A method and device for visibility correction integrated forecast

Abstract

The present invention discloses a method and device for visibility correction integrated forecasting, wherein the method adopts a preset probability density matching method and a sliding double weight method respectively to correct the original forecast product to obtain two first visibility correction forecast products for the future n0-n2 forecast time; for the n0-n1 forecast time, the two first visibility correction forecast products are corrected twice based on the first-order Kalman filter combined with the historical visibility actual situation to obtain two second visibility correction forecast products; the two first visibility correction forecast products are used as visibility correction forecast products for the n1-n2 forecast time; the two second visibility correction forecast products and the visibility correction forecast product are integrated. The accuracy of the forecast is improved by the method.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a method for correcting integrated forecasting of visibility. Background Art

[0002] The causes of visibility are complex, especially the occurrence and development of low visibility weather is not only related to weather conditions and atmospheric circulation, but also affected by many factors such as underlying surface, climate environment and human activities, making it difficult to forecast. In addition to reflecting the transparency of the atmosphere, visibility also affects people's work, life and transportation to a certain extent. Especially when low visibility events occur, it often leads to an increase in traffic accidents and delays in aircraft takeoffs and landings, bringing many inconveniences and hazards to daily life and travel.

[0003] Early forecasts of visibility were mainly based on meteorological methods, which were based on meteorological analysis and diagnosis of its formation conditions, combined with the forecaster's experience and the method of extrapolation of actual conditions. With the development of numerical forecasting technology, numerical forecast products at home and abroad have become the main reference for forecasters. Due to the different dynamic frameworks and physical parameterization schemes between numerical models, the model system errors have certain differences, so the accuracy of visibility forecasts in numerical forecasts is low. Research by the U.S. Meteorological Development Laboratory shows that the forecast performance of the model output statistics method (MOS) based on the global forecast system is significantly better than the direct results of the model. Due to the incomplete parameterization scheme of the numerical model itself, there are systematic deviations in the forecasting ability of certain elements. At present, researchers mostly use MOS methods such as frequency matching, probability density matching, Kalman filtering and neural networks to correct the errors of model forecasts. However, visibility has a complex relationship with various meteorological elements, especially in the nowcast, especially the forecast within 24 hours, the impact of meteorological elements on visibility is more obvious, and random errors greatly reduce the effect of model forecasts.

[0004] At present, whether it is the traditional statistical method, the interpretation of numerical forecast products, or artificial intelligence technology based on deep learning, the forecast effect and degree of businessization of visibility forecast are not ideal. Summary of the invention

[0005] The present application provides a method for visibility correction integrated forecast to solve the problems existing in the related art.

[0006] In a first aspect, the present invention provides a method for processing visibility forecast errors, comprising respectively using a preset probability density matching method and a sliding double weight method to correct the original forecast product to obtain the future n 0 -n 2 Forecast timeliness: two first visibility correction forecast products; for n 0 -n 1Based on the first-order Kalman filter and the historical visibility conditions, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products; the two first visibility correction forecast products are used as the forecast products for n 1 -n 2 A visibility correction forecast product with a forecast period; integrating the two second visibility correction forecast products and the visibility correction forecast product.

[0007] Optionally, the preset probability density matching method and sliding double weight method are used to revise the original forecast product to obtain the future n 0 -n 2 The two first visibility revised forecast products of forecast time include: with respect to the probability density function of the actual visibility, adjusting the corresponding probability density function of the visibility forecasted by the original forecast product at each forecasting site to make it consistent with the probability density distribution of the actual visibility;

[0008] Based on the observation increment double-weighted average, the visibility forecasted by the original product at each prediction station is corrected using the sliding double-weighted average method.

[0009] Optionally, with respect to the probability density function of actual visibility, adjusting the corresponding probability density function of the visibility predicted by the original forecast product at each prediction site includes: dividing the forecast visibility and the actual visibility of the original forecast product into m threshold intervals; calculating a first frequency when the actual visibility appears at different visibility thresholds to determine the probability density function of the actual visibility and the frequency of the actual visibility; calculating a second frequency when the forecast visibility appears at different visibility thresholds; and determining the revised forecast value of visibility based on the second frequency and the probability density function of the actual visibility.

[0010] Optionally, based on the observation increment double weighted average, the original product is corrected at each forecast site using a sliding double weighted average method, including: based on The forecast visibility is corrected by: is the double-weighted average of the observation increments at the kth station.

[0011] Optionally, in the calculation When including: in, For the kth station The median of is the corresponding median absolute deviation, that is The median of is the weight, when When Each station Double weighted average for:

[0012] Optionally, for n 0 -n 1 Forecast timeliness, based on the first-order Kalman filter combined with the historical visibility actual situation, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products including: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value; when t=1, B t-1 =0.

[0013] Optionally, the two second visibility correction forecast products and the visibility correction forecast product are integrated, including: integrating the two second visibility correction forecast products respectively to obtain n 0 -n 1 A forecast product with a forecast timeliness, wherein during integration, a test is performed based on the TS score to obtain a first TS score result; based on the size of the first TS score result, weight coefficients corresponding to each second visibility correction forecast product are determined; integration is performed based on the weight coefficients; the visibility correction forecast products are respectively integrated to obtain n 1 -n 2 A forecast product with a forecast timeliness; wherein, during integration, a test is performed based on the TS score to obtain a second TS score result; based on the magnitude of the second TS score result, a weight coefficient corresponding to each second visibility correction forecast product is determined; and integration is performed based on the weight coefficient

[0014] In a second aspect, the present invention provides a method and apparatus for processing visibility correction integrated forecast, comprising a first correction unit, configured to respectively use a preset probability density matching method and a sliding double weight method to correct the original forecast product to obtain the future n 0 -n 2 Forecast timeliness: two first visibility revised forecast products;

[0015] The second correction unit is configured to correct n 0 -n 1Based on the first-order Kalman filter and the historical visibility conditions, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products; the two first visibility correction forecast products are used as the forecast products for n 1 -n 2 Visibility correction forecast products with forecast validity;

[0016] An integration unit is configured to integrate the two second visibility correction forecast products and the visibility correction forecast product.

[0017] In a third aspect, the present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of the first aspects is implemented.

[0018] In a fourth aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method provided in the first aspect when executing the program.

[0019] The present invention discloses a method for correcting integrated visibility forecasting, wherein the method adopts a preset probability density matching method and a sliding double weight method to correct the original forecast product to obtain the future n 0 -n 2 Forecast timeliness: two first visibility correction forecast products; for n 0 -n 1 Based on the first-order Kalman filter and the historical visibility conditions, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products; the two first visibility correction forecast products are used as the forecast products for n 1 -n 2 Visibility correction forecast products for different forecast periods; the two second visibility correction forecast products and the visibility correction forecast products are integrated. Visibility element forecast products such as numerical models and objective methods suitable for current weather business are used, and the visibility forecast effects of different forecast periods and actual weather forecast business needs are fully considered. Different modeling and forecasting technical route methods are adopted to optimize and reduce the adverse effects of systematic errors and random errors on visibility forecast effects, avoiding the limitations of a single method and the inaccuracy of the forecast. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0021] Figure 1A method for processing visibility correction integrated forecast is provided for this application;

[0022] Figure 2 A schematic diagram of the implementation logic of the method for processing visibility correction integrated forecast in this application;

[0023] Figure 3 A method corresponding to the present application Figure 1 Schematic diagram of an electronic device. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0025] The following is combined with Figure 1 The method for processing visibility forecast errors is exemplified. The method comprises the following steps:

[0026] Step 101: Use the preset probability density matching method and sliding double weight method to revise the original forecast product to obtain the future n 0 -n 2 Forecast timeliness: two first visibility correction forecast products.

[0027] In this embodiment, the forecast performance of the recent model visibility forecast is fully considered, and the probability density function matching method (PDF) and the sliding double weight method (MBC) are respectively adopted. According to different forecast time limits and starting times, the rolling modeling (training cycle: past 20 days) is adopted to obtain the future visibility correction forecast products based on the above two methods, such as the future 3-168 hours visibility correction forecast products starting from 08:00 and 20:00 every day.

[0028] Step 102: For n 0 -n 1 Forecast timeliness, based on the first-order Kalman filter combined with the historical visibility situation, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products

[0029] In this embodiment, reference Figure 2Taking the next 3-24 hours as an example, the first-order Kalman filter (1D-Kalman) method is used and combined with the actual data of the visibility hourly prediction site in the past 24 hours, the two sets of visibility correction forecast products generated in step 101 are corrected again to further eliminate the influence of random errors caused by the numerical model, and form two sets of visibility correction forecast products for the next 24 hours, namely "PDF+1D-Kalman" and "MBC+1D-Kalman" forecast products.

[0030] Step 103: Use the two first visibility correction forecast products as the 1 -n 2 Visibility correction forecast product for forecast period.

[0031] In this embodiment, since the actual visibility results of the past 24 hours are used in step 102, only the random error of the next 24 hours can be reduced. For the visibility forecast of the next 27-168 hours, the result of step (1) is maintained, that is, the "PDF" and "MBC" forecast products, respectively, and are integrated after inspection and evaluation.

[0032] Step 104: Integrate the two second visibility correction forecast products and the visibility correction forecast product.

[0033] In this embodiment, forecast products of different forecast time periods may be integrated respectively, and the integration results of the different forecast time periods may be used as the final integration result.

[0034] As an optional implementation method of this embodiment, with respect to the probability density function of the actual visibility, the corresponding probability density function of the visibility predicted by the original forecast product at each forecast site is adjusted to make it consistent with the probability density distribution of the actual visibility; and based on the observed incremental double-weighted average value, the visibility predicted by the original product at each forecast site is corrected by using a sliding double-weighted average method.

[0035] In this optional implementation, the model results are adjusted to be consistent with the probability density distribution of the actual observations based on the probability density matching method, so as to achieve the purpose of correcting the systematic errors of the data. The sliding double weighted average method corrects the forecast for each station separately based on the double weighted average of the observation increments. The double weighted average is generally similar to the average, but can reduce the impact of outliers on the average.

[0036] As an optional implementation method of this embodiment, with respect to the probability density function of actual visibility, adjusting the corresponding probability density function of the visibility predicted by the original forecast product at each prediction site includes: dividing the forecast visibility and the actual visibility of the original forecast product into m threshold intervals; calculating the first frequency when the actual visibility appears at different visibility thresholds to determine the actual visibility and the probability density function of the frequency of the actual visibility; calculating the second frequency when the forecast visibility appears at different visibility thresholds; and determining the revised forecast value of visibility based on the second frequency and the probability density function of the actual visibility.

[0037] In this optional implementation, the specific implementation example is as follows: (1) the visibility actual and forecast values ​​are divided into 9 threshold intervals, namely, 0.05 km, 0.2 km, 0.5 km, 1.0 km, 3.0 km, 5.0 km, 10.0 km, 15.0 km and 25.0 km; (2) the frequency of different visibility thresholds in the actual observed visibility is calculated, and the probability density function f(x) of the actual visibility and the actual visibility frequency is established; (3) the formula is used to calculate the probability density function f(x) of the actual visibility and the actual visibility frequency. By calculating the frequency of different visibility thresholds in the visibility forecast by the model and substituting it into f(x), the revised visibility forecast value can be calculated.

[0038] In addition, since the terrain in the forecast area, such as Xinjiang, is complex and changeable and has an important impact on the spatial distribution of visibility, the probability density function is established for each station instead of a unified function for the whole Xinjiang. In addition, this study fully considers the forecast performance of the recent model, and adopts a rolling modeling method for different forecast timeliness and starting time to establish probability density functions respectively. The training period is the actual situation and model forecast results of the past 20 days. Among them, A i is the number of stations with a certain level of visibility threshold, B is the total number of stations, and P i is the frequency of occurrence of a certain visibility threshold.

[0039] As an optional implementation of this embodiment, based on the observation increment double weighted average, the visibility forecasted by the original product at each forecast site is corrected by using the sliding double weighted average method, including: The forecast visibility is corrected by: is the double-weighted average of the observation increments at the kth station.

[0040] As an optional implementation of this embodiment, in the calculation When including: in, For the kth station The median of is the corresponding median absolute deviation, that is The median of is the weight, when When Each station Double weighted average for:

[0041] In this optional implementation, the double-weighted average is generally similar to the average, but can reduce the impact of outliers on the average: is the double-weighted average of the observation increments at the kth station, and is calculated as follows:

[0042] In the formula, For the kth station The median of is the corresponding median absolute deviation, that is of the median. is the weight, when When Each station Double weighted average for:

[0043] In order to unify the sliding period with the probability density matching method, the sliding correction period n of the sliding double weighted average method is the past 20 days. When correcting, the correction amount is first calculated based on the model forecast and observation values ​​of the previous n days, and then the current forecast field is corrected.

[0044] As an optional implementation of this embodiment, for n 0 -n 1 Forecast timeliness, based on the first-order Kalman filter combined with the historical visibility actual situation, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products including: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value; when t=1, B t-1 =0.

[0045] In this optional implementation, the first-order Kalman filter method is also called the average decreasing method, which is to estimate the decreasing average deviation of the correction time by weighted average and iterative update of the deviations of different forecast time periods. This method has the characteristics of self-adaptation, and can correct the forecast results in real time by adding data, and is not affected by the forecast mode. Its method and principle are as follows: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value. Using "cold start", that is, when t = 1, B t-1 =0. The size of the weight coefficient w reflects the error memory capacity of the correction model and directly affects the correction forecast result. This method adopts a step-by-step increase scheme. The weight coefficient w is increased in a loop from 0 to 1 with a step length of 0.01 to calculate the optimal weight coefficient corresponding to the lowest visibility forecast error for each station in the past 20 days.

[0046] As an optional implementation of this embodiment, integrating the two second visibility correction forecast products and the visibility correction forecast product includes: integrating the two second visibility correction forecast products respectively to obtain n 0 -n 1 A forecast product with a forecast timeliness, wherein during integration, a test is performed based on the TS score to obtain a first TS score result; based on the size of the first TS score result, weight coefficients corresponding to each second visibility correction forecast product are determined; integration is performed based on the weight coefficients; the visibility correction forecast products are respectively integrated to obtain n 1 -n 2 Forecast products with forecast effectiveness; wherein, during integration, a test is performed based on the TS score to obtain a second TS score result; based on the size of the second TS score result, the weight coefficients corresponding to each second visibility correction forecast product are determined respectively; and integration is performed based on the weight coefficients.

[0047] In this optional implementation, in order to further evaluate the recent forecast performance of different correction methods, the optimal weight coefficients of different correction methods are assigned. This method uses the TS (Threat Score) score as the inspection index of the visibility forecast effect. The TS score is between [0,1], which reflects the accuracy of the forecast. The specific inspection index calculation formula is as follows:

[0048] Further, refer to Figure 2, when integrating, the corrected forecast products of different forecast time periods are integrated respectively, and the integrated results of different forecast time periods are used as the final integration. For example, based on the two sets of forecast products "PDF+1D-Kalman" and "MBC+1D-Kalman" in the past 15 days, the TS inspection and evaluation index can be used to carry out visibility classification inspection and evaluation for the above two methods within the 24-hour forecast time period and the 27-168 forecast time period, and according to the TS score, the forecast weight coefficients of the two methods are obtained, and the weight coefficients are used in the current forecast product, and finally the products within 24 and 27-168 forecast time periods are spliced ​​together, that is, the final visibility correction integrated forecast result within the 24-hour forecast time period = "PDF+1D-Kalman" visibility correction result * "PDF+1D-Kalman" weight coefficient + "MBC+1D-Kalman" visibility correction result * "MBC+1D-Kalman" weight coefficient. The visibility correction integrated forecast result for the forecast period of 27-168 hours = "PDF" visibility correction result * "PDF" weight coefficient + "MBC" visibility correction result * "MBC" weight coefficient.

[0049] This embodiment avoids the limitations of a single method: atmospheric visibility is closely related to the content of water vapor and aerosol in the air, and is also affected by local wind speed, wind direction, precipitation, and particle concentration. Whether it is a short-term forecast, a short-term forecast, or a medium- and long-term forecast, there are many factors that need to be considered in the visibility forecast modeling process. Existing studies are mostly based on a single station or focus on a certain area, and only focus on a single method to correct the model visibility elements.

[0050] This method uses a variety of objective correction techniques and fully combines the actual visibility observation data, so that the impact of systematic errors and random errors on visibility forecast deviations has been significantly reduced, and the forecast errors have been treated differently. In this method, a real-time verification step is added to optimize the weight coefficients of different correction methods, that is, the principle of "good forecasts have a large weight and poor forecasts have a small weight", and finally an optimal solution is obtained, saving the time cost of weather forecasters to select different methods.

[0051] The present application also provides a visibility forecast error processing device, comprising a first correction unit, configured to respectively use a preset probability density matching method and a sliding double weight method to correct the original forecast product to obtain the future n 0 -n 2 Forecast validity: two first visibility correction forecast products; the second correction unit is configured for n 0 -n 1Based on the first-order Kalman filter and the historical visibility conditions, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products; the two first visibility correction forecast products are used as the forecast products for n 1 -n 2 A visibility correction forecast product with a forecast period; an integration unit configured to integrate the two second visibility correction forecast products and the visibility correction forecast product.

[0052] Optionally, the preset probability density matching method and sliding double weight method are used to revise the original forecast product to obtain the future n 0 -n 2 The two first visibility correction forecast products of forecast time include: adjusting the probability density function of the visibility predicted by the original forecast product at each forecast site according to the probability density function of the actual visibility so as to make it consistent with the probability density distribution of the actual visibility; and correcting the visibility predicted by the original product at each forecast site by using the sliding double weighted average method based on the observed incremental double weighted average.

[0053] Optionally, with respect to the probability density function of actual visibility, adjusting the corresponding probability density function of the visibility predicted by the original forecast product at each prediction site includes: dividing the forecast visibility and the actual visibility of the original forecast product into m threshold intervals; calculating a first frequency when the actual visibility appears at different visibility thresholds to determine the probability density function of the actual visibility and the frequency of the actual visibility; calculating a second frequency when the forecast visibility appears at different visibility thresholds; and determining the revised forecast value of visibility based on the second frequency and the probability density function of the actual visibility.

[0054] Optionally, based on the observation increment double weighted average, the original product is corrected at each forecast site using a sliding double weighted average method, including: based on The forecast visibility is corrected by: is the double-weighted average of the observation increments at the kth station.

[0055] Optionally, in the calculation When including: in, For the kth station The median of is the corresponding median absolute deviation, that is The median of is the weight, when When Each station Double weighted average for:

[0056] Optionally, for n 0 -n 1 Forecast timeliness, based on the first-order Kalman filter combined with the historical visibility actual situation, the two first visibility correction forecast products are corrected twice to obtain two second visibility correction forecast products including: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value; when t=1, B t-1 =0.

[0057] Optionally, integrating the two second visibility correction forecast products and the visibility correction forecast product comprises: integrating the two second visibility correction forecast products respectively to obtain n 0 -n 1 A forecast product with a forecast timeliness, wherein during integration, a test is performed based on the TS score to obtain a first TS score result; based on the size of the first TS score result, weight coefficients corresponding to each second visibility correction forecast product are determined; integration is performed based on the weight coefficients; the visibility correction forecast products are respectively integrated to obtain n 1 -n 2 Forecast products with forecast effectiveness; wherein, during integration, a test is performed based on the TS score to obtain a second TS score result; based on the size of the second TS score result, the weight coefficients corresponding to each second visibility correction forecast product are determined respectively; and integration is performed based on the weight coefficients.

[0058] Figure 3 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the electronic device 50 includes: a processor 501 (processor), a memory 502 (memory) and a bus 503;

[0059] The processor 501 and the memory 502 communicate with each other via the bus 503 ; the processor 501 is used to call program instructions in the memory 502 to execute the methods provided by the above-mentioned method embodiments.

[0060] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions. The computer instructions enable a computer to execute the methods provided by the above method embodiments.

[0061] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc., various storage media that can store program codes.

[0062] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0063] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0064] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A method for correcting integrated visibility forecast, characterized in that: include: The preset probability density matching method and the sliding double weight method are respectively used to correct the original forecast product to obtain two first visibility correction forecast products for the future n0-n2 forecast period, wherein, with respect to the probability density function of the actual visibility, the corresponding probability density function of the visibility predicted by the original forecast product at each forecast site is adjusted to make it consistent with the probability density distribution of the actual visibility; and based on the observation increment double weight average, the sliding double weight average method is used to correct the visibility predicted by the original product at each forecast site; with respect to the actual The probability density function of the actual visibility is obtained by adjusting the corresponding probability density function of the visibility predicted by the original forecast product at each forecasting site, including: dividing the forecast visibility and the actual visibility of the original forecast product into m threshold intervals; calculating the first frequency when the actual visibility appears at different visibility thresholds to determine the probability density function of the actual visibility and the frequency of the actual visibility; calculating the second frequency when the forecast visibility appears at different visibility thresholds; and determining the forecast value of visibility after correction based on the second frequency and the probability density function of the actual visibility; According to the n0-n1 forecast timeliness, the two first visibility revised forecast products are revised twice based on the first-order Kalman filter combined with the historical visibility conditions to obtain two second visibility revised forecast products; The two first visibility correction forecast products are used as the visibility correction forecast products for the n1-n2 forecast period; The two second visibility correction forecast products and the visibility correction forecast product are integrated, wherein the two second visibility correction forecast products are respectively integrated to obtain a forecast product with a forecast time of n0-n1, wherein during integration, a test is performed based on the TS score to obtain a first TS score result; based on the size of the first TS score result, the weight coefficient corresponding to each second visibility correction forecast product is determined; based on the weight coefficient, the integration is performed; the visibility correction forecast products are respectively integrated to obtain a forecast product with a forecast time of n1-n2; wherein, during integration, a test is performed based on the TS score to obtain a second TS score result; based on the size of the second TS score result, the weight coefficient corresponding to each second visibility correction forecast product is determined; based on the weight coefficient, the integration is performed; Based on the observation increment double weighted average, the visibility forecasted by the original product at each forecast station is corrected using the sliding double weighted average method, including: based on The forecast visibility is corrected by: is the double-weighted average correction term of the observation increment at the kth station; In calculation When including: in, For the kth station the median of is the corresponding median absolute deviation, that is the median of is the weight, when When Each station The double weighted average correction term for: According to the n0-n1 forecast timeliness, the two first visibility correction forecast products are corrected twice based on the first-order Kalman filter combined with the historical visibility conditions, and the two second visibility correction forecast products are obtained, including: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value; when t=1, B t-1 =0.

2. A method and apparatus for processing visibility correction integrated forecast, characterized in that: include: The first correction unit is configured to respectively adopt a preset probability density matching method and a sliding double weight method to correct the original forecast product to obtain two first visibility correction forecast products for the future n0-n2 forecast period, wherein, with respect to the probability density function of the actual visibility, the corresponding probability density function of the visibility predicted by the original forecast product at each forecast site is adjusted to make it consistent with the probability density distribution of the actual visibility; and based on the observation increment double weight average, the visibility predicted by the original product at each forecast site is corrected by using the sliding double weight average method. Positive; with respect to the probability density function of the actual visibility, adjusting the corresponding probability density function of the visibility predicted by the original forecast product at each forecasting site includes: dividing the forecast visibility and the actual visibility of the original forecast product into m threshold intervals; calculating the first frequency when the actual visibility appears at different visibility thresholds to determine the probability density function of the actual visibility and the frequency of the actual visibility; calculating the second frequency when the forecast visibility appears at different visibility thresholds; determining the forecast value of visibility after correction based on the second frequency and the probability density function of the actual visibility; The second correction unit is configured to perform secondary correction on the two first visibility correction forecast products for the n0-n1 forecast time period based on the first-order Kalman filter combined with the historical visibility conditions to obtain two second visibility correction forecast products; and use the two first visibility correction forecast products as visibility correction forecast products for the n1-n2 forecast time period; The integration unit is configured to integrate the two second visibility correction forecast products and the visibility correction forecast product, wherein the two second visibility correction forecast products are respectively integrated to obtain a forecast product with a forecast time of n0-n1, wherein during integration, a test is performed based on a TS score to obtain a first TS score result; based on the size of the first TS score result, a weight coefficient corresponding to each second visibility correction forecast product is determined; integration is performed based on the weight coefficient; the visibility correction forecast products are respectively integrated to obtain a forecast product with a forecast time of n1-n2; wherein during integration, a test is performed based on a TS score to obtain a second TS score result; based on the size of the second TS score result, a weight coefficient corresponding to each second visibility correction forecast product is determined; integration is performed based on the weight coefficient; Based on the observation increment double weighted average, the visibility forecasted by the original product at each forecast station is corrected using the sliding double weighted average method, including: based on The forecast visibility is corrected by: is the double-weighted average correction term of the observation increment at the kth station; In calculation When including: in, For the kth station the median of is the corresponding median absolute deviation, that is the median of is the weight, when When Each station The double weighted average correction term for: According to the n0-n1 forecast timeliness, the two first visibility correction forecast products are corrected twice based on the first-order Kalman filter combined with the historical visibility conditions, and the two second visibility correction forecast products are obtained, including: B t =(1-w)×B t-1 +w×(F t -A t ), where t is the current forecast time; B t is the deviation correction value at time t; B t-1 is the deviation correction value of the previous moment; w is the decreasing average weight parameter; A t is the actual value at time t, F t is the corresponding forecast value; when t=1, B t-1 =0.

3. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in claim 1 is implemented.

4. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to claim 1 is implemented.

Images