion in each monitoring unit tends to be consistent, and the prediction error caused by the heterogeneity of the internal area of the monitoring unit can be reduced.

[0082] On this basis, taking the monitoring unit as the smallest modeling unit, a large area can be modularized and processed in blocks, reducing data redundancy and computational redundancy caused by unified modeling of the entire area, supporting parallel computing and local updating, significantly improving system prediction efficiency and real-time response capability, avoiding unnecessary remote sensing image processing and global inversion, improving the timeliness and scalability of the model, and adapting to the deployment needs of large areas and multiple scenarios. The monitoring unit divided based on ecological and geographical properties has clear semantic boundaries, which facilitates subsequent pollutant source analysis, risk warning, and the development of pollution control strategies, as well as the fine implementation of policies and regional differentiated management.

[0083] Preferably, in the step S1, a plurality of division schemes of the monitoring units can be formed according to land use, ground cover type and / or vegetation distribution data, and then the best division scheme of the monitoring units can be determined according to needs.

[0084] It should be noted that in determining the optimal monitoring unit division scheme, different user needs and emphases can be freely selected, such as the calculation efficiency, accuracy, stability, etc. of the model as a reference for selection, wherein the calculation efficiency of the model can complete a simulation or training time as a measurement index, the accuracy of the model can be the closeness of the model prediction result to the real situation or observation data, such as the deviation of the prediction from the real situation as a measurement index, and the stability of the model can be the anti-interference ability of the model as a measurement index.

[0085] Further, the step S2 specifically comprises:

[0086] S21, acquiring or constructing geographic information of the monitoring area: acquiring or constructing geographic information of the monitoring area through satellite remote sensing images, DEM terrain elevation model, field surveying and mapping, etc. The geographic information includes elevation data, land use, ground cover type and vegetation distribution data, etc.

[0087] S22, identifying and classifying potential atmospheric pollution sources: the potential atmospheric pollution sources can be marked or identified by referring to local emission inventory, national pollution source survey data, remote sensing identification method, etc. The potential atmospheric pollution sources include industrial production facilities such as factories, boilers, power plants, waste incineration plants, etc., transportation facilities such as highways, transportation hubs, etc., and incidental facilities such as construction sites, forest fires, vegetation volatile organic compounds, etc.

[0088] S23, marking the potential atmospheric pollution sources as fixed monitoring point candidates; such as importing the positions of the atmospheric pollution sources into the monitoring area topographic map, adding attribute fields to each potential atmospheric pollution source, such as pollution source type, emission species, emission level, emission time characteristics, etc., and visualizing the pollution intensity using hierarchical symbol method, such as using large red dots to represent heavy pollution sources and small red dots to represent light pollution sources.

[0089] S24, merging and screening the fixed monitoring point candidates to determine the final fixed monitoring points: such as merging multiple fixed monitoring point candidates with close distances into one fixed monitoring point according to the distances of the fixed monitoring point candidates; canceling fixed monitoring points with few people, low emissions, and low monitoring value; or referring to historical meteorological information, historical pollution information and topography of the monitoring area, analyzing the convergence zone of heavy pollution area, ventilation corridor, pollution transmission channel, canceling the fixed monitoring points in the low atmospheric pollution risk area, etc., determining the final fixed monitoring points through artificial or machine intelligent screening;

[0090] S25, determining the specific position of the fixed monitoring point: when laying the monitoring points, the specific positions of the candidate points of the fixed monitoring points are arranged near the point-like emission sources such as industrial parks, factories, boilers, power stations, waste incineration plants, construction sites and the like; the population-dense or sensitive areas (such as schools and hospitals) near the line-like emission sources such as highways and traffic hubs; and the downwind of the area of the surface-like emission sources such as mountain fires and vegetation volatile organic compounds.

[0091] In the present application, by identifying, screening and scientifically laying the fixed monitoring points through steps S21-S25, more targeted and effective technical optimization can be achieved at the level of pollution source identification and monitoring network construction. Starting from the actual situation of pollution source distribution, the candidate set of the fixed monitoring points is dynamically determined to achieve comprehensive coverage of typical emission sources. Instead of relying on fixed standard grid points, the fixed monitoring points are laid based on the distribution characteristics of the pollution sources; and by labeling the attributes such as the type of pollution source, the type of emission, the intensity and the time characteristics, a more representative emission characteristic database can be constructed, which is helpful for hierarchical visualization, further optimization of monitoring point screening and subsequent visual analysis, avoidance of blind point laying and improvement of the regional adaptability of the pollution identification and prediction model.

[0092] On this basis, by the strategy of merging and screening the candidate points of the fixed monitoring points, the redundant and low-value monitoring points can be effectively filtered, the limited sampling resources can be laid in the areas with the most representative and monitoring value, the redundancy of the monitoring points can be reduced, the cost of point laying can be controlled, and the resource use efficiency and economy of the overall monitoring system can be improved.

[0093] Further, the present application also adopts a classification point laying strategy for different types of emission sources such as point-like, line-like and surface-like emission sources, which can fully reflect the differences in diffusion behaviors of various pollution sources, improve the model identification and tracing ability, and improve the adaptability and spatio-temporal sensitivity of the model to complex pollution scenarios with multiple sources.

[0094] In addition, in the step S2, the setting of the fixed monitoring points can realize the continuous monitoring and dynamic tracking of the key pollution sources in the monitoring area, real-time and continuous acquisition of the pollutant concentration data, tracking of the time variation law of the pollutant emission, accurate identification of the main pollutant types and their sources, and provision of high-quality data support for subsequent model correction, pollutant diffusion prediction, pollution tracing analysis and the like. At the same time, the accuracy of the initial conditions and boundary conditions of the atmospheric pollutant diffusion model can be enhanced, the fixed monitoring points provide spatially reasonable and temporally continuous data sources, which can be used as the data input of the atmospheric pollutant diffusion model, improve the response ability of the model to the actual pollutant migration, realize the rolling correction and dynamic optimization of the pollution variation trend, reduce the prediction error of the model, and improve the warning accuracy and timeliness, overcoming the problem of large prediction deviation caused by the dependence of the existing model on static input or incomplete meteorological driving.

[0095] In addition, the setting of the fixed monitoring point can also support the accurate triggering of the pollution early warning mechanism. When the concentration of pollutants at a certain fixed monitoring point reaches or approaches the preset threshold, the short-term early warning mechanism can be automatically triggered to start the local pollution emergency response process. At the same time, the setting of the fixed monitoring point can also be used to support pollution tracing, evaluation and decision analysis.

[0096] Further, in the step S3, on the basis of the fixed monitoring point layout, the number of random monitoring points in each monitoring unit can be further determined according to the atmospheric pollutant prediction accuracy requirement, based on the area of each monitoring unit and the gas diffusion anomaly factor, and then the random monitoring points and the fixed monitoring points are coupled and associated to establish a high-precision monitoring network.

[0097] Specifically, the number of random monitoring points arranged in each monitoring unit is dynamically determined according to the area of the monitoring unit and the gas diffusion anomaly factor, and the specific process is as follows:

[0098] (1) First, a dynamic evaluation mechanism of the gas diffusion anomaly factor is established, wherein the gas diffusion anomaly factor of each monitoring unit = historical pollutant content standard deviation / historical pollutant content average value; the historical pollutant content can be the monthly or annual average pollutant content of the monitoring unit;

[0099] (2) Then, the basic number of random monitoring points in each monitoring unit is determined according to the size of the gas diffusion anomaly factor. Generally, the larger the gas diffusion anomaly factor, the more the basic number of random monitoring points arranged in the corresponding monitoring unit. For example, when the gas diffusion anomaly factor is large, such as greater than 0.5, the number of random monitoring points arranged in the monitoring unit can be appropriately increased, such as 6-10; when the gas diffusion anomaly factor is small, such as less than 0.2, the number of random monitoring points arranged in the monitoring unit can be appropriately reduced, such as 1-3; when the gas diffusion anomaly factor is moderate, such as between 0.2 and 0.5, the number of random monitoring points arranged in the monitoring unit can be set to 4-5;

[0100] (3) Then, the basic number of random monitoring points is adjusted in combination with the area of each monitoring unit. Generally, when the area of the monitoring unit is large, the number of random monitoring points can be appropriately increased, and vice versa, when the area of the monitoring unit is small, the number of random monitoring points can be appropriately reduced. For example, the monitoring units can be first classified according to the size of the area, such as large area monitoring units (>100km 2 ), medium area monitoring units (50-100km 2 ), and small area monitoring units (<50km 2), and then the basic number of random monitoring points is adjusted according to the classification, for example, the arrangement number of random monitoring points of a large-area monitoring unit can be adjusted to the basic number of monitoring points*(2-5), the arrangement number of random monitoring points of a medium-area monitoring unit can be adjusted to the basic number of monitoring points*(1.2-2.5), and the arrangement number of random monitoring points of a small-area monitoring unit can be adjusted to the basic number of monitoring points*(0.8-1).

[0101] Further, the arrangement position of the random monitoring point can be determined by a random sampling model or the like.

[0102] As some examples of the present application, the arrangement position of the random monitoring point can be generated by a Monte Carlo random sampling algorithm to generate a monitoring sampling point set satisfying a preset number and meeting a spatial distribution constraint, wherein a plurality of monitoring points in the same monitoring unit tend to be evenly distributed.

[0103] Preferably, after generating the position and number of random monitoring points, the positions of the random monitoring points and the fixed monitoring points can be compared, and the repeated monitoring points caused by the individual distance being too close can be cancelled.

[0104] In the present application, the introduction of the random monitoring point can fill the blank of the fixed monitoring point in the non-key area or the pollution variable area, improve the spatial continuity and accuracy of the prediction model, at the same time, the random point arrangement can provide more abundant and diverse training samples for the model, reduce the dependence on specific points, prevent overfitting, and improve the generalization ability.

[0105] Further, in the present application, the gas diffusion anomaly factor is used as the basis for point arrangement, which can accurately identify the monitoring unit with rapid pollution change and perform "encryption monitoring", thereby improving the sensitivity and accuracy of the overall prediction. At the same time, the present application determines the number of random points by area and anomaly factor, so that the model has the ability to identify spatial heterogeneity, thereby more accurately simulating the pollution migration and diffusion track.

[0106] On this basis, the extraction method of the monitoring point used in the present application can combine the accuracy of the fixed monitoring point with the wide coverage of the random monitoring point, make up for the short board of sparse distribution of fixed points, at the same time avoid redundant deployment, and improve the point arrangement economy and information coverage.

[0107] As some examples of the present invention, in step S4, the geographic information includes elevation DEM data, road network and traffic density map, river and lake distribution, administrative boundaries and grid division, etc.; the historical meteorological information includes data such as wind speed and direction, solar radiation and sunshine duration, humidity, temperature and temperature gradient, precipitation and precipitation intensity, etc.; the historical pollutant information includes spatial distribution map / raster data of pollutants, pollution source location, emission intensity, pollution concentration, type, diffusion trajectory, distribution, etc. The specific composition of the geographic information, historical meteorological information and historical pollutant information can be selected according to the needs of the model construction.

[0108] Preferably, based on the post-correction mechanism of the present invention, when constructing the atmospheric pollutant diffusion model, the contribution of each feature parameter to the prediction result can be sorted, and then several feature parameters with larger contributions, such as 5 to 8 feature parameters, can be selected for training to construct the atmospheric pollutant diffusion model of the monitoring area. In this process, the influence of feature parameters with lower contributions on the prediction result can be ignored, and their influence on the prediction result can be uniformly reflected through the post-correction procedure, making the model structure more reasonable and concise, avoiding the introduction of too many feature parameters, which would lead to model structure redundancy, low computational efficiency, and poor generalization ability, making the model lighter, with lower training cost, and more reasonable structure.

[0109] Furthermore, in step S4, after acquiring historical meteorological and pollutant information of the monitoring area, the data can be preprocessed first. The data preprocessing process includes at least: data format unification (converting data from different sources, such as CSV, GeoTIFF, and NetCDF, into a uniformly processable format), time alignment (aligning meteorological and pollutant data to the same time resolution), spatial interpolation / rasterization (performing spatial interpolation on pollutant location data, such as Kriging and IDW, to generate a continuous pollution field), and data cleaning (removing outliers and processing missing values ​​using interpolation methods, etc.). After the data preprocessing, the obtained dataset is used to train a preset machine learning model to obtain an atmospheric pollutant diffusion model.

[0110] Furthermore, the detailed process of constructing the atmospheric pollutant diffusion model in step S4 has been described in the prior art and will not be repeated here.

[0111] It should be noted that this invention mainly focuses on optimizing the input parameters of the prediction model and correcting the prediction results of the model. Therefore, the specific type of atmospheric pollutant diffusion model used in this invention is not limited.

[0112] As some examples of the present application, the atmospheric pollutant diffusion model can be a physical mechanism model based on aerodynamics and diffusion equation, such as Gaussian plume model, Lagrangian model, WRF-Chem model, etc.; a statistical and machine learning model based on historical data rules, such as Random Forest, SVM, XGBoos, Kriging model, etc.; and a deep learning model or a multi-model fusion model.

[0113] Further, on the basis of the high-precision monitoring network composed of random monitoring points and fixed monitoring points, in the step S5, only the atmospheric pollutant data of the fixed monitoring points and the random monitoring points based on the remote sensing images of the fixed monitoring points and the random monitoring points need to be obtained, so that the representative data of each monitoring unit can be collected through the accurate selection of the monitoring points, not only avoiding the large amount of data redundancy caused by the traditional method of uniformly and fully covering the data acquisition, but also reducing the difficulty of subsequent data processing process, and significantly improving the processing efficiency of the atmospheric pollutant diffusion model and the timeliness of the monitoring results. In fact, in the process of atmospheric pollutant concentration monitoring and prediction, the difference of atmospheric pollutants in a short distance and a small area is often not obvious, so it is not necessary to fully cover the remote sensing images of the entire monitoring area. The traditional grid method often needs to splice the images due to the large image coverage area, resulting in large amount of data processing, complex processing process and low efficiency, which directly affects the cycle and timeliness of atmospheric pollution prediction.

[0114] In the present application, by setting the fixed monitoring points and the random monitoring points and constructing the model based on the information of the fixed monitoring points and the random monitoring points, the problems of limited point positions, insufficient spatial coverage, difficulty in fully reflecting the dynamic migration process of pollutants in a large range of atmosphere, and lack of coupling ability with terrain, topography and real-time weather conditions in the traditional fixed monitoring method are avoided. At the same time, the prediction accuracy is also taken into account.

[0115] Further, in the step S5, after obtaining the prediction result of the atmospheric pollution situation of the monitoring area in the future period of time, the prediction result of the atmospheric pollution situation can be analyzed and processed first, and the specific process includes the steps of:

[0116] S51, obtaining the prediction result of the pollutant distribution, determining the boundary and center of the pollutant distribution;

[0117] S52, taking the center of the pollutant distribution as the starting point and the boundary line of the pollutant distribution as the ending point, drawing a pollutant distribution vector diagram C corresponding to the prediction result of the future atmospheric pollution situation.

[0118] As some examples of the present application, the process of determining the pollutant distribution boundary and center according to the pollutant distribution prediction result can be: first, the pollutant concentration of each region is compared with the preset concentration threshold, and the pollutant concentration exceeding the concentration threshold is the pollution area, and vice versa, and after determining the pollution area, the centroid or geometric center of the pollution area is calculated and taken as the pollutant distribution center point of the region.

[0119] Further, the step S6 specifically comprises:

[0120] S61, obtaining the atmospheric pollution situation prediction result of the current moment output by the atmospheric pollutant diffusion model;

[0121] S62, obtaining the atmospheric pollution situation measurement result of the current moment based on the remote sensing image monitoring;

[0122] S63, respectively analyzing the atmospheric pollution situation prediction result and the atmospheric pollution situation measurement result of the current moment, determining the pollutant distribution boundary and center, and then taking the pollutant distribution center as the starting point and the pollutant distribution boundary line as the end point, drawing the pollutant distribution vector diagram A corresponding to the current atmospheric pollution situation prediction result and the pollutant distribution vector diagram B corresponding to the current atmospheric pollution situation measurement result;

[0123] S64, determining the prediction deviation of the atmospheric pollutant diffusion model based on the pollutant distribution vector diagram A and B results;

[0124] S65, correcting the atmospheric pollution situation prediction result of the monitoring area in the future period of time according to the determined prediction deviation, to obtain the corrected atmospheric pollution situation prediction result.

[0125] Preferably, in the steps S52 and S63, the drawn pollutant distribution vector diagram comprises a plurality of one-to-one corresponding and same-angle sub-vectors, and specifically as shown in Figure 2 The sub-vectors are drawn along the east, west, south and north directions, or the north direction is taken as 0°, and the sub-vectors are drawn at intervals of 30°, 45° or 60°, Figure 2 The sub-vectors drawn at intervals of 45° are shown It can be understood that the more sub-vectors drawn, the more accurate the prediction result obtained through correction will be, but the calculation amount of the correction process will be larger, therefore, the number of sub-vectors should be appropriate, and preferably 6-12.

[0126] It should be noted that in the present application, the historical moment corresponds to the last monitoring period, the current moment corresponds to the current monitoring period, and the future period of time corresponds to the next monitoring period, and the time intervals between the monitoring periods are equal.

[0127] As some examples of the present application, in the step S62, the current atmospheric pollution situation measurement result based on remote sensing image monitoring can be the atmospheric pollution situation measurement result obtained by the present application or others through remote sensing image full coverage in real time. When the present application uses the remote sensing image monitored by itself to output the atmospheric pollution situation measurement result of the current time, the pollutant distribution area can be obtained according to the atmospheric pollution situation prediction result of the corresponding current time output by the atmospheric pollution diffusion model in step S61, and then the atmospheric pollution situation measurement result meeting the subsequent use requirement is obtained by dynamically adjusting and appropriately expanding the remote sensing image coverage area with the predicted pollutant distribution area as the reference, and specifically, the actual pollutant cloud corresponding to the pollutant cloud in the prediction result is appropriate. It can be understood that the atmospheric pollution situation measurement result to be obtained only needs to cover the pollution area, and full coverage detection of the monitoring area is not necessary.

[0128] Further, in the step S64, the process of determining the prediction deviation of the atmospheric pollution diffusion model based on the pollutant distribution vector graphs A and B is as follows:

[0129] S641, the deviation coefficient a of the pollutant distribution area in the pollutant distribution vector graphs A and B is calculated, wherein the deviation coefficient a = (1-K)*100%, the K is the coincidence ratio, and the coincidence ratio K = the coincident area of the vector graphs A and B / the total area of the pollutant distribution area in the graph B;

[0130] S642, the relative sizes of the deviation coefficient a and the preset threshold a 阈 are compared:

[0131] If a < a 阈 , the deviation coefficient a is smaller, indicating that the consistency of the atmospheric pollution situation prediction result and the atmospheric pollution situation measurement result corresponding to the current time is higher, the atmospheric pollution situation prediction result of the next period output by the atmospheric pollution diffusion model does not need to be corrected, and the step S65 is continued to directly output the atmospheric pollution situation prediction result of the atmospheric pollution diffusion model;

[0132] If a ≥ a 阈 , the deviation coefficient a is larger, indicating that the consistency of the atmospheric pollution situation prediction result and the atmospheric pollution situation measurement result corresponding to the current time is lower, the atmospheric pollution situation prediction result of the next period output by the atmospheric pollution diffusion model needs to be corrected, and the step S643 is continued to be executed;

[0133] S643, the center point of the pollutant distribution area in the pollutant distribution vector graph A is taken as the starting point, the center point of the pollutant distribution area in the pollutant distribution vector graph B is taken as the ending point, and the prediction deviation vector between the center points of the pollutant distribution areas in the pollutant distribution vector graphs A and B is drawn ;

[0134] Then, the pollutant distribution vector diagram A is translated from the start point to the end point of the vector , and a translated pollutant distribution vector diagram A' is obtained.

[0135] The difference between each component vector in the pollutant distribution vector diagram B and the pollutant distribution vector diagram A' is calculated respectively, and a set of component vector deviation vectors corresponding to each angle is obtained.

[0136] The predicted deviation vector between the center points obtained by the above calculation , the component vector deviation vector can represent the predicted deviation of the atmospheric pollutant diffusion model.

[0137] As some examples of the present application, the size of the preset threshold α 阈 and the subsequent α 阈 ' can be set according to the accuracy requirement. Generally, the higher the accuracy requirement, the smaller the values of the preset threshold α 阈 and α 阈 '.

[0138] Further, the step S65 includes:

[0139] When α < α 阈 , the atmospheric pollution prediction result of the atmospheric pollutant diffusion model is directly outputted.

[0140] When α ≥ α 阈 , first, the center point of the pollutant distribution vector diagram C is translated along the vector , then the component vector in the pollutant distribution vector diagram C is added to the corresponding component vector deviation vector to obtain a corrected component vector, and the boundary of the pollutant distribution vector diagram C is adjusted according to the corrected component vector to obtain a pollutant distribution vector diagram C', and the pollutant distribution vector diagram C' is outputted as the final atmospheric pollution prediction result.

[0141] Further, in the process of formulating the pollutant distribution vector diagram C', when the corrected component vector is determined, the boundary line of the pollutant distribution vector diagram C can be first adjusted according to the corrected component vector, and the boundary line near the adjusted area is processed for smooth transition.

[0142] As an example of the present application, the above process is described in detail for the correction process of a pollution cloud group containing only one pollution cloud, when the pollution distribution vector diagram contains multiple pollution cloud groups, the boundaries and the center points of each pollution cloud group can also be determined respectively according to the above process, and then the correction is performed respectively in a similar manner.

[0143] Further, the step S65 further comprises: when α ≥ α 阈 , judging whether the prediction result needs to be corrected in depth, and the specific process is as follows:

[0144] Firstly, the pollution distribution vector diagram A is translated along the prediction deviation vector , to obtain a pollution distribution vector diagram A', so that the center point of the translated pollution distribution vector diagram A' coincides with the center point in the pollution distribution vector diagram B, and then the deviation coefficient α' of the pollution distribution area in the pollution distribution vector diagrams A' and B is calculated, wherein the deviation coefficient α' = (1-K') * 100%, and the K' is the coincidence ratio, and the coincidence ratio K = the overlapping area of the vector diagrams A' and B / the total area of the pollution distribution area in the diagram B;

[0145] Then, the relative size of the deviation coefficient α' and the preset threshold α 阈 ' is compared:

[0146] If α' < α 阈 ', the pollution distribution vector diagram C does not need to be corrected in depth, at this time, the center point of the pollution distribution vector diagram C can be translated along the vector firstly, and then the partial vector in the pollution distribution vector diagram C is added to the corresponding partial vector deviation vector to obtain the corrected partial vector, and the boundary of the pollution distribution vector diagram C is adjusted according to the corrected partial vector to obtain a pollution distribution vector diagram C', and the pollution distribution vector diagram C' is output as the final atmospheric pollution prediction result;

[0147] If α' ≥ α 阈 ', the pollution distribution vector diagram C needs to be corrected in depth, and the depth correction process is as follows:

[0148] Firstly, the boundary line of the pollution distribution vector diagram B is analyzed to obtain extreme points on the boundary line, and the extreme points include maximum values and minimum values, and then the extreme value vectors of the pollution distribution vector diagram B are drawn respectively with the center point of the pollution distribution vector diagram B as the starting point and each extreme point on the boundary line as the ending point, and the vector set composed of the extreme value vectors is called an extreme value vector set;

[0149] Then, the extreme value vectors corresponding to the angles of the extreme value vectors are drawn on the pollution distribution vector diagram A;

[0150] Then, the difference between each extreme value vector in the pollutant distribution vector diagram B and the pollutant distribution vector diagram A is calculated respectively to obtain a set of extreme value deviation vectors corresponding to each extreme value

[0151] And the predicted deviation vector between the center points calculated , the sub-vector deviation vector And the extreme value deviation vector Indicate the predicted deviation of the atmospheric pollutant diffusion model;

[0152] When α' ≥ α 阈 , first draw the extreme value vector corresponding to the angle of the extreme value vector on the pollutant distribution vector diagram C, then translate the center point of the pollutant distribution vector diagram C along the vector , and then add the sub-vector in the pollutant distribution vector diagram C and the corresponding sub-vector deviation vector After the addition, the corrected sub-vector is obtained, and the extreme value vector in the pollutant distribution vector diagram C and the corresponding extreme value deviation vector After the addition, the corrected extreme value vector is obtained.

[0153] And adjust the boundary of the pollutant distribution vector diagram C according to the corrected sub-vector and extreme value vector to obtain the pollutant distribution vector diagram C', and output the pollutant distribution vector diagram C' as the final atmospheric pollution prediction result.

[0154] As an example of the present application, the present application has no limitation on the remote sensing image used, which can be a remote sensing image obtained by regularly observing the earth through a multi-spectral, multi-spectral sensor meteorological satellite, an environmental monitoring satellite, or a remote sensing image obtained by using an airplane, a drone, a spectral imager, a laser radar and other sensors.

[0155] Preferably, the area of a single image of the remote sensing image used by the present application should be not less than 8km 2 , and the spatial resolution should be not less than 1m / pixel.

[0156] Compared with the existing data-driven model (LSTM), the prediction deviation correction mechanism based on the pollutant distribution vector diagram proposed by the present application comprehensively corrects the model prediction result from multiple dimensions such as the pollutant distribution center, the boundary, the sub-vector and the extreme value vector, which has the following advantages:

[0157] (1) By calculating the deviation vector of the measured and predicted pollution area center point corresponding to the current time, and accordingly translating the prediction result of the next period, the deviation of the prediction result from the actual distribution in geographical space can be significantly reduced, and the center drift problem caused by prediction error of the traditional model can be solved;

[0158] (2) The boundary profile and key inflection point are respectively corrected by using the partial vector and extreme value vector, the flexible adjustment of the pollution map profile is realized instead of the deformation in a single scale, and the geometric fidelity and fitting degree of the prediction map boundary are improved;

[0159] (3) Based on the predicted value and the actual value at the current time, by introducing the deviation vector of the center point and the partial vector and the extreme value vector, the rolling correction and dynamic optimization of the predicted value of the next period are realized, which has the advantages of simple calculation process, no risk of "error amplification", not affected by external conditions, no need to frequently retrain, good timeliness, etc.;

[0160] (4) The deviation coefficient α / α'mechanism and the quantized prediction error are introduced, a clear and measurable way is provided to evaluate the difference between the prediction result and the actual observation, and whether to correct, the depth and amplitude of correction are determined;

[0161] (5) The error judgment control mechanism with adjustable threshold is formed by presetting parameters such as threshold α threshold, α threshold', which allows users to flexibly set the tolerance error range according to the accuracy requirement, ensures the balance between resources and accuracy, avoids blind correction, and improves the system efficiency;

[0162] (5) According to the size of the deviation coefficient α / α', a multi-level and multi-granularity adaptive correction strategy is realized, the ability to independently correct multiple pollution clouds is provided, which is suitable for the complex and multi-source situation in real atmospheric pollution, and the adaptability and expansibility of the model are enhanced;

[0163] (6) Through the vector grid with uniformly distributed angles, the model can more comprehensively capture the pollution diffusion deviation in all directions, and effectively avoid directional omission;

[0164] (7) The present application is a post-correction mechanism, which can not change the original pollution diffusion model structure, and can be independently embedded in the existing platform as a plug-in or post-processing module, and has good compatibility and low modification cost.

[0166] The high-precision atmospheric pollution prediction method according to the present application is used to predict the atmospheric pollution situation in Ningbo, Zhejiang province, and the process is as follows:

[0167] (1) First, the digital topographic map of the monitoring area is obtained based on the public data, and the monitoring area is divided into 187 monitoring units based on the public topographic map, land cover type and vegetation distribution data;

[0168] (2) Then determine the potential atmospheric pollutant emission sources in the monitoring area according to the disclosed information, and mark them as fixed monitoring points, a total of 768 fixed monitoring points are marked;

[0169] (3) Establish a dynamic evaluation mechanism for gas diffusion anomaly factors, and determine the number of random monitoring points in each monitoring unit according to the size of the gas diffusion anomaly factor:

[0170] When the gas diffusion anomaly factor is greater than 0.8, the basic number of random monitoring points is 7;

[0171] When the gas diffusion anomaly factor is between 0.2 and 0.8, the basic number of random monitoring points is 4;

[0172] When the gas diffusion anomaly factor is less than 0.2, the basic number of random monitoring points is 2;

[0173] And verify the sample space independence by the Moran index, realize the precision control of sampling error ≤8%, and verify the spatial representativeness (|I|≤0.2) by the Moran index.

[0174] Then adjust the number of random monitoring points based on the area of the monitoring unit:

[0175] When the monitoring unit is greater than 100km 2 , the final number of random monitoring points is: basic number * 3;

[0176] When the area of the monitoring unit is between 50-100km 2 , the final number of random monitoring points is: basic number * 1.5;

[0177] When the area of the monitoring unit is less than 50km 2 , the final number of random monitoring points is: basic number * 1;

[0178] The final number of random monitoring points is a total of 1334;

[0179] (4) Based on the geographic information, historical meteorological information and historical pollutant information of the fixed monitoring points and random monitoring points, an atmospheric pollutant diffusion model (Lagrangian envelope model) of the monitoring area is constructed;

[0180] (5) Based on the remote sensing images of the fixed monitoring points and random monitoring points, real-time atmospheric pollutant data of the fixed monitoring points and random monitoring points are obtained, and input into the atmospheric pollutant diffusion model, to predict the future atmospheric pollution situation, and obtain the prediction result of the atmospheric pollution situation of the monitoring area in the future period of time;

[0181] (6) determining the prediction deviation of the atmospheric pollutant diffusion model based on the current monitoring data and the prediction result of the historical atmospheric pollution situation, and correcting the prediction result of the atmospheric pollution situation of the monitoring area in a future period of time according to the determined prediction deviation to obtain the corrected prediction result of the atmospheric pollution situation.

[0182] The obtained prediction result is compared with the measured value and the traditional gridding, full-coverage modeling and remote sensing data fusion technology respectively, and the results as shown in Figure 2 are obtained, wherein the red dotted line represents the measured value of the monitoring point, the black dot represents the corrected prediction value of the monitoring point, and the green triangle represents the prediction value obtained by the traditional gridding, full-coverage modeling and remote sensing data fusion technology:

[0183] According to Figure 2 It can be known that the high-precision atmospheric pollution prediction method can obtain the remote sensing image of the selected area by the monitoring point sampling method, construct a prediction model, and obtain the atmospheric pollution prediction result with high timeliness and high accuracy by correcting the prediction value.

[0184] In addition, compared with the traditional full-coverage analysis and prediction method, the embodiment can effectively reduce the data processing amount by fixed-point and sampling monitoring of the specific monitoring area, for example, for a monitoring area of 10,000 square kilometers: the number of remote sensing images can be reduced from nearly 10,000 to about 2,000, greatly reducing the data processing amount, and avoiding the splicing of data, which can simplify the data processing process and improve the processing efficiency, reduce the prediction period from 3-8 hours to 30-60 minutes, and greatly improve the timeliness of atmospheric pollution.

[0185] Furthermore, the present application can ensure the comprehensiveness of monitoring and highlight the focus of monitoring by combining fixed monitoring points and random monitoring points, thereby improving the monitoring accuracy.

[0186] The embodiments of the present application are described above in combination with the drawings, and the embodiments and features in the embodiments in the present application can be combined with each other without conflict, and the present application is not limited to the above-mentioned specific embodiments, which are only illustrative and not restrictive, and a person of ordinary skill in the art can make many forms under the inspiration of the present application without departing from the scope of the present application and the protection scope of the claims.

Claims

1. A high-precision method for predicting air pollution, characterized in that, Including the following steps: S1, the monitoring area is divided into several monitoring units based on land use, land cover type and / or vegetation distribution data; S2, identify potential sources of air pollutant emissions in the monitoring area and mark them as fixed monitoring points; S3. Based on the accuracy requirements for air pollutant prediction, the number of random monitoring points in each monitoring unit is determined according to the area of ​​each monitoring unit and the gas diffusion variation factor. Random monitoring points and fixed monitoring points are coupled to establish a high-precision monitoring network. S4. Construct an atmospheric pollutant diffusion model for the monitoring area based on the geographical information, historical meteorological information, and historical pollutant information of the fixed and random monitoring points. S5. Based on remote sensing images of fixed and random monitoring points, real-time air pollutant data of fixed and random monitoring points are obtained and input into the air pollutant diffusion model to predict future air pollution and obtain the prediction results of air pollution in the monitoring area for a period of time in the future. S6. Based on current monitoring data and historical air pollution prediction results, determine the prediction bias of the air pollutant diffusion model, and correct the air pollution prediction results for the monitoring area in the future period according to the determined prediction bias, so as to obtain the corrected air pollution prediction results.

2. The high-precision air pollution prediction method according to claim 1, characterized in that, Step S2 specifically includes: S21, Obtain or construct geographic information of the monitoring area; S22, Identify and classify potential sources of air pollution; S23 identifies potential sources of air pollutant emissions as candidate sites for fixed monitoring points; S24, merge and screen the candidate fixed monitoring points to determine the final fixed monitoring points; S25, Determine the specific location of the fixed monitoring point.

3. The high-precision air pollution prediction method according to claim 1, characterized in that, In step S3, the process of determining the number of random monitoring points in each monitoring unit is as follows: (1) Establish a dynamic assessment mechanism for gas diffusion variation factors, wherein the gas diffusion variation factor of each monitoring unit = standard deviation of historical pollutant content / average value of historical pollutant content; (2) Determine the basic number of random monitoring points for each monitoring unit based on the magnitude of the gas diffusion variation factor; (3) Adjust the basic number of random monitoring points based on the area of ​​each monitoring unit.

4. The high-precision air pollution prediction method according to claim 1, characterized in that, In step S5, after obtaining the predicted air pollution situation in the monitoring area for a future period, the predicted air pollution situation is first analyzed and processed. The specific process includes the following steps: S51, Obtain the pollutant distribution prediction results and determine the pollutant distribution boundary and center; S52, starting from the pollutant distribution center and ending at the pollutant distribution boundary line, draw a pollutant distribution vector map C corresponding to the predicted future air pollution situation.

5. The high-precision air pollution prediction method according to claim 4, characterized in that, Step S6 specifically includes: S61, Obtain the atmospheric pollution situation prediction results corresponding to the current moment from the output of the atmospheric pollutant diffusion model; S62, Obtain the measured results of air pollution at the current moment based on remote sensing image monitoring; S63, analyze the predicted air pollution situation and the measured air pollution situation at the current time respectively, determine the pollutant distribution boundary and center, and then draw the pollutant distribution vector map A corresponding to the current air pollution situation and the pollutant distribution vector map B corresponding to the current air pollution situation, with the pollutant distribution center as the starting point and the pollutant distribution boundary line as the ending point. S64, Determine the prediction bias of the atmospheric pollutant diffusion model based on the results of pollutant distribution vector maps A and B; S65, based on the determined prediction deviation, correct the prediction results of air pollution in the monitoring area for a future period of time to obtain the corrected prediction results of air pollution.

6. The high-precision air pollution prediction method according to claim 5, characterized in that, In steps S52 and S63, the drawn pollutant distribution vector map includes multiple one-to-one corresponding sub-vectors with the same angle. The sub-vectors are drawn at predetermined angular intervals, with the pollutant distribution center as the starting point and the pollutant distribution boundary line as the ending point.

7. The high-precision air pollution prediction method according to claim 6, characterized in that, In step S64, the process of determining the prediction bias of the atmospheric pollutant diffusion model based on the pollutant distribution vector map results A and B is as follows: S641, calculate the deviation coefficient α of the pollutant distribution area in the pollutant distribution vector maps A and B, wherein the deviation coefficient α = (1-K)*100%, and K is the overlap ratio, which is K = the overlap area of ​​vector maps A and B / the total area of ​​the pollutant distribution area in map B. S642, compare the deviation coefficient α and the preset threshold α 阈 Relative size: If α < α 阈 Instead of correcting the air pollution prediction results for the next period output by the air pollutant diffusion model, step S65 is executed to directly output the air pollution prediction results of the air pollutant diffusion model. If α≥α 阈 The atmospheric pollution forecast results for the next period output by the atmospheric pollutant diffusion model need to be corrected, and step S643 should be continued. S643, using the center point of the pollutant distribution area in pollutant distribution vector map A as the starting point and the center point of the pollutant distribution area in pollutant distribution vector map B as the ending point, draw the prediction deviation vector between the center points of the pollutant distribution areas in pollutant distribution vector maps A and B. Then along the vector The pollutant distribution vector map A is translated from the starting point to the ending point to obtain the translated pollutant distribution vector map A'; Calculate the differences between each component vector in the pollutant distribution vector map B and the pollutant distribution vector map A' respectively, to obtain a set of component vector deviation vectors corresponding to each angle. And use the calculated prediction deviation vector between the center points Component vector deviation vector This indicates the prediction bias of the atmospheric pollutant diffusion model.

8. The high-precision air pollution prediction method according to claim 7, characterized in that, Step S65 includes: When α < α 阈 At that time, the air pollution prediction results of the air pollutant diffusion model are directly output; When α≥α 阈 First, the center point of the pollutant distribution vector diagram C is aligned with the vector... After translation, the component vectors in the pollutant distribution vector map C are compared with the corresponding component vector deviation vectors. The corrected component vectors are obtained by adding them together. The boundaries of the pollutant distribution vector map C are then adjusted according to the corrected component vectors to obtain the pollutant distribution vector map C'. The pollutant distribution vector map C' is then output as the final prediction result of air pollution.

9. The high-precision air pollution prediction method according to claim 8, characterized in that, Step S65 further includes: when α≥α 阈 Then, determine whether a deep correction is needed for the prediction results: First, map the pollutant distribution vector A along the prediction deviation vector. The pollutant distribution vector map A' is translated to obtain a vector map of pollutant distribution A', such that the center point of the translated vector map A' coincides with the center point of the pollutant distribution vector map B. Then, the deviation coefficient α' of the pollutant distribution area in the vector maps A' and B is calculated, where the deviation coefficient α' = (1-K')*100%, and K' is the overlap ratio, which is equal to the overlap area of ​​vector maps A' and B / the total area of ​​the pollutant distribution area in map B. Then compare the deviation coefficient α' with the preset threshold α. 阈 Relative size of ': If α' < α 阈 ', No depth correction is required for the pollutant distribution vector map C; If α'≥α 阈 ', It is necessary to perform depth correction on the pollutant distribution vector map C.

10. The high-precision air pollution prediction method according to claim 9, characterized in that, The depth correction process is as follows: First, the boundary line of the pollutant distribution vector map B is analyzed to obtain the extreme points on the boundary line. Then, the extreme value vectors of the pollutant distribution vector map B are drawn with the center point of the pollutant distribution vector map B as the starting point and each extreme point on the boundary line as the ending point. The vector set composed of the extreme value vectors is called the extreme value vector set. Then, draw the extreme value vectors on the pollutant distribution vector map A, which correspond one-to-one with the angles of the extreme value vectors; Next, the differences between the extreme value vectors in pollutant distribution vector map B and pollutant distribution vector map A are calculated respectively, resulting in a set of extreme value deviation vectors corresponding to each extreme value. In α'≥α 阈 First, draw the extreme value vectors on the pollutant distribution vector map C, which correspond one-to-one with the angles of the extreme value vectors. Then, draw the center point of the pollutant distribution vector map C along the vector... After translation, the component vectors in the pollutant distribution vector map C are compared with the corresponding component vector deviation vectors. ...After adding them together, we obtain the corrected component vector, which is then used to compare the extreme value vector in the pollutant distribution vector map C with the corresponding extreme value deviation vector. The corrected extreme value vector is obtained by adding them together; The boundaries of the pollutant distribution vector map C are adjusted according to the corrected component vectors and extreme value vectors to obtain the pollutant distribution vector map C', and the pollutant distribution vector map C' is output as the final prediction result of air pollution.

Images