A method and system for improving the effectiveness of medium- and long-term air quality forecasting

Abstract

This invention belongs to the field of air quality numerical forecasting technology, and relates to a method and system for improving the performance of medium- and long-term air quality forecasts. The method includes: determining the weather patterns for each sub-region and each buffer zone at each time period; determining the optimal parameterization scheme combination under different weather patterns; determining the future weather patterns for each time period within a future timeframe; determining the ensemble forecast field for the previous time period; determining the ensemble forecast field for the next time period; determining the ensemble forecast field for all time periods in the region; determining the ensemble forecast field for all time periods in all sub-regions and buffer zones; determining the temporal and spatial coupled meteorological field of the inner nested regions of each sub-region; determining the meteorological field of the large region; and determining the meteorological field of each nested region. It utilizes weather classification, multiple initial field perturbations, and multi-regional, multi-time period forecast result coupling techniques to improve the performance of medium- and long-term air quality forecasts by enhancing the medium- and long-term forecasting performance of meteorological models.

Description

Technical Field

[0001] This invention belongs to the field of numerical air quality forecasting technology, and relates to a method and system for improving the effectiveness of medium- and long-term air quality forecasting. Background Technology

[0002] With the development of numerical models, numerical modeling has become an important method in pollutant concentration prediction. In practical operational forecasting, three generations of air quality models are mainly used to construct air quality forecasting systems. The mainstream three generations of air quality models today include the MODEL-3 / CAMQ (Community Multiscale Air Quality Modeling System), the CAMx (Community Atmosphere Model), the WRF-CHEM model, and the NAQPMS (Nestled Grid Air Quality Prediction System). These three generations of air quality models incorporate complex and comprehensive gas-phase chemical and photochemical mechanisms, and have good simulation and prediction capabilities for the spatiotemporal distribution of pollutants.

[0003] However, regardless of which third-generation air quality model is used to construct the air quality forecasting system, a complete air quality forecasting system mainly consists of three parts: an emission source processing system (providing emission source input), a meteorological model (providing meteorological fields such as temperature, pressure, humidity, and wind), and an air quality model (simulating the spatiotemporal distribution of pollutants). Therefore, the meteorological field is a crucial input to the air quality forecasting system and significantly impacts its forecast accuracy. Consequently, the accuracy of the meteorological model's predictions has a substantial influence on the accuracy of air quality forecasts.

[0004] Meanwhile, meteorological models are highly nonlinear systems. The initial fields and physical parameterization schemes in the models are approximations of actual physical processes, and both contain certain errors. These errors will nonlinearly discretize as time integrates. For example... Figure 1 As shown, at the initial reporting time t0, the error between the initial atmospheric state and the actual atmospheric state is very small. However, as the model integration time increases, this error is amplified, and by time t0+n, the difference between the predicted and actual atmospheric states becomes very large. This indicates that in medium- and long-term forecasts, the large forecasting error of meteorological models affects the forecasting performance of air quality models. Therefore, improving the performance of medium- and long-term forecasts by meteorological models can improve the forecasting performance of air quality models.

[0005] In meteorology, ensemble forecasting is primarily used to improve the effectiveness of medium- and long-term forecasts. Chinese invention patent 202110873280.7 first obtains a relatively accurate meteorological field through ensemble forecasting of meteorological models, and then uses this meteorological field to drive an air quality model to improve its forecasting performance. This patent considers the impact of meteorological model forecast accuracy on air quality models and addresses the influence of large-error meteorological forecasts on air quality forecasts, which is significant for improving the forecasting effectiveness of air quality models. However, this patent does not consider the significant differences in the optimal parameterization scheme combination for different time periods. While the meteorological forecast field obtained through ensemble forecasting is generally better than that of a single deterministic forecast, its forecasting performance in each time period is lower than that of its optimal ensemble forecast members. If the optimal parameterization scheme combination for different time periods can be obtained, it is unnecessary to construct physical process ensemble members. Furthermore, this patent does not consider how to process the meteorological field of the simulated area of ​​interest and surrounding areas. The meteorological field required by meteorological departments only needs to reflect the local forecasting performance to be used in the actual local meteorological forecasting system. Pollutant transport has a significant impact; for example, in a medium-sized city, particulate matter transport contributes approximately 60%–70%, and ozone transport contributes approximately 70%–80%. Therefore, air quality models must consider the contribution of transport, requiring the meteorological model to not only provide good forecasts for the local area but also for surrounding areas. If the forecast for surrounding areas is poor, the air quality model's forecast error for those areas will be larger, leading to a larger error in the transport contribution from surrounding areas to the local area, ultimately affecting the air quality model's forecast accuracy for the local area. Assuming a relatively accurate initial field, the main factor influencing the forecast accuracy for each region is the physical parameterization scheme of the meteorological model.

[0006] Meteorological models incorporate numerous physical parameterization schemes, which primarily describe various atmospheric physical processes through mathematical modeling. For example, the WRF meteorological model includes over ten types of parameterization schemes, such as cumulus convection, boundary layer, microphysical, and land surface processes. Each type offers multiple options; for instance, the boundary layer schemes include MYJ, MRF, ACM2, QNSE, and MYNN, among others. This necessitates selecting the most suitable scheme from various physical parameterization schemes based on local conditions. For instance, it's necessary to choose the scheme with the best local simulation and prediction performance from various boundary layer schemes, and vice versa. However, long-term testing results show that for any given region, there is no single scheme with consistently good forecasting performance.

[0007] In summary, existing meteorological models have significant errors in medium- and long-term forecasts, making it difficult to meet the needs of medium- and long-term air quality forecasting. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention proposes a method to improve the effectiveness of medium- and long-term air quality forecasts. This method utilizes weather classification technology, initial field multiple perturbation technology, and multi-regional, multi-time period simulation result coupling technology. By enhancing the accuracy of medium- and long-term forecasts in meteorological models, it further improves the forecasting effectiveness of air quality models.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for improving the effectiveness of medium- and long-term air quality forecasts, characterized by comprising the following steps:

[0011] 1) Divide the large area into multiple sub-regions and set up buffer zones between two adjacent sub-regions. Based on weather classification technology and combined with historical reanalysis meteorological data over many years, identify the weather patterns of each sub-region and each buffer zone at each time period.

[0012] 2) Use meteorological models and reanalysis data to simulate weather over many years to identify the optimal combination of parameterization schemes for each sub-region and buffer zone under different weather types;

[0013] 3) Based on meteorological model-driven field meteorological data and combined with weather typology technology, identify the future weather patterns of each sub-region and each buffer zone for a period of time in the future;

[0014] 4) Forecast the weather of a certain area in each sub-region and each buffer region for the previous period to obtain the ensemble forecast field for the previous period.

[0015] 5) Based on the ensemble forecast field of the previous time period, the weather of the next time period adjacent to the previous time period is predicted to obtain the ensemble forecast field of the next time period.

[0016] 6) Repeat step 5) to obtain the ensemble forecast field for all time periods in the region;

[0017] 7) Repeat steps 4) to 6) for each sub-region and other regions in each buffer region except the sub-region itself to obtain the ensemble forecast field for all time periods of all sub-regions and buffer regions;

[0018] 8) Perform temporal and spatial coupling on the ensemble forecast fields of all time periods of all sub-regions and the inner nested regions of the buffer region to obtain the temporal and spatial coupled meteorological fields of the inner nested regions of each sub-region;

[0019] 9) The meteorological fields of the inner nested regions of each sub-region, after temporal and spatial coupling, are spliced ​​together to obtain the meteorological field of the large region;

[0020] 10) The large region is truncated according to the geographical range of each nested region of the target region under the air quality model, and the meteorological field of each nested region of the target region under the air quality model is obtained based on the meteorological field of the truncated large region.

[0021] Preferably, step 4) specifically includes:

[0022] 41) Perturb the initial field of a certain region in each sub-region and each buffer region in the previous time period to generate multiple first initial fields and construct a series of first initial value set forecast members.

[0023] 42) Based on the future weather pattern of the region in the previous period, the optimal parameterization scheme combination of the region under the weather pattern is used and combined with the forecast members of the first initial value set to predict the future weather, so as to obtain the first meteorological forecast field of each of the first initial value forecast members of the region in the previous period.

[0024] 43) The ensemble average of the first meteorological forecast fields of each first initial value forecast member in the previous period of the region is used to obtain the ensemble forecast field of the region in the previous period.

[0025] Preferably, step 5) specifically includes:

[0026] 51) Perform ensemble averaging on the ensemble forecast field of the previous time period in the region and use the result of the ensemble averaging as the initial field of the next time period in the region.

[0027] 52) Perturb the initial field of the region in the next time period to generate multiple second initial fields and construct a series of second initial value set forecast members;

[0028] 53) Based on the future weather pattern of the region in the next period, use the optimal parameterization scheme combination of the region under the weather pattern and combine it with the second initial value set forecast members to predict the future weather, and obtain the second meteorological forecast field of each second initial value forecast member of the region in the next period.

[0029] 54) The ensemble average of the second meteorological forecast fields of each second initial value forecast member for the next period of time in the region is used to obtain the ensemble forecast field for the next period of time in the region.

[0030] Preferably, in step 4), the ensemble forecast field of the region in the previous period includes an ensemble forecast field that exceeds the previous period by a certain time, and in step 5), the ensemble forecast field of the region in the next period includes an ensemble forecast field that precedes the next period by a certain time, so that the ensemble forecast field of the previous period and the ensemble forecast field of the next period have an overlap period.

[0031] Preferably, in step 8), the temporal coupling of the ensemble forecast fields for all time periods of all sub-regions and the inner nested regions of the buffer region involves fitting the ensemble forecast fields during the overlap period of the previous and subsequent time periods using a sine and cosine probability density function. The fitting formula is as follows:

[0032]

[0033] In the formula, For the ensemble forecast field of time t within the overlap period, This is the ensemble forecast field for the time period preceding time t within the overlap period. This is the ensemble forecast field for the time period following time t within the overlap period, where n is the overlap period and t represents any time within the overlap period, taking values ​​of 1, 2, 3, ..., n. is a constant, where, The radian is equal to 180°. , .

[0034] Preferably, the spatial fitting in step 8) specifically includes:

[0035] 81) Combine the meteorological fields after time coupling of the inner nested regions of the buffer zones between the east-west sub-regions and the meteorological fields after time coupling of the inner nested regions of each sub-region to perform east-west fitting, and obtain the meteorological fields after east-west fitting of the inner nested regions of each sub-region.

[0036] 82) By combining the meteorological field after temporal coupling of the inner nested region of the buffer zone between the north-south sub-regions and the meteorological field after east-west fitting of the inner nested region of each sub-region, the meteorological field after temporal and spatial coupling of the inner nested region of each sub-region is obtained by performing north-south fitting.

[0037] Preferably, step 10) specifically includes:

[0038] 101) Based on the geographical range of the inner nested region of the target area under the air quality model, the large area is extracted, and the meteorological field of the extracted large area is used as the meteorological field of the inner nested region of the target area under the air quality model.

[0039] 102) Based on the geographical range of other nested regions of the target area under the air quality model, the large area is extracted, and based on the meteorological field of the extracted large area, the meteorological field of other nested regions of the target area under the air quality model is processed to obtain the meteorological field of other nested regions of the target area under the air quality model.

[0040] Furthermore, the present invention also provides a system for improving the effectiveness of medium- and long-term air quality forecasts, characterized in that it comprises:

[0041] The weather pattern determination module is used to divide a large area into multiple sub-regions and set buffer zones between two adjacent sub-regions. Based on weather patterning technology and combined with reanalysis meteorological data from many years of history, it identifies the weather pattern of each sub-region and each buffer zone for each time period.

[0042] The optimal parameterization scheme combination determination module is used to simulate weather over many years using meteorological models and reanalysis data, and to identify the optimal parameterization scheme combination for each sub-region and buffer zone under different weather types.

[0043] The future weather pattern determination module is used to identify the future weather pattern of each sub-region and each buffer zone for a period of time in the future, based on meteorological data from the driving field of the meteorological model and weather classification technology.

[0044] The previous period ensemble forecast field determination module is used to predict the weather of a certain area in each sub-region and each buffer region in the previous period, and obtain the previous period ensemble forecast field of that area.

[0045] The ensemble forecast field determination module for the next time period is used to predict the weather of the next time period adjacent to the previous time period based on the ensemble forecast field of the previous time period, and obtain the ensemble forecast field of the next time period.

[0046] A regional ensemble forecast field determination module is used to obtain the ensemble forecast field for all time periods of the region based on the ensemble forecast field determination module for the previous time period and the ensemble forecast field determination module for the next time period.

[0047] The temporal and spatial coupling module is used to perform temporal and spatial coupling on the ensemble forecast fields of all time periods of each sub-region and the inner nested regions of each buffer region, so as to obtain the temporal and spatial coupled meteorological fields of the inner nested regions of each sub-region;

[0048] A large-area meteorological field determination module is used to stitch together the meteorological fields of the inner nested regions of each sub-region after temporal and spatial coupling to obtain the meteorological field of the large area;

[0049] The nested region meteorological field determination module is used to extract the large region according to the geographical range of each nested region of the target region under the air quality mode, and obtain the meteorological field of each nested region of the target region under the air quality mode based on the meteorological field of the extracted large region.

[0050] Furthermore, the present invention also provides a device for improving the effectiveness of medium- and long-term air quality forecasts, characterized in that it comprises:

[0051] One or more processors;

[0052] Memory, used to store one or more programs;

[0053] When the one or more programs are executed by the one or more processors, the one or more processors implement the method described above for improving the effectiveness of medium- and long-term air quality forecasts.

[0054] Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the program is executed by a processor, it implements the steps of the method for improving the effectiveness of medium- and long-term air quality forecasts as described above.

[0055] Compared with the prior art, the method and system for improving the medium- and long-term air quality forecasting effect of the present invention have one or more of the following beneficial technical effects:

[0056] 1. This invention uses weather morphology technology to identify weather patterns for different time periods within a future timeframe in actual forecasting. It then selects the optimal parameterization scheme combination based on these weather patterns, and simultaneously perturbs the initial field to construct ensemble forecast members. This process simulates and predicts future weather. During weather pattern transitions, a new initial field is generated based on the ensemble average. This new initial field is then perturbed again to generate ensemble members, and the optimal parameterization scheme combination is selected again based on the future weather pattern to simulate and predict future weather. This process is repeated, selecting different parameterization scheme combinations based on weather morphology technology and perturbing the initial field at different time periods to generate forecast results for each time period. Finally, the forecast results are coupled across multiple regions and time periods to obtain the meteorological field driving the air quality model.

[0057] 2. This invention improves the accuracy of long-term air quality forecasts by enhancing the performance of long-term forecasts in meteorological models. Attached Figure Description

[0058] Figure 1 The graph shows the changes in the actual atmospheric conditions and the predicted atmospheric conditions over time, as presented by existing technologies.

[0059] Figure 2 This is a flowchart of the method for improving the effectiveness of medium- and long-term air quality forecasting according to the present invention.

[0060] Figure 3 A schematic diagram of the large-area division of the present invention is shown.

[0061] Figure 4 A schematic diagram of nested sub-regions of the present invention is shown.

[0062] Figure 5 A schematic diagram of the buffer area configuration of the present invention is shown.

[0063] Figure 6 A schematic diagram of the spatial coupling of the present invention is shown.

[0064] Figure 7 A schematic diagram of the meridional winds of each grid in the outer nested region of the present invention is shown.

[0065] Figure 8 A schematic diagram of the zonal winds for each grid in the outer nested region of the present invention is shown.

[0066] Figure 9 A schematic diagram is shown illustrating the prediction results of other meteorological elements for each grid in the outer nested region determined by the present invention.

[0067] Figure 10 This is a schematic diagram of the system structure for improving the medium- and long-term air quality forecasting effect of the present invention.

[0068] in, Figure 7-9 In the image, the left side shows a grid diagram of a large area, and the right side shows a grid diagram of the outer nested area of ​​the air quality model. One grid in the outer nested area includes nine grids within the large area. Detailed Implementation

[0069] Before detailing any embodiment of the invention, it should be understood that the invention, in its application, is not limited to the details of the construction and arrangement of the components set forth in the following description or illustrated in the following figures. The invention can have other embodiments and can be practiced or carried out in various ways. Furthermore, it should be understood that the wording and terminology used herein are for descriptive purposes and should not be considered limiting. The use of “comprising” or “having” and variations thereof is intended to cover the items set forth below and their equivalents, as well as any additional items. Unless otherwise specified or limited, the terms “installation,” “connection,” “support,” and “linkage,” and variations thereof are used broadly and cover both direct and indirect installation, connection, support, and linking. Moreover, “connection” and “linkage” are not limited to physical or mechanical connections or links.

[0070] Furthermore, firstly, in the disclosure of this invention, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the above terms should not be construed as limiting this invention. Secondly, the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple. The term "a" should not be construed as a limitation on the quantity.

[0071] To address the shortcomings of existing technologies, this invention provides a method and system for improving the effectiveness of medium- and long-term air quality forecasts. Based on fully considering the characteristics of the meteorological field required by air quality models, it uses weather classification technology, initial field multiple perturbation technology, and multi-time period and multi-region simulation result coupling technology to solve the discontinuity of meteorological field forecasts in time and space by meteorological models. By improving the effectiveness of medium- and long-term forecasts by meteorological models, the accuracy of air quality model forecasts is improved.

[0072] Figure 2 A flowchart illustrating the method for improving medium- and long-term air quality forecasting according to the present invention is shown. Figure 2 As shown, the method for improving the effectiveness of medium- and long-term air quality forecasting according to the present invention includes the following steps:

[0073] 1. Determine the weather type.

[0074] 1. Divide the large region into multiple sub-regions, and identify the weather patterns of each sub-region at each time period based on weather classification technology and historical meteorological data over many years. Organize the time periods of each sub-region according to the weather patterns.

[0075] In this invention, the specific implementation is illustrated by dividing a large area (e.g., the entire China) into four sub-regions, with each sub-region implementing a double-nesting structure (the outer nested region has a 9km grid, i.e., a square grid with a side length of 9km, also known as a 9km resolution grid; the inner nested region has a 3km grid, i.e., a square grid with a side length of 3km, also known as a 3km resolution grid; thus, one grid of the outer nested region includes nine grids of the inner nested region). In practical applications, multiple sub-regions (e.g., 8 or 16) and multiple nesting structures (e.g., triple nesting, with the outer nested region having a 27km grid, the middle nested region having a 9km grid, and the inner nested region having a 3km grid) can be set according to actual needs, with the specific processing method being the same as for four sub-regions and double nesting.

[0076] For ease of expression, such as Figure 3 As shown, this invention divides the large region into four sub-regions (northeast, northwest, southeast, and southwest, denoted as NE, NW, SE, and SW, respectively). Meanwhile, as... Figure 4 As shown, each sub-region includes an inner nested region and an outer nested region. Furthermore, during setup, it is ensured that the innermost nested regions of each sub-region are tightly connected and do not overlap (e.g., ...). Figure 3 The outer nested regions can overlap. Because the outer nested regions in this invention only provide initial values ​​and boundary values ​​to the inner nested regions, and do not subsequently process the data in the outer nested regions, this is for ease of demonstration. Figure 3 The outer nested area has been hidden, and only the inner nested area is displayed.

[0077] In this invention, weather patterns for each time period in four sub-regions are identified based on weather classification technology combined with meteorological data from the past three or more years. The historical time periods for each sub-region are then organized according to weather patterns. For example, the weather pattern for the Northeast sub-region from [date] to [date] is specified.

[0078] Weather classification technology primarily uses the analysis of temperature, humidity, isopotential height lines (aerial weather maps) or isobars (surface weather maps), wind direction, and wind speed as a basis to identify atmospheric weather systems (troughs, ridges, shear lines, cold vortices, low-pressure systems, etc.) and surface weather systems (pre-frontal, post-frontal, rear of high-pressure systems, uniform fields, topographic troughs, inverted troughs, North China low pressure systems, etc.) and surface weather systems to determine the local weather pattern. Specific weather patterns generally exhibit certain patterns in their weather types, and analyzing local weather patterns to forecast local weather is a common technique in artificial weather forecasting. Weather classification technology involves manually analyzing a large number of historical weather systems and then training a large weather classification model based on AI technology; this is a current technology.

[0079] 2. Set up buffer zones between two adjacent sub-regions, and identify the weather patterns of each buffer zone for each time period based on weather classification technology and historical meteorological data over many years, and organize the time periods of each buffer zone according to the weather patterns.

[0080] In the examples of this invention, determining the weather pattern for each buffer zone at each time period specifically includes:

[0081] 1) such as Figure 5 As shown, a buffer region 1 (also set as a double nested region) is set between the northeast and northwest sub-regions, denoted as HC1. The inner nested region of buffer region 1 needs to completely overlap with the mesh of the inner nested regions of the northeast and northwest sub-regions (e.g., ...). Figure 5Because the outer nested region of this invention only provides initial values ​​and boundary values ​​to the inner nested region, and does not subsequently process the data of the outer nested region, this is for ease of demonstration. Figure 5 The outer nested region was hidden, and only the inner nested region was displayed. Based on weather classification technology and combined with meteorological data from the past 3 years or more, the weather of buffer zone 1 was classified to obtain the weather type of buffer zone 1 for each time period, and the time periods of buffer zone 1 were organized according to the weather type.

[0082] 2) Similarly, buffer zone 2 is set between the southeast and southwest sub-regions, buffer zone 3 is set between the northeast and southeast sub-regions, and buffer zone 4 is set between the southwest and northwest sub-regions (buffer zones 2, 3, and 4 are also nested). It can be seen that buffer zone 1 and buffer zone 2 are buffer zones between sub-regions divided in the east-west direction, and buffer zone 3 and buffer zone 4 are buffer zones between sub-regions divided in the north-south direction. Similar to step 1), based on weather classification technology combined with historical meteorological data from the past 3 years or more, weather classification is performed on buffer zones 2, 3, and 4 to obtain the weather patterns for each time period in buffer zones 2, 3, and 4. The time periods of buffer zones 2, 3, and 4 are then organized according to the weather patterns.

[0083] In this invention, the purpose of setting a buffer zone is to establish a transition zone between two adjacent sub-regions, allowing meteorological elements to evolve slowly and avoiding abnormal discontinuities in the meteorological field at the boundary between the two adjacent sub-regions. For example, if the northeastern and northwestern sub-regions are directly joined, and the two sub-regions use different weather patterns, there will be abnormal discontinuities in the meteorological field at the boundary. These abnormal discontinuities may form a false convergence zone at the boundary, and using this meteorological field to drive the air quality model will increase the simulation error.

[0084] II. Determine the optimal combination of parameterization schemes.

[0085] 1. Use meteorological models to simulate each time period of each sub-region and identify the optimal combination of parameterization schemes for each sub-region under different weather types.

[0086] In this invention, since the weather patterns of each sub-region at each time period are obtained through step one, and the types of weather that will occur under specific weather conditions generally follow certain patterns, the optimal parameterization scheme combination under a specific weather pattern is relatively fixed. Therefore, by using meteorological models to simulate each time period of each sub-region, the optimal parameterization scheme combination of each sub-region under different weather patterns can be identified.

[0087] When using the WRF meteorological model, the optimal parameterization scheme combination includes more than 10 types of parameterization schemes, such as cumulus convection parameterization scheme, boundary layer parameterization scheme, microphysical parameterization scheme, and land surface process parameterization scheme. Each parameterization scheme includes the optimal parameterization scheme adopted.

[0088] 2. Use meteorological models to simulate each buffer zone at different times, and identify the optimal combination of parameterization schemes for each buffer zone under different weather types.

[0089] Similarly, in this invention, since the weather patterns corresponding to each buffer zone in each time period are obtained through step one, and the types of weather that will occur under specific weather conditions generally follow certain patterns, the optimal parameterization scheme combination under a specific weather pattern is relatively fixed. Therefore, by using meteorological models to simulate each buffer zone in each time period, the optimal parameterization scheme combination for each buffer zone under different weather patterns can be identified.

[0090] 3. Determine the future weather pattern.

[0091] When forecasting future air quality, the weather patterns for each sub-region and buffer zone in the future are first identified based on meteorological data from the driving field of the meteorological model, combined with weather classification technology.

[0092] Currently, there are various existing meteorological models driving field data. All current meteorological models driving field data meet the requirements of weather classification. By combining the meteorological data of the driving field of the meteorological models with weather classification technology, the future weather patterns of each sub-region and buffer region within a certain period of time (90 days) can be identified.

[0093] Since this invention forecasts air quality in the medium to long term, and the medium to long term forecasts are for the next three months, and because weather patterns are constantly evolving, there will be multiple weather types in the next three months, when making predictions for the future, the weather patterns for the next three months need to be segmented. For example, three months is 90 days, and days 1-10 are one period, corresponding to one weather type; days 11-20 are another period, corresponding to one weather type; and so on.

[0094] IV. Determine the ensemble forecast field for the previous time period.

[0095] The weather forecast for the previous period is obtained by predicting the weather of a certain area in each sub-region and each buffer region.

[0096] Specifically, the initial field of a certain region (e.g., the Northeast sub-region) for the first time period is perturbed using methods such as the singular vector method or the growing model propagation method to generate multiple first initial fields, constructing a series of initial value ensemble forecast members. Then, each initial value ensemble forecast member is combined with the future weather pattern of the region for the first time period obtained in step three and the optimal parameterization scheme under that weather pattern determined in step two to predict future weather, obtaining the first meteorological forecast field of each first initial value forecast member for the region for the first time period. Finally, the ensemble average of the first meteorological forecast fields of each first initial value forecast member for the region for the first time period is performed to obtain the ensemble forecast field for the region for the first time period.

[0097] In this invention, the prediction for the first time period extends beyond the first time period by a certain time, for example, 6 hours. For instance, if the first time period for a certain region (Northeast sub-region) is from 00:00 on March 1st to 00:00 on March 5th, then the prediction for the first time period should begin at 06:00 on March 5th. Simultaneously, the prediction for the second time period should begin a certain time earlier, for example, 6 hours earlier. For instance, if the second time period for a certain region (Northeast sub-region) is from 00:00 on March 5th to 00:00 on March 10th, then the prediction for the second time period needs to start from 18:00 on March 4th. This facilitates subsequent time fitting and eliminates the temporal discontinuities in the meteorological field caused by different combinations of optimal parameterization schemes.

[0098] 5. Determine the ensemble forecast field for the next time period.

[0099] Based on the ensemble forecast field of the previous time period, the weather of the next time period connected to the previous time period is predicted to obtain the ensemble forecast field of the next time period.

[0100] Specifically, the ensemble forecast field of the previous time period in the region is first subjected to ensemble averaging, and the result of the ensemble averaging is used as the initial field of the next time period in the region. For example, the ensemble forecast field predicted for 18:00 on March 4 (the start time of the second time period) in the previous time period is subjected to ensemble averaging, and the result of the ensemble averaging is used as the initial field of the forecast for the next time period.

[0101] Because the integration error increases significantly after a period of time in the previous forecast, the ensemble forecast field at the initial moment of the next time period in the ensemble forecast field of the previous time period is the average forecast field after ensemble averaging. Using it as the initial field of the next time period can reduce the simulation error.

[0102] Then, the initial field of the next time period is perturbed using methods such as singular vector method or growth mode propagation method to generate multiple initial fields of the next time period, and a second set of initial value prediction members is constructed.

[0103] Finally, based on the initial ensemble forecast members constructed in the second phase and the optimal parameterized scheme combination under the corresponding weather type in the region for the next period, the future weather is ensemble forecasted to obtain the ensemble forecast field for the region for the next period.

[0104] This invention focuses on medium- to long-term forecasting. Current ensemble forecasting techniques for this purpose typically construct ensemble forecast members by perturbing an initial field at a specific moment, simulating and predicting weather for the next three months. This avoids generating multiple initial fields and performing multiple perturbations within the three-month forecast period. In contrast, this invention divides the forecast period into multiple time periods, generating an initial field for each period, perturbing it, and then performing an ensemble forecast. For example, in a three-month forecast, the initial field is perturbed based on the original start time, resulting in a series of initial ensemble members, which are then used for ensemble forecasting. After a period of forecasting, as the integration time increases, the forecast errors of each ensemble member deviate significantly from the actual weather field. At this point, the members are ensemble-averaged, resulting in a forecast field with minimal error from the actual weather field at all times. The weather field at a specific moment (the initial moment of the next time period) of this ensemble-averaged weather field is then processed to generate a new initial field with minimal error from the actual weather field. Based on this new initial field, a second perturbation is performed to generate another series of ensemble members, and another ensemble forecast is performed. After a period of forecasting, ensemble averaging is performed again to generate an initial field, which is then perturbed to generate ensemble members for combined forecasting. This process is repeated until the forecasting period ends. This method requires multiple initial value perturbations when predicting the next 90 days, while traditional ensemble forecasting only perturbs at the first moment and forecasts the next 90 days at once. Therefore, this invention can achieve better forecasting results.

[0105] VI. Determine the ensemble forecast field for all time periods in the region.

[0106] Repeat step five to obtain the ensemble forecast fields for all time periods in the region (Northeast sub-region). Specifically, in the first repetition, the ensemble forecast field at the initial moment of the second time period from the ensemble forecast field of the first time period is ensemble-averaged, and the result is used as the initial field of the second time period. The initial field of the second time period is perturbed, and an ensemble forecast is performed for the second time period to obtain the ensemble forecast field for the second time period. In the second repetition, the ensemble forecast field at the initial moment of the third time period from the ensemble forecast field of the second time period is ensemble-averaged, and the result is used as the initial field of the third time period. The initial field of the third time period is perturbed, and an ensemble forecast is performed for the third time period to obtain the ensemble forecast field for the third time period. This process is repeated to obtain the ensemble forecast fields for all time periods in the region.

[0107] 7. Determine the ensemble forecast field for all time periods of all sub-regions and buffer zones.

[0108] Repeat steps four through six for each sub-region and buffer region except for the region (Northeast sub-region) where the ensemble forecast field for all time periods has been determined through steps four through six, to obtain the ensemble forecast fields for all time periods of all sub-regions and buffer regions.

[0109] Because each sub-region and each buffer region is nested multiple times (double nesting in the example of this invention), the ensemble forecast field for all time periods of all sub-regions and buffer regions determined above includes both the ensemble forecast field for all time periods of the inner nested regions of each sub-region and each buffer region, and the ensemble forecast field for all time periods of the outer nested regions of each sub-region and each buffer region. However, in subsequent processing, this invention only uses the ensemble forecast field for all time periods of the inner nested regions of each sub-region and each buffer region.

[0110] 8. Temporal coupling and spatial coupling.

[0111] 1. Perform temporal coupling on the ensemble forecast fields for all time periods of each sub-region and the inner nested regions of each buffer region.

[0112] For each sub-region and buffer region, as shown in steps four and five, the ensemble forecast fields of the two time periods have a 12-hour overlap. This overlap period is mainly used for time fitting to avoid abnormal discontinuities in the meteorological field over time. For example, time coupling of relative humidity mainly involves fitting the ensemble forecast fields of two weather patterns. In the last 12 hours of the first weather pattern, the weight of the relative humidity prediction results for each hour of that weather pattern gradually decreases, while the weight of the second weather pattern gradually increases.

[0113] Meanwhile, the fitting process must consider not only the continuity of the meteorological field but also the non-uniformity of specific characteristic quantities. Therefore, this invention uses a sine and cosine probability density function that can characterize the non-uniform distribution of characteristic quantities for fitting.

[0114] The fitting formula is as follows:

[0115]

[0116] In the formula, Let be the fitted value of the relative humidity at time t within the n-hour overlap period. The predicted relative humidity for the weather pattern preceding time t within an n-hour overlap period. This is the predicted relative humidity for the next weather pattern within an n-hour overlap period, where t represents any time within the overlap period and its value ranges from 1, 2, 3, ..., n. is a constant, where, The radian is equal to 180°. , .

[0117] The formula shows that at times adjacent to the previous weather pattern, the previous weather pattern has a larger weight, and as time progresses, the weight of the previous weather pattern gradually decreases while the weight of the subsequent weather pattern gradually increases. Similarly, by temporally coupling other meteorological elements in the ensemble forecast field (e.g., temperature, air pressure, radial wind, zonal wind, etc.), the temporally coupled ensemble forecast field of each sub-region and the inner nested regions of each buffer region can be obtained.

[0118] 2. Spatial coupling is performed on the ensemble forecast field after temporal coupling of each sub-region and the inner nested region of the buffer region.

[0119] 1) By combining the time-coupled ensemble forecast fields of the inner nested regions of the buffer zones between the east-west divided sub-regions with the time-coupled ensemble forecast fields of the inner nested regions of each sub-region, east-west fitting is performed to obtain the east-west fitted meteorological fields of the inner nested regions of each sub-region. The specific implementation steps are as follows:

[0120] ① The buffer zone between the east-west divided sub-regions is divided into a central region and two edge regions. The central region and the two edge regions are also double-nested. Specifically, the ensemble forecast field resulting from the temporal coupling of the inner nested region of the buffer zone is used as the east-west fitted meteorological field of the inner nested region of the central region.

[0121] For example, such as Figure 6 As shown, buffer region 1 is divided into a central region C and two edge regions D and E. The central region C and the edge regions D and E are also double-nested. The width of the central region C and the two edge regions D and E can each occupy one-third of the width of buffer region 1. The ensemble forecast field after time coupling of the inner nested regions of buffer region 1 is used as the east-west fitted meteorological field of the inner nested region C in the middle of buffer region 1.

[0122] ②The fitting results of the time-coupled ensemble forecast field of the inner nested region of the buffer zone and the time-coupled ensemble forecast field of the inner nested region of the sub-region that overlaps with the edge zone are used as the east-west fitted meteorological field of the inner nested region of the edge zone.

[0123] For example, such as Figure 6 As shown, the areas outside region C in the middle of buffer zone 1, namely region D which overlaps with the northeast sub-region and region E which overlaps with the northwest sub-region, need to be fitted to solve the problem of abnormal discontinuity in the east-west direction of the meteorological field in adjacent regions and improve the accuracy of the meteorological field.

[0124] For each grid within the inner nested region of region D, there are two time-coupled ensemble forecast fields, namely, the forecast results of meteorological elements, specifically the forecast results of meteorological elements in the inner nested region of the Northeast sub-region and the forecast results of meteorological elements in the inner nested region of buffer region 1. When fitting region D, to ensure the continuity of the meteorological field, the forecast results of meteorological elements (temperature, air pressure, humidity, radial wind, zonal wind, etc., more than 100 meteorological elements) in the inner nested region of buffer region 1 should have a higher weight than the grids in the inner nested region of region C. Similarly, the forecast results of meteorological elements in the inner nested region of the Northeast sub-region should have a higher weight than the grids in the inner nested region of region B. From region C to region B, the weight of the forecast results of meteorological elements in the inner nested region of buffer region 1 gradually decreases, while the weight of the forecast results in the inner nested region of the Northeast sub-region gradually increases. Furthermore, the fitting process must consider not only the continuity of the meteorological field but also the uniformity of specific characteristic quantities. Therefore, this invention uses a sine and cosine probability density function, which can characterize the non-uniform distribution of characteristic quantities, for fitting.

[0125] The fitting formula is as follows:

[0126]

[0127] In the formula, This is the prediction result of a meteorological element for the i-th grid in a certain row of grids in the east-west direction of the inner nested region of edge region D, starting from the central region. Here, n is the total number of grids in that row in the east-west direction of the inner nested region of edge region D, and i is 1, 2, 3...n from region C to region B. This refers to the prediction results of meteorological elements for the corresponding grid in the inner nested region of buffer zone 1. This represents the predicted meteorological elements for the corresponding grid within the inner nested region of the Northeast sub-region. is a constant, where, The radian is equal to 180°. , .

[0128] By using the above fitting formula to fit all the grids in a row of grids in the east-west direction of the inner nested region of the edge area, the prediction results of meteorological elements for all grids in that row can be obtained.

[0129] As can be seen from the formula, the first grid in the inner nested region to the left of region D, near region C, uses the meteorological element prediction results from the inner nested region of buffer region 1. Conversely, the first grid in the inner nested region to the right of region D, near region B, uses the meteorological element prediction results from the inner nested region of the northeastern sub-region. From region C to region B, the weight of the meteorological element prediction results from the inner nested region of buffer region 1 gradually decreases, while the weight of the meteorological element prediction results from the inner nested region of the northeastern sub-region gradually increases. Furthermore, the change in weight is non-uniform.

[0130] By using the above fitting formula to fit all the grids in each row of the inner nested region of the edge area in the east-west direction, the meteorological element prediction results of all grids in the inner nested region of the edge area can be obtained.

[0131] Using the same method, the forecast results of meteorological elements in the inner nested region of region E can be obtained.

[0132] ③ The ensemble forecast field after time coupling of the inner nested regions of each sub-region is used as the east-west fitted meteorological field of the inner nested region of the part of each sub-region that does not overlap with the buffer region.

[0133] For example, outside of buffer zone 1, the meteorological element forecasts for region B in the northeast sub-region and region A in the northwest sub-region are used from the inner nested regions of their respective sub-regions.

[0134] ④ Integrate the east-west fitted meteorological fields of the inner nested regions of the central region, the east-west fitted meteorological fields of the inner nested regions of the edge region, and the east-west fitted meteorological fields of the inner nested regions of each sub-region that do not overlap with the buffer region to obtain the east-west fitted meteorological fields of the inner nested regions of each sub-region.

[0135] In other words, the predicted meteorological elements of the inner nested regions of region C, region D, region E, region A, and region B are integrated to obtain the predicted meteorological elements of the inner nested regions of the new northeast and northwest sub-regions, i.e., the east-west fitted meteorological fields of the inner nested regions of the northeast and northwest sub-regions.

[0136] Similarly, by fitting the inner nested region of buffer zone 2 between the southwest and southeast sub-regions using the above method, we can obtain the prediction results of meteorological elements in the inner nested regions of the new southwest and southeast sub-regions, that is, the east-west fitted meteorological field of the inner nested regions of the southwest and southeast sub-regions.

[0137] However, the east-west fitted meteorological field is only the inner nested region of the buffer zone between the northeast and northwest sub-regions and the buffer zone between the southeast and southwest sub-regions. The prediction results of meteorological elements in the north-south direction of the inner nested region between the northeast and southeast sub-regions and the northwest and southwest sub-regions may still differ greatly, which will cause discontinuity in the meteorological field. Therefore, north-south fitting is required.

[0138] 2) By combining the temporal coupling of the inner nested regions of the buffer zones between the north-south sub-regions and the east-west fitted meteorological fields of the inner nested regions of each sub-region, a north-south fitting is performed to obtain the temporal and spatial coupled meteorological fields of the inner nested regions of each sub-region, that is, the prediction results of meteorological elements of each grid in the inner nested regions of each sub-region.

[0139] In this invention, similar to the east-west fitting method, the meteorological field after east-west fitting of the inner nested regions of buffer region 4 and the northwest and southwest sub-regions is fitted in the north-south direction, and the meteorological field after east-west fitting of the inner nested regions of buffer region 3 and the northeast and southeast sub-regions is fitted in the north-south direction to obtain the meteorological field after temporal and spatial coupling of the inner nested regions of each sub-region, that is, the prediction results of meteorological elements of each grid in the inner nested regions of each sub-region.

[0140] 9. Determine the meteorological field of the large area.

[0141] After obtaining the temporal and spatial coupled meteorological fields of the inner nested regions of each sub-region, the meteorological fields of the inner nested regions of each sub-region are spliced ​​together to obtain the meteorological field of the large region.

[0142] In this invention, since the previous processing was based on the inner nested region, it can be seen that the meteorological field of this large area is only the meteorological field of the inner nested region, that is, the meteorological field with a 3-kilometer grid resolution.

[0143] 10. Determine the meteorological field for each nested region.

[0144] When conducting air quality model simulations, the first step is to determine the area to be simulated, i.e., the target area in the air quality model. For example, if an air quality model simulation is conducted for Hebei Province, the target area would be the geographical area of ​​Hebei Province and its surrounding regions within a certain distance.

[0145] Meanwhile, air quality model simulations are generally multi-layered, such as double-layered. In this case, the outer nested region (e.g., Hebei Province) is larger with a larger grid, for example, 9km, but with lower resolution. The inner nested region (e.g., Shijiazhuang City) is smaller than the outer nested region, and its grid is smaller than the outer nested region's grid, for example, 3km, but with higher resolution. Of course, triple-layered or other heavily nested regions are also possible. In a triple-layered simulation, the outer nested region has a grid of 27km, the middle nested region has a grid of 9km, and the inner nested region has a grid of 3km.

[0146] Through the processing steps one through nine described above, this invention can obtain a meteorological field with a 3km resolution grid for a large area (e.g., the entire China). Thus, when performing air quality model simulations, the target area in the air quality model (e.g., Hebei Province, Beijing, etc.) can directly obtain its own 3km resolution grid meteorological field based on the meteorological field of the larger region, according to its geographical scope. This eliminates the need for each target area to obtain its own 3km resolution grid meteorological field through steps one through nine separately, saving significant costs.

[0147] In this invention, when performing air quality model simulation, the large area can be truncated according to the geographical range of each nested region of the target area under the air quality model, and the meteorological fields of each nested region of the target area under the air quality model can be obtained based on the meteorological field of the truncated large area, which specifically includes:

[0148] 1. Based on the geographical range of the inner nested region of the target area under the air quality model, the large area is directly extracted, and the meteorological field of the extracted large area is used as the meteorological field of the inner nested region of the target area under the air quality model.

[0149] For example, based on the size of the inner nested region of the target area in the air quality model, a section is directly cut off on the large area, and the meteorological field of each grid in the cut-off large area (i.e., each 3km resolution grid of the large area) is used as the meteorological field of each grid (which is also a 3km resolution grid) in the inner nested region of the target area in the air quality model.

[0150] 2. Based on other nested regions of the air quality model, for example, the geographical range of the outer nested region is cut off from the large region, and the meteorological field of each grid in the large region contained by each grid of the other nested regions is processed to obtain the meteorological field of each grid of the other nested regions of the air quality model.

[0151] For example, in this invention, the grid of the outer nested region of the target area in the air quality model is a 9km resolution grid, and the grid of the large area is a 3km resolution grid. Therefore, one grid of the outer nested region of the target area in the air quality model includes 9 grids in the large area. By processing the meteorological field of the 9 grids in the large area, the meteorological field of the grid of the outer nested region of the target area in the air quality model can be obtained.

[0152] In this invention, the process of obtaining the meteorological field of each grid in the other nested regions of the target area under the air quality model, based on the meteorological field of each grid within the large region contained in each of the other nested regions, specifically includes:

[0153] 1) The meridional wind in the meteorological field of each grid in the other nested regions is obtained by weighted averaging the meridional wind in the meteorological field of the westernmost column of grids in the large region contained by each grid in the other nested regions.

[0154] For example, such as Figure 7 As shown, the meridional wind u in a grid 1 of the outer nested region is determined by the westernmost column of nine grids within the large area encompassed by grid 1 of the outer nested region, that is, Figure 7 The weighted average of the meridional winds of grid 1, grid 2, and grid 3 within the large region is obtained.

[0155] 2) The zonal wind in the meteorological field of each grid in the other nested regions is obtained by weighted averaging the zonal wind in the meteorological field of the northernmost grid in the large region contained by each grid in the nested gas layer region.

[0156] For example, such as Figure 8 As shown, the zonal wind v in grid 1 of the outer nested region is determined by the westernmost column of the nine grids within the large area encompassed by grid 1 of the outer nested region. Figure 8 The weighted average of the zonal winds of grid 1, grid 2, and grid 3 within the large region is obtained.

[0157] 3) The prediction results of the remaining meteorological elements in the meteorological field of each grid in the other nested regions are obtained by weighted averaging the prediction results of the corresponding meteorological elements in the meteorological field of all grids in the large region contained by each grid in the outer nested region.

[0158] For example, such as Figure 9 As shown, the prediction results of meteorological elements other than meridional wind u and zonal wind v in grid 1 of the outer nested region are derived from the nine grids within the large area contained in grid 1 of the outer nested region, that is, Figure 9 The weighted average of the prediction results of the corresponding meteorological elements in grid 1, grid 2, grid 3, ..., grid 9 within the large area is obtained.

[0159] In general meteorological model simulations, the outer nested region (using a 9km resolution grid) is larger than the inner nested region (using a 3km resolution grid), providing initial and boundary values ​​to the inner nested region. Simultaneously, the inner nested region provides feedback to the outer nested region, ensuring that meteorological elements in the overlapping areas of the two nested regions match. However, this invention, when obtaining the prediction results of meteorological elements from the inner nested region's grid, reduces the impact of discontinuities in the meteorological field between adjacent sub-regions by setting a buffer and performing secondary fitting. This results in a deviation between the predicted meteorological elements of the overlapping area's grid and the original meteorological element prediction results. If a similar method is used to process and stitch together the outer nested region to obtain the prediction results of its meteorological elements, then the predicted meteorological elements of each grid in the overlapping or adjacent areas of the two sub-regions in the outer nested region will also deviate from the original meteorological elements. This leads to a mismatch between the predicted meteorological elements of the 9km resolution grid and the 3km resolution grid in the overlapping areas of the inner and outer nested regions, thus preventing the direct processing of the outer nested region. Therefore, the prediction results of meteorological elements in the inner and outer nested regions of this invention are all based on the prediction results of meteorological elements in the 3km resolution grid (i.e., the 3km resolution grid in the large area of ​​the meteorological model obtained in step nine), and can be arbitrarily extracted according to the needs of air quality model simulation.

[0160] Figure 10 A schematic diagram of the system configuration for improving the medium- and long-term air quality forecasting performance of the present invention is shown. Figure 10 As shown, the system for improving medium- and long-term air quality forecasting according to the present invention includes:

[0161] 1. Weather type determination module.

[0162] The weather pattern determination module is used to divide a large area into multiple sub-regions and set up buffer zones between two adjacent sub-regions. Based on weather patterning technology and combined with reanalysis meteorological data from many years of history, it identifies the weather patterns of each sub-region and each buffer zone for each time period.

[0163] 2. Optimal parameterization scheme combination determination module.

[0164] The optimal parameterization scheme combination determination module is used to simulate weather over many years using meteorological models and reanalysis data, and to identify the optimal parameterization scheme combination for each sub-region and buffer zone under different weather types.

[0165] 3. Future weather prediction module.

[0166] The future weather pattern determination module is used to identify the future weather pattern of each sub-region and each buffer region for a period of time in the future, based on meteorological data from the driving field of the meteorological model and weather classification technology.

[0167] 4. Module for determining the ensemble forecast field of the previous period.

[0168] The previous period ensemble forecast field determination module is used to predict the weather of a certain area in each sub-region and each buffer region in the previous period, and obtain the previous period ensemble forecast field of that area.

[0169] 5. The module for determining the ensemble forecast field for the next time period.

[0170] The ensemble forecast field determination module for the next time period is used to predict the weather of the region in the next time period adjacent to the previous time period, and obtain the ensemble forecast field of the region in the next time period.

[0171] 6. Regional ensemble forecast field determination module.

[0172] The regional ensemble forecast field determination module is used to obtain the ensemble forecast field for all time periods of the region based on the ensemble forecast field determination module for the previous time period and the ensemble forecast field determination module for the next time period.

[0173] 7. Time and space coupling module.

[0174] The time-space coupling module is used to perform time-space coupling on the ensemble forecast field of all time periods of each sub-region and the inner nested region of each buffer region, so as to obtain the time-space coupled meteorological field of the inner nested region of each sub-region.

[0175] 8. Module for determining the meteorological field in a large area.

[0176] The large-area meteorological field determination module is used to splice together the meteorological fields of the inner nested regions of each sub-region after temporal and spatial coupling to obtain the meteorological field of the large area.

[0177] 9. Meteorological field determination module for each nested region.

[0178] The nested region meteorological field determination module is used to extract the large region according to the geographical range of each nested region of the target region under the air quality mode, and obtain the meteorological field of each nested region of the target region under the air quality mode based on the meteorological field of the extracted large region.

[0179] Furthermore, the present invention also provides an apparatus for improving the effectiveness of medium- and long-term air quality forecasts, comprising: one or more processors; a memory for storing one or more programs; and, when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the method for improving the effectiveness of medium- and long-term air quality forecasts as described above.

[0180] Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for improving the effectiveness of medium- and long-term air quality forecasts as described above.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Those skilled in the art can modify or make equivalent substitutions to the technical solutions of the present invention based on the concept of the present invention, without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for improving the effectiveness of medium- and long-term air quality forecasts, characterized in that, Includes the following steps: 1) Divide the large area into multiple sub-regions and set up buffer zones between two adjacent sub-regions. Based on weather classification technology and combined with reanalysis meteorological data from many years of history, identify the weather patterns of each sub-region and each buffer zone at each time period. 2) Use meteorological models and reanalysis data to simulate weather over many years to identify the optimal combination of parameterization schemes for each sub-region and buffer zone under different weather types; 3) Based on meteorological model-driven field meteorological data and combined with weather typology technology, identify the future weather patterns of each sub-region and each buffer zone for a period of time in the future; 4) Forecast the weather of a certain area in each sub-region and each buffer region for the previous period to obtain the ensemble forecast field for the previous period. 5) Based on the ensemble forecast field of the previous time period, the weather of the next time period adjacent to the previous time period is predicted to obtain the ensemble forecast field of the next time period. 6) Repeat step 5) to obtain the ensemble forecast field for all time periods in the region; 7) Repeat steps 4) to 6) for each sub-region and other regions in each buffer region except the sub-region itself to obtain the ensemble forecast field for all time periods of all sub-regions and buffer regions; 8) Perform temporal and spatial coupling on the ensemble forecast fields for all time periods of all sub-regions and the inner nested regions of the buffer region to obtain the temporally and spatially coupled meteorological fields of the inner nested regions of each sub-region. Temporal coupling of the ensemble forecast fields for all time periods of all sub-regions and the inner nested regions of the buffer region involves fitting the ensemble forecast fields during the overlap period of the previous and subsequent time periods using a sine and cosine probability density function. The fitting formula is as follows: In the formula, For the ensemble forecast field of time t within the overlap period, This is the ensemble forecast field for the time period preceding time t within the overlap period. Let n be the ensemble forecast field for the time period following time t within the overlap period, where n is the overlap period and t represents any time within the overlap period, taking values ​​of 1, 2, 3, ..., n. is a constant, where, The radian is equal to 180°. , ; The spatial fitting specifically includes: 81) combining the meteorological fields after temporal coupling of the inner nested regions of the buffer zones between the east-west divided sub-regions and the meteorological fields after temporal coupling of the inner nested regions of each sub-region to perform east-west fitting, thereby obtaining the meteorological fields after east-west fitting of the inner nested regions of each sub-region; 82) combining the meteorological fields after temporal coupling of the inner nested regions of the buffer zones between the north-south divided sub-regions and the meteorological fields after east-west fitting of the inner nested regions of each sub-region to perform north-south fitting, thereby obtaining the meteorological fields after temporal and spatial coupling of the inner nested regions of each sub-region. 9) The meteorological fields of the inner nested regions of each sub-region, after temporal and spatial coupling, are spliced ​​together to obtain the meteorological field of the large region; 10) The large region is truncated according to the geographical range of each nested region of the target region under the air quality model, and the meteorological field of each nested region of the target region under the air quality model is obtained based on the meteorological field of the truncated large region.

2. The method for improving the effectiveness of medium- and long-term air quality forecasting according to claim 1, characterized in that, Step 4) specifically includes: 41) Perturb the initial field of a certain region in each sub-region and each buffer region in the previous time period to generate multiple first initial fields and construct a series of first initial value set forecast members. 42) Based on the future weather pattern of the region in the previous period, the optimal parameterization scheme combination of the region under the weather pattern is used and combined with the forecast members of the first initial value set to predict the future weather, so as to obtain the first meteorological forecast field of each of the first initial value forecast members of the region in the previous period. 43) The ensemble average of the first meteorological forecast fields of each first initial value forecast member in the previous period of the region is used to obtain the ensemble forecast field of the region in the previous period.

3. The method for improving the effectiveness of medium- and long-term air quality forecasting according to claim 2, characterized in that, Step 5) specifically includes: 51) Perform ensemble averaging on the ensemble forecast field of the previous time period in the region and use the result of the ensemble averaging as the initial field of the next time period in the region. 52) Perturb the initial field of the region in the next time period to generate multiple second initial fields and construct a series of second initial value set forecast members; 53) Based on the future weather pattern of the region in the next period, use the optimal parameterization scheme combination of the region under the weather pattern and combine it with the second initial value set forecast members to predict the future weather, and obtain the second meteorological forecast field of each second initial value forecast member of the region in the next period. 54) The ensemble average of the second meteorological forecast fields of each second initial value forecast member for the next period of time in the region is used to obtain the ensemble forecast field for the next period of time in the region.

4. The method for improving the effectiveness of medium- and long-term air quality forecasting according to claim 3, characterized in that, In step 4), the ensemble forecast field of the region in the previous period includes an ensemble forecast field that exceeds the previous period by a certain time. In step 5), the ensemble forecast field of the region in the next period includes an ensemble forecast field that precedes the next period by a certain time, so that the ensemble forecast field of the previous period and the ensemble forecast field of the next period have an overlap period.

5. The method for improving the effectiveness of medium- and long-term air quality forecasting according to claim 1, characterized in that, Step 10) specifically includes: 101) Based on the geographical range of the inner nested region of the target area under the air quality model, the large area is extracted, and the meteorological field of the extracted large area is used as the meteorological field of the inner nested region of the target area under the air quality model. 102) Based on the geographical range of other nested regions of the target area under the air quality model, the large area is extracted, and based on the meteorological field of the extracted large area, the meteorological field of other nested regions of the target area under the air quality model is processed to obtain the meteorological field of other nested regions of the target area under the air quality model.

6. A system for improving the effectiveness of medium- and long-term air quality forecasting, characterized in that, include: The weather pattern determination module is used to divide a large area into multiple sub-regions and set buffer zones between two adjacent sub-regions. Based on weather patterning technology and combined with reanalysis meteorological data from many years of history, it identifies the weather pattern of each sub-region and each buffer zone for each time period. The optimal parameterization scheme combination determination module is used to simulate weather over many years using meteorological models and reanalysis data, and to identify the optimal parameterization scheme combination for each sub-region and buffer zone under different weather types. The future weather pattern determination module is used to identify the future weather pattern of each sub-region and each buffer zone for a period of time in the future, based on meteorological data from the driving field of the meteorological model and weather classification technology. The previous period ensemble forecast field determination module is used to predict the weather of a certain area in each sub-region and each buffer region in the previous period, and obtain the previous period ensemble forecast field of that area. The ensemble forecast field determination module for the next time period is used to combine the ensemble forecast field of the previous time period to predict the weather of the next time period connected to the previous time period, so as to obtain the ensemble forecast field of the next time period for the region. A regional ensemble forecast field determination module is used to obtain the ensemble forecast field for all time periods of the region based on the ensemble forecast field determination module for the previous time period and the ensemble forecast field determination module for the next time period. The temporal and spatial coupling module is used to perform temporal and spatial coupling on the ensemble forecast fields of all time periods of each sub-region and the inner nested regions of each buffer region, so as to obtain the temporal and spatial coupled meteorological fields of the inner nested regions of each sub-region; The large-area meteorological field determination module is used to stitch together the temporal and spatially coupled meteorological fields of the inner nested regions of each sub-region to obtain the meteorological field of the large region. Specifically, temporal coupling of the ensemble forecast fields for all time periods of all sub-regions and the inner nested regions of the buffer region involves fitting the ensemble forecast fields within the overlap period of the previous and subsequent time periods using a sine and cosine probability density function. The fitting formula is as follows: In the formula, For the ensemble forecast field of time t within the overlap period, This is the ensemble forecast field for the time period preceding time t within the overlap period. Let n be the ensemble forecast field for the time period following time t within the overlap period, where n is the overlap period and t represents any time within the overlap period, taking values ​​of 1, 2, 3, ..., n. is a constant, where, The radian is equal to 180°. , ; The spatial fitting specifically includes: 1) combining the meteorological fields after temporal coupling of the inner nested regions of the buffer zones between the east-west divided sub-regions and the meteorological fields after temporal coupling of the inner nested regions of each sub-region to perform east-west fitting, thereby obtaining the meteorological fields after east-west fitting of the inner nested regions of each sub-region; 2) combining the meteorological fields after temporal coupling of the inner nested regions of the buffer zones between the north-south divided sub-regions and the meteorological fields after east-west fitting of the inner nested regions of each sub-region to perform north-south fitting, thereby obtaining the meteorological fields after temporal and spatial coupling of the inner nested regions of each sub-region. The nested region meteorological field determination module is used to extract the large region according to the geographical range of each nested region of the target region under the air quality mode, and obtain the meteorological field of each nested region of the target region under the air quality mode based on the meteorological field of the extracted large region.

7. A device for improving the effectiveness of medium- and long-term air quality forecasts, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for improving the effectiveness of medium- and long-term air quality forecasts as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method for improving the effectiveness of medium- and long-term air quality forecasts as described in any one of claims 1-5.

Images