A method and device for intensity correction of static infrared cloud images

Abstract

The application discloses a kind of intensity revision methods for static infrared forecast cloud chart, comprising: obtaining spatial one-to-one corresponding original forecast cloud chart, original measured cloud chart, original total cloud amount forecast chart, clear sky background brightness temperature field, to be revised forecast cloud chart and future total cloud amount forecast chart;Cloud area identification is carried out to original measured cloud chart according to clear sky background brightness temperature field, and first cloud area identification result is obtained;According to first preset cloud amount threshold and original total cloud amount forecast chart, the cloud area of original forecast cloud chart is identified, and second cloud area identification result is obtained;According to second preset cloud amount threshold and future total cloud amount forecast chart, the cloud area of to be revised forecast cloud chart is identified, and third cloud area identification result is obtained;According to first cloud area identification result, second cloud area identification result, third cloud area identification result and clear sky background brightness temperature field, intensity revision is carried out to to be revised forecast cloud chart, and target forecast cloud chart is obtained.The method can combine the different error characteristics of cloud area and clear sky area to improve the revision accuracy.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to an intensity correction method and apparatus for stationary infrared forecast cloud images. Background Technology

[0002] With advancements in numerical weather prediction and satellite remote sensing technologies, forecast cloud images have demonstrated significant application potential in meteorological monitoring and short-term forecasting. However, due to inherent forecast biases in numerical models and uncertainties in radiative transfer process simulations, forecast cloud images generated by existing methods still exhibit significant brightness temperature errors, particularly in cloud regions, where errors can reach tens of Kelvin, severely limiting the direct application value of the products. Current technologies typically involve directly performing global statistical corrections on the forecast field; however, these methods fail to fully consider the fundamental differences in radiative characteristics between clouds and clear-sky underlying surfaces, leading to blurred cloud structure or distorted radiative balance in clear-sky regions after correction. Therefore, the insufficient correction accuracy of existing methods stems from their failure to distinguish the different error characteristics of cloud and clear-sky regions. Summary of the Invention

[0003] In view of the above problems, this application provides an intensity correction method and apparatus for stationary infrared forecast cloud images, which can solve the problem of insufficient correction accuracy caused by the failure of existing methods to distinguish the different error characteristics of cloud areas and clear sky areas.

[0004] Firstly, this application provides an intensity correction method for stationary infrared forecast cloud images, including: Acquire spatially corresponding original forecast cloud images, original measured cloud images, original total cloud cover forecast images, clear sky background brightness temperature field, forecast cloud images to be corrected, and future total cloud cover forecast images; wherein, the original forecast cloud images and the original total cloud cover forecast images are forecast data at historical target times, the original measured cloud images are measured data at the historical target times, and the forecast cloud images to be corrected and the future total cloud cover forecast images are forecast data at future target times; Based on the clear sky background brightness temperature field, cloud areas are identified in the original measured cloud map to obtain the first cloud area identification result; The cloud regions in the original forecast cloud map are identified based on the first preset cloud cover threshold and the original total cloud cover forecast map to obtain the second cloud region identification result; The cloud regions of the forecast cloud map to be corrected are identified based on the second preset cloud cover threshold and the forecast map of total future cloud cover, and the third cloud region identification result is obtained. The intensity of the forecast cloud map to be corrected is corrected based on the first cloud area identification result, the second cloud area identification result, the third cloud area identification result, and the clear sky background brightness temperature field to obtain the target forecast cloud map.

[0005] In the above technical solution, the method can determine the real cloud area of ​​the original measured cloud map by combining the clear sky background brightness temperature field, and determine the forecast cloud area of ​​the original forecast cloud map and the corrected forecast cloud map by combining the cloud cover threshold. Then, by analyzing the difference between the real cloud area and the forecast cloud area, the cloud area intensity of the future forecast cloud map to be corrected is corrected using the difference information, so as to obtain a target forecast cloud map that is more in line with the actual observation.

[0006] In some implementations, the step of performing intensity correction on the forecast cloud image to be corrected based on the first cloud region identification result, the second cloud region identification result, the third cloud region identification result, and the clear sky background brightness temperature field to obtain the target forecast cloud image includes: A grading correction model is established based on the identification results of the first cloud area, the identification results of the second cloud area, and the preset grading interval; The predicted cloud map to be corrected is corrected according to the tiered correction model to obtain the tiered corrected cloud map. Based on the clear sky background brightness temperature field and the third cloud area identification result, the graded correction cloud map is corrected for clear sky conditions to obtain the target forecast cloud map.

[0007] In the above technical solution, the method can perform differentiated correction of cloud areas by graded intervals, and perform clear sky correction of non-cloud areas by combining clear sky background brightness temperature field, thereby achieving hierarchical and refined intensity correction and further improving the accuracy of target forecast cloud images.

[0008] In some implementations, establishing a tiered correction model based on the first cloud region identification result, the second cloud region identification result, and a preset tiering interval includes: Based on the first cloud area identification result and the second cloud area identification result, a set of pixel samples for tiered statistics is selected from the original forecast cloud map; the pixel samples in the set of pixel samples are simultaneously marked as cloud areas or clear sky areas in the first cloud area identification result and the second cloud area identification result. The grading error is calculated based on the pixel sample set and the preset grading interval to obtain the grading error corresponding to each of the multiple gradings. The tiered correction model includes multiple tiers and the tiered errors corresponding to each of the multiple tiers.

[0009] In the above technical solution, the method can perform grading error statistics by screening pixel samples whose forecast and actual cloud conditions are consistent, effectively avoiding the interference of cloud area and clear sky area confusion on error calculation; at the same time, by using the grading correction model built based on grading error to accurately update the brightness temperature value of the pixel to be corrected, the reliability and accuracy of grading correction can also be effectively improved.

[0010] In some embodiments, the step of performing tiered correction on the forecast cloud map to be corrected according to the tiered correction model to obtain a tiered corrected cloud map includes: Based on the tiered correction model, multiple correction errors are determined that correspond one-to-one with the multiple pixels to be corrected in the forecast cloud image to be corrected; For the predicted brightness temperature values ​​and correction errors that have a one-to-one correspondence among the plurality of pixels to be corrected, calculate a plurality of corrected brightness temperature values ​​that correspond one-to-one with all the plurality of pixels to be corrected. The brightness temperature values ​​of the forecast cloud map to be corrected are updated based on all corrected brightness temperature values ​​to obtain graded corrected cloud maps.

[0011] In the above technical solution, the method can perform tiered correction of the predicted brightness temperature value of the cloud map to be corrected through a tiered correction model, thereby eliminating the prediction deviation characteristics under different brightness temperature ranges and providing a high-precision data foundation for subsequent intensity correction.

[0012] In some implementations, the step of calculating the grading error based on the pixel sample set and a preset grading interval to obtain the grading error corresponding to each of the multiple grading intervals includes: Obtain the predicted brightness temperature values ​​of all pixel samples in the original predicted cloud image and the measured brightness temperature values ​​in the original measured cloud image; Based on the predicted brightness temperature values ​​of all pixel samples, all pixel samples are divided into multiple corresponding sub-sub ... Calculate the average error between the predicted brightness temperature value and the measured brightness temperature value of all pixel samples in each graded sample set to obtain the graded error corresponding to each of the multiple grades.

[0013] In the above technical solution, the method can classify the samples according to the predicted brightness temperature value and use the average error between the predicted and measured brightness temperature of the samples in each class as the classification error, thereby accurately characterizing the prediction deviation characteristics under different brightness temperature ranges, and providing a high-precision error benchmark with clear physical meaning for subsequent intensity correction.

[0014] In some embodiments, the step of performing clear-sky correction on the graded corrected cloud map based on the clear-sky background brightness temperature field and the third cloud region identification result to obtain the target forecast cloud map includes: Based on the third cloud region identification result, determine multiple clear sky region pixels in the forecast cloud image to be corrected, and multiple corrected brightness temperature values ​​corresponding to the multiple clear sky region pixels. Based on the clear sky background brightness temperature field, obtain multiple clear sky background field brightness temperature values ​​that correspond one-to-one with the multiple clear sky area pixels; The absolute difference between the plurality of corrected brightness temperature values ​​and the plurality of clear sky background field brightness temperature values ​​is calculated to obtain the plurality of absolute differences corresponding to the plurality of clear sky area pixels. In the graded correction cloud map, the clear-sky area pixels corresponding to the absolute difference value that is less than or equal to the preset difference value are determined as pixels to be processed; The brightness temperature value of the pixel to be processed in the graded correction cloud map is replaced with the corresponding brightness temperature value of the clear sky background field to obtain the clear sky corrected target forecast cloud map.

[0015] In the above technical solution, the method can optimize the brightness temperature value of some pixels in the classification correction result by using the brightness temperature field of clear sky background, so as to make the brightness temperature value of the pixels more accurate, and thus obtain a more accurate target forecast cloud map.

[0016] In some embodiments, the step of identifying cloud regions based on the clear-sky background brightness temperature field of the original measured cloud image to obtain a first cloud region identification result includes: Obtain pixel information of all measured pixels in the original measured cloud image; wherein, each pixel information includes a pixel position and a measured brightness temperature value with a one-to-one correspondence. Obtain the brightness temperature value of the clear sky background field for all grid cells in the clear sky background brightness temperature field; wherein, each clear sky background field brightness temperature value corresponds one-to-one with each cell position; Multiple brightness temperature difference values ​​are obtained by calculating the difference between all measured brightness temperature values ​​and all clear sky background field brightness temperature values; All measured pixels are marked according to the multiple brightness temperature difference values ​​and the preset cloud detection threshold to obtain measured pixels in the cloud area and measured pixels in the clear sky area. The labeling results of the measured pixels in the cloud area and the measured pixels in the clear sky area are binarized to obtain the first cloud area identification result.

[0017] In the above technical solution, the method can accurately divide the cloud area and the clear sky area in the original measured cloud map by comparing the difference between the measured brightness temperature and the clear sky background brightness temperature and combining it with threshold judgment, thereby obtaining a more reliable first cloud area identification result.

[0018] In some implementations, the step of identifying cloud regions in the original forecast cloud map based on a first preset cloud cover threshold and the original total cloud cover forecast map to obtain a second cloud region identification result includes: Obtain the total cloud cover forecast data of the original forecast cloud map based on the original total cloud cover forecast map; Based on the total cloud cover forecast data, determine the cloud cover value of all forecast pixels in the original forecast cloud map; All forecast pixels are labeled based on the cloud cover value and the first preset cloud cover threshold to obtain cloud area forecast pixels and clear sky area forecast pixels. The labeling results of the cloud area forecast pixels and the clear sky area forecast pixels are binarized to obtain the second cloud area identification result.

[0019] In the above technical solution, the method can determine the cloud amount value of each pixel based on the total cloud amount forecast data of the original forecast cloud map, and achieve objective division between the forecast cloud area and the clear sky area through the cloud amount threshold, thereby obtaining a more reliable second cloud area identification result.

[0020] Secondly, this application provides an intensity correction device for stationary infrared forecast cloud images, the intensity correction device for stationary infrared forecast cloud images comprising: The acquisition unit is used to acquire, in a spatially corresponding manner, the original forecast cloud image, the original measured cloud image, the clear sky background brightness temperature field, and the forecast cloud image to be corrected; wherein, the original forecast cloud image is the forecast data at the historical target time, the original measured cloud image is the measured data at the historical target time, and the forecast cloud image to be corrected is the forecast data at the future target time. The first identification unit is used to identify cloud areas in the original measured cloud map based on the clear sky background brightness temperature field, and obtain the first cloud area identification result. The second identification unit is used to identify cloud areas in the original forecast cloud map according to the first preset cloud cover threshold, and obtain the second cloud area identification result. An intensity correction unit is used to correct the intensity of the forecast cloud map to be corrected based on the first cloud area identification result, the second cloud area identification result, and the clear sky background brightness temperature field, so as to obtain the target forecast cloud map.

[0021] In the above technical solution, the device can determine the real cloud area of ​​the original measured cloud map by combining the clear sky background brightness temperature field, and determine the forecast cloud area of ​​the original forecast cloud map and the corrected forecast cloud map by combining the cloud cover threshold. Then, by analyzing the difference between the real cloud area and the forecast cloud area, the device can use the difference information to correct the cloud area intensity of the future forecast cloud map to be corrected, thereby obtaining a target forecast cloud map that is more in line with actual observation.

[0022] Thirdly, this application provides an electronic device including a memory and a processor, the memory storing a computer program, the processor running the computer program to cause the electronic device to perform the intensity correction method for stationary infrared forecast cloud images as described in any of the first aspects.

[0023] Fourthly, this application provides a readable storage medium storing a computer program, which, when executed by a processor, performs the intensity correction method for stationary infrared forecast cloud images as described in any one of the first aspects.

[0024] Fifthly, this application provides a computer program product comprising a computer program that, when executed by a processor, performs the intensity correction method for stationary infrared forecast cloud images as described in any one of the first aspects.

[0025] The beneficial effects of this application are as follows: By combining tiered correction and clear-sky correction strategies, the problem of large forecast cloud image errors is effectively solved. Specifically, tiered correction makes the forecast brightness temperature distribution closer to the actual measurement and improves the forecast for low-temperature areas, while clear-sky correction optimizes the performance in high-temperature areas. The combination of the two significantly reduces the overall forecast error. At the same time, through a real-time call mechanism, this application can also realize dynamic correction of forecast cloud images, meeting the needs of operational applications for real-time and high-precision forecasts. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating an intensity correction method for a stationary infrared forecast cloud image according to an embodiment of this application. Figure 2 This is a schematic diagram of the statistical results of brightness temperature error grading in an embodiment of this application; Figure 3 This is a schematic diagram of a classification correction result in an embodiment of this application; Figure 4 This is a schematic diagram of a clear sky correction result in an embodiment of this application; Figure 5 This is a comparative chart of the statistical distribution of brightness temperature bars in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an intensity correction device for a stationary infrared forecast cloud image according to an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0028] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0030] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more (including two), similarly, "multiple sets" refers to two or more sets (including two sets), and "multiple pieces" refers to two or more pieces (including two pieces) unless otherwise explicitly defined.

[0031] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0032] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0033] Currently, although advancements in numerical weather prediction, radiative transfer, and machine learning technologies have enabled forecast cloud images to possess characteristics similar to observed cloud images, the process of mapping forecast cloud images from numerical weather prediction products to a satellite perspective still involves the superposition of errors from the numerical weather prediction itself and model conversion. Therefore, these errors inevitably lead to corresponding forecast biases. For example, the root mean square error (RMSE) of the FY-4 satellite's longwave infrared channel is as high as 10–27 K, while cloud area errors are even more significant.

[0034] To address the aforementioned technical problems, this application provides an intensity correction method for stationary infrared forecast cloud images. This method utilizes the deviation patterns between historical forecasts and historical realities to correct the current forecast, thereby restoring a result closer to the current reality. Furthermore, this method can further approximate the forecast distribution to actual measurements by tiered correction to fit the overall deviation, and by using clear-sky correction to constrain high-temperature clear-sky areas, thus achieving accurate restoration of the current forecast cloud image.

[0035] like Figure 1 As shown, some embodiments of this application provide an intensity correction method for stationary infrared forecast cloud images, which includes: S100. Acquire spatially corresponding original forecast cloud images, original measured cloud images, original total cloud cover forecast images, clear sky background brightness temperature field, forecast cloud images to be corrected, and future total cloud cover forecast images; wherein, the original forecast cloud images and original total cloud cover forecast images are forecast data at historical target times, the original measured cloud images are measured data at historical target times, and the forecast cloud images to be corrected and future total cloud cover forecast images are forecast data at future target times.

[0036] In this embodiment, the original forecast cloud image and the original measured cloud image can both be related data from the FY-4 long-wave infrared channel (e.g., 10.3~11.3μm); or they can both be related data from the FY-4 visible light channel (e.g., 0.5~0.75μm).

[0037] In this embodiment, the historical target time corresponds to the time 2 hours ago (represented on the number line as -2 to 0h, where 0 represents the current time), and the future target time corresponds to the time 2 hours later (represented on the number line as 0 to 2h). Alternatively, the historical and future target times can be the time 6 hours ago (-6 to 0h) and the time 6 hours later (0 to 6h). It should be understood that the historical and future target times can be any custom times determined based on timeliness requirements and the stability of forecast error statistics.

[0038] In this embodiment, the forecast cloud image to be corrected can be obtained through the operational system or generated through numerical weather prediction models and radiation transfer models.

[0039] As an optional implementation, this method can also support dynamic invocation and updating: When both measured and forecasted cloud images exist simultaneously, the intensity correction process is initiated. Combined with FY-4 measured cloud Figure 1 With an update frequency of 15 minutes, the statistical results of grading error and clear sky error can be dynamically updated as quickly as every 15 minutes.

[0040] This implementation method enables high-frequency dynamic correction of forecast cloud images, thereby continuously ensuring the timeliness and accuracy of target forecast cloud images.

[0041] S200. Based on the brightness temperature field of the clear sky background, cloud areas are identified in the original measured cloud map to obtain the first cloud area identification result.

[0042] In this embodiment, the clear-sky background brightness temperature field can be composed of the maximum brightness temperature values ​​of each satellite pixel at the same time over the past few days, and this maximum brightness temperature value is identified as the clear-sky brightness temperature of the corresponding pixel. Alternatively, it can be an average brightness temperature field established based on historical multi-year climate data, or a simulated clear-sky radiation field output by a numerical weather prediction model.

[0043] In this embodiment, cloud area identification employs the brightness temperature background map method, which determines cloud areas by comparing the difference between the measured pixel brightness temperature value and the corresponding brightness temperature value in the clear sky background brightness temperature field. Alternatively, cloud area identification can also utilize a multi-channel threshold-based method or a cloud detection algorithm combining texture features and a machine learning model.

[0044] S300. Based on the first preset cloud cover threshold and the original total cloud cover forecast map, the cloud areas of the original forecast cloud map are identified to obtain the second cloud area identification result.

[0045] In this embodiment, the original forecast cloud image can be composed of multiple pixels, and the numerical model will assign a value of 0 to 10% of the total cloud cover to each pixel (representing the proportion of the pixel covered by clouds).

[0046] In this embodiment, the method can use a more stringent threshold in this step. For example, only if the assigned value of a pixel is greater than 80% (e.g., 82%, 90%, 100%) will the pixel be included in the graded correction cloud area sample; there will be many such pixels that meet the conditions in the whole image, which together constitute the cloud area sample set.

[0047] In this embodiment, based on the same principle, the method can include pixels with an amplitude of 0 into the graded correction clear sky area sample; there will be many such pixels in the whole image that meet the conditions, which together constitute the clear sky area sample set.

[0048] S400. Based on the second preset cloud cover threshold and the future total cloud cover forecast map, identify the cloud areas in the forecast cloud map to be corrected, and obtain the third cloud area identification result.

[0049] In this embodiment, the forecast cloud image to be corrected corresponds spatially to the original forecast cloud image, so the two have the same pixel arrangement and number of pixels. At the same time, for the forecast cloud image to be corrected, the numerical model will also assign a value of 0 to 10% of the total cloud cover to each pixel (representing the proportion of the pixel covered by clouds).

[0050] In this embodiment, the method can use a relatively lenient threshold in this step. For example, as long as the value of the pixel is greater than 30%, the pixel is considered a cloud area pixel; as long as the amplitude of the pixel is less than or equal to 30%, the pixel is considered a clear sky area pixel.

[0051] S500: Based on the identification results of the first cloud region, the second cloud region, the third cloud region, and the brightness temperature field of the clear sky background, the intensity of the forecast cloud map to be corrected is corrected to obtain the target forecast cloud map.

[0052] In this embodiment, intensity correction is a numerical correction of the brightness temperature value in the forecast cloud image. This correction specifically includes two steps: tiered correction and clear sky correction, which are organically combined to achieve refined intensity correction of the near-term (0-2h) forecast cloud image.

[0053] In the above embodiments, the method can determine the real cloud area of ​​the original measured cloud map by combining the clear sky background brightness temperature field, and determine the forecast cloud area of ​​the original forecast cloud map and the corrected forecast cloud map by combining the cloud cover threshold. Then, by analyzing the difference between the real cloud area and the forecast cloud area, the cloud area intensity of the future forecast cloud map to be corrected is corrected using the difference information, so as to obtain a target forecast cloud map that is more in line with the actual observation.

[0054] In some embodiments, step S200 may include: S210. Obtain pixel information of all measured pixels in the original measured cloud image; wherein, each pixel information includes the pixel position and measured brightness temperature value with a one-to-one correspondence.

[0055] In this embodiment, the pixel information of the original measured cloud image can be directly extracted from satellite observation data. This method requires ensuring that the positional information of each pixel accurately matches the measured brightness temperature value, thereby providing fundamental data support for subsequent cloud area identification.

[0056] S220. Obtain the brightness temperature value of the clear sky background field for all grid cells in the clear sky background brightness temperature field; wherein, each clear sky background field brightness temperature value corresponds one-to-one with the position of each cell.

[0057] In this embodiment, the brightness temperature value of the clear sky background field can be obtained by statistically analyzing the maximum brightness temperature values ​​of each pixel of the satellite at the same time over the past few days. Each pixel location corresponds to a unique brightness temperature value of the clear sky background field, forming a complete clear sky background brightness temperature field.

[0058] S230. Calculate the difference between all measured brightness temperature values ​​and all clear sky background brightness temperature values ​​to obtain multiple brightness temperature difference values.

[0059] In this embodiment, the brightness temperature difference value can be obtained by subtracting the brightness temperature value of the clear sky background field at the corresponding location from the measured brightness temperature value of each measured pixel. This method can achieve quantification of differences across all pixels through pixel-by-pixel calculation.

[0060] S240. Based on multiple brightness temperature difference values ​​and a preset cloud detection threshold, all measured pixels are marked to obtain measured pixels in the cloud area and measured pixels in the clear sky area.

[0061] In this embodiment, when the brightness temperature difference is greater than the preset cloud detection threshold, the pixel is determined to be covered by clouds and marked as a measured pixel in the cloud area; otherwise, it is marked as a measured pixel in the clear sky area, thereby achieving the initial division between the cloud area and the clear sky area.

[0062] In this embodiment, the preset cloud detection threshold can be determined based on factors such as historical data statistical analysis or experience.

[0063] S250. The labeling results of the measured pixels in the cloud area and the measured pixels in the clear sky area are binarized to obtain the first cloud area identification result.

[0064] In this embodiment, the binarization process uses a 0 / 1 encoding rule. A value of 1 indicates that the pixel is a measured pixel in a cloud area, and a value of 0 indicates that the pixel is a measured pixel in a clear sky area. This allows for the final standardized first cloud area identification result.

[0065] In the above embodiments, the method can accurately divide the cloud area and the clear sky area in the original measured cloud map by comparing the difference between the measured brightness temperature and the clear sky background brightness temperature and combining it with threshold judgment, thereby obtaining a more reliable first cloud area identification result.

[0066] In some embodiments, step S300 may include: S310. Obtain the total cloud cover forecast data from the original total cloud cover forecast map.

[0067] In this embodiment, the total cloud cover forecast data can be obtained from numerical weather prediction models.

[0068] In this embodiment, the data format of the total cloud cover forecast data can be a quantitative index of 0-10, which is used to directly reflect the cloud cover forecast information of each region.

[0069] S320. Based on the total cloud cover forecast data, determine the cloud cover value of all forecast pixels in the original forecast cloud map.

[0070] In this embodiment, the method can decompose the total cloud cover forecast data according to the pixel position, so that each forecast pixel corresponds to a unique cloud cover value, ensuring accurate spatial matching of cloud cover information.

[0071] S330. Mark all forecast pixels according to the cloud cover value of all forecast pixels and the first preset cloud cover threshold to obtain cloud area forecast pixels and clear sky area forecast pixels.

[0072] In this embodiment, the original forecast cloud image can be composed of multiple pixels, and the numerical model will assign a value of 0 to 10% of the total cloud cover to each pixel (representing the proportion of the pixel covered by clouds).

[0073] In this embodiment, the method can use a more stringent threshold in this step. For example, only if the assigned value of a pixel is greater than 80% (e.g., 82%, 90%, 100%) will the pixel be included in the graded correction cloud area sample; there will be many such pixels that meet the conditions in the whole image, which together constitute the cloud area sample set.

[0074] In this embodiment, based on the same principle, the method can include pixels with an amplitude of 0 into the graded correction clear sky area sample; there will be many such pixels in the whole image that meet the conditions, which together constitute the clear sky area sample set.

[0075] S340. The labeling results of cloud area forecast pixels and clear sky area forecast pixels are binarized to obtain the second cloud area identification result.

[0076] In this embodiment, the method can adopt the same 0 / 1 binary encoding rule as the first cloud area identification result. Here, a value of 1 represents a cloud area forecast pixel, and a value of 0 represents a clear sky area forecast pixel, thus ensuring the format consistency of the identification results for both clear sky and cloud areas.

[0077] In the above embodiments, the method can determine the cloud cover value of each pixel based on the total cloud cover forecast data of the original forecast cloud map, and achieve objective division between the forecast cloud area and the clear sky area through the cloud cover threshold, thereby obtaining a more reliable second cloud area identification result.

[0078] In some embodiments, step S500 may include: S510. Based on the identification results of the first cloud area, the identification results of the second cloud area, and the preset grading interval, establish a grading correction model.

[0079] In this embodiment, the preset interval for the grading is 1K (Kelvin).

[0080] In this embodiment, the original measured cloud map corresponds to a total brightness temperature range (A, B). This method can divide this range (A, B) into multiple smaller ranges according to a preset interval; each smaller range is a interval.

[0081] In this embodiment, the core essence of the tiered correction model is to discretize the continuous brightness temperature forecast values ​​into several tiers at fixed intervals, independently statistically analyze the deviation pattern between the forecast and the actual measurement for each tier, and then use the deviation pattern of that tier to correct the forecast values ​​of the same tier.

[0082] To effectively improve the reliability and accuracy of the classification correction, step S510 may further include: S511. Based on the first cloud area identification result and the second cloud area identification result, select a set of pixel samples for tiered statistics from the original forecast cloud map; the pixel samples in the set of pixel samples are simultaneously marked as cloud areas or clear sky areas in the first cloud area identification result and the second cloud area identification result.

[0083] S512. Calculate the grading error based on the pixel sample set and the preset grading interval to obtain the grading error corresponding to each of the multiple gradings; wherein, the multiple gradings and the grading errors corresponding to each of the multiple gradings constitute the grading correction model.

[0084] In this embodiment, the tiered error calculation must meet the statistical condition that the number of samples in each tier is at least 10. Furthermore, this method only calculates the error for tiers that meet this condition, ensuring the statistical significance of the error results.

[0085] As an optional implementation, the grading error is calculated based on the pixel sample set and a preset grading interval to obtain the grading error corresponding to each of the multiple grading intervals, including: Obtain the predicted brightness temperature values ​​of all pixel samples in the original predicted cloud image and the measured brightness temperature values ​​in the original measured cloud image; Based on the predicted brightness temperature values ​​of all pixel samples, all pixel samples are divided into multiple corresponding sub-sub ... Calculate the average error between the predicted brightness temperature value and the measured brightness temperature value of all pixel samples in each grade sample set to obtain the grade error corresponding to each grade.

[0086] In this embodiment, the average error is expressed by the formula The calculation yielded the result. Where N... k Let S be the total number of samples in the k-th rank. i Let O be the predicted brightness temperature value for the i-th sample. i Let be the measured brightness temperature value of the i-th sample.

[0087] S520. Based on the tiered correction model, the forecast cloud map to be corrected is tiered and corrected to obtain the tiered corrected cloud map.

[0088] In this embodiment, the core of the graded correction is to use the statistical results of the brightness temperature error between the forecast and the actual measurement at the historical target time (-2~0h) to make targeted corrections to the brightness temperature of the pixels to be corrected at the future target time (0~2h).

[0089] In this embodiment, the target of the tiered correction is all pixels in the forecast cloud image to be corrected, that is, the subsequent pixels to be corrected are any pixel in the forecast to be corrected.

[0090] To eliminate forecast bias characteristics across different brightness temperature ranges, step S520 may further include: S521. Based on the tiered correction model, determine multiple correction errors that correspond one-to-one with multiple pixels to be corrected in the forecast cloud image to be corrected.

[0091] In this embodiment, the method can determine the classification index k to which the predicted brightness temperature value of the pixel to be corrected belongs (k=index(S)). 预报值 The average error corresponding to the classification is taken as the classification error of the pixel to be corrected.

[0092] In this embodiment, one pixel to be corrected corresponds to one unique correction error. However, multiple correction errors corresponding to multiple pixels to be corrected may be the same, that is, the same classification error.

[0093] For example, the first pixel to be corrected corresponds to the first correction error, and the second pixel to be corrected corresponds to the second correction error. The first correction error can be equal to the second correction error. In this case, both the first correction error and the second correction error are the same level error.

[0094] S522. For the predicted brightness temperature values ​​and correction errors that have a one-to-one correspondence among multiple pixels to be corrected, calculate multiple corrected brightness temperature values ​​that correspond one-to-one with all multiple pixels to be corrected.

[0095] In this embodiment, the corrected brightness temperature value can be calculated using the following formula: ; This method can add the predicted brightness temperature value of the pixel to be corrected to the corresponding binning error to obtain the binning corrected brightness temperature value.

[0096] S523. Based on all corrected brightness temperature values, update the brightness temperature values ​​of the forecast cloud map to be corrected to obtain graded corrected cloud maps.

[0097] In this embodiment, the method can replace the original predicted brightness temperature value of each pixel in the predicted cloud map to be corrected with the calculated corrected brightness temperature value, complete the brightness temperature update of the whole map, and form a graded corrected cloud map so that the predicted brightness temperature is closer to the actual measurement.

[0098] S530. Based on the clear sky background brightness temperature field and the third cloud area identification results, the graded correction cloud map is corrected for clear sky conditions to obtain the target forecast cloud map.

[0099] In this embodiment, clear sky correction only corrects the clear sky pixels in the third cloud area identification result. Its core purpose is to reduce the error of the target forecast cloud map in the clear sky area.

[0100] In this embodiment, the method can achieve precise correction of the clear sky area by comparing it with the brightness temperature field of the clear sky background.

[0101] To reduce the error of the target forecast cloud image in the clear sky area, step S530 may further include: S531. Based on the third cloud region identification results, determine multiple clear-sky region pixels in the forecast cloud image to be corrected, as well as multiple corrected brightness temperature values ​​corresponding to the multiple clear-sky region pixels.

[0102] In this embodiment, the third cloud area identification result may include two parts: clear sky area pixels and cloud area pixels. As mentioned earlier, the identification threshold used in the third cloud area identification result is not a strict threshold, so the third cloud area identification result may only include clear sky area pixels and cloud area pixels. Correspondingly, the second cloud area identification result should include cloud areas, clear sky areas, and other areas.

[0103] S532. Obtain multiple clear sky background field brightness temperature values ​​that correspond one-to-one with multiple clear sky area pixels based on the clear sky background brightness temperature field.

[0104] In this embodiment, the brightness temperature value of the clear sky background field corresponds one-to-one with the grid cell position of the graded correction cloud map. Each cell to be corrected can be matched with a corresponding clear sky background brightness temperature value.

[0105] In this embodiment, the method only needs to obtain the brightness temperature value of the clear sky Beijing field corresponding to the pixel in the clear sky area.

[0106] S533. Calculate the absolute difference between multiple corrected brightness temperature values ​​and multiple clear sky background field brightness temperature values ​​to obtain multiple absolute differences corresponding to multiple clear sky area pixels.

[0107] In this embodiment, the absolute difference is calculated using the following formula: ; Among them, S 分档订正 The brightness temperature value after grading correction, O b This represents the brightness temperature value of the corresponding clear sky background field.

[0108] S534. In the graded correction cloud map, the clear-sky area pixels corresponding to the absolute difference value that is less than or equal to the preset difference value are identified as pixels to be processed.

[0109] In this embodiment, the preset difference value can be determined to be 5K through experimental statistical analysis. Wherein, when the absolute difference value e clr If the pixel is ≤5K, it is determined to be a pixel to be processed.

[0110] S535. Replace the corrected brightness temperature value of the pixel to be processed in the graded correction cloud map with the corresponding clear sky background field brightness temperature value to obtain the clear sky corrected target forecast cloud map.

[0111] In this embodiment, the brightness temperature value after clear sky correction is determined according to the following formula: ; This method can be based on this to perform brightness temperature replacement only on the pixels to be processed in the clear sky area, thereby preserving the cloud classification correction results.

[0112] In the above embodiments, the method can perform differentiated correction of cloud areas by graded intervals, and perform clear sky correction of non-cloud areas by combining clear sky background brightness temperature field, thereby achieving hierarchical and refined intensity correction and further improving the accuracy of target forecast cloud images.

[0113] To verify the effectiveness of the forecast cloud image intensity correction method proposed in this application, this embodiment uses the FY-4 long-wave infrared channel (10.3~11.3μm) forecast cloud image as the research object, and conducts a targeted experiment from 4:00 to 8:00 (UTC) on June 8, 2023. The target time division of the experiment strictly matches the time dimension setting of this method: 4:00 to 6:00 corresponds to the historical reference target time (-2~0h) in the algorithm, used for error statistical analysis; 6:00 to 8:00 corresponds to the forecast correction target time (0~2h) in the algorithm, used to verify the correction effect. Details are as follows: (1) Verification of the effect of graded correction Based on historical data from 4:00 AM to 6:00 AM on June 8, 2023, brightness temperature errors were statistically analyzed in different tiers (the results of the brightness temperature error tiered statistics can be found in [link to relevant documentation]). Figure 2 ).

[0114] Statistics show that the original forecast brightness temperature distribution is generally biased towards the high-temperature area, and the error differences between different regions are significant: the average deviation in the low-value area (i.e., the cloud area) is about 20K, and the average deviation in the high-value area (i.e., the clear sky area) is about 10K. The forecast error of brightness temperature in the cloud area is significantly greater than that in the clear sky area, which confirms the necessity of carrying out targeted intensity correction.

[0115] After using the method described in this application to categorize and correct the forecast cloud images for 6:00 and 8:00 on June 8, 2023 (the correction results can be found in [reference]), the cloud images were corrected by category. Figure 3 ).in, Figure 3The results of the two groups of graded corrections are shown: Group 1 (the three images at the top): the brightness temperature distribution of the observation map (left), the forecast map (middle), and the graded correction forecast map (right) at 6:00; Group 2 (the three images at the bottom): the brightness temperature distribution of the observation map (left), the forecast map (middle), and the graded correction forecast map (right) at 8:00.

[0116] By comparing the original predicted brightness temperature with the measured brightness temperature distribution, we can know that: The brightness temperature data after tiered correction shows a significant improvement in its alignment with measured values, especially in key areas such as fronts and typhoon cloud systems, where the forecast brightness temperature values ​​are effectively reduced. This demonstrates that the method can accurately correct the high-temperature offset in the original forecast, fully proving the significant optimization effect of tiered correction on cloud areas (low-temperature areas).

[0117] (2) Verification of the correction effect under clear skies Based on the tiered correction, further clear-sky corrections were performed on the cloud images at the aforementioned target times (results can be found in...). Figure 4 ).in, Figure 4 The image shows a comparison of the forecast brightness temperature distribution at 6:00 (left) and 8:00 (right) for clear skies.

[0118] By comparing the graded correction results, it can be seen that the clear-sky correction has a relatively mild impact on the overall forecast cloud imagery, but statistical data shows that: The forecast accuracy for clear-sky areas has been specifically improved, with slight improvements in the overall correlation coefficient and average error index, thus enabling precise optimization for high-temperature areas (clear-sky areas). It should also be noted that although the root mean square error has slightly worsened after the clear-sky correction, it has not affected the positive improvement in the overall correction effect.

[0119] (3) Statistical comparison of overall effects Figure 5 The following is a histogram showing the brightness temperature distribution at 6:00 AM and 8:00 AM on June 8, 2023. Figure 5 The overall correction effect of this method is presented intuitively: After the grading correction, the overall distribution of the forecast brightness temperature is closer to the distribution of the measured brightness temperature, and the error in the low temperature area (cloud area) is significantly improved. The clear sky correction focuses on high-temperature areas (clear sky areas) to achieve targeted optimization.

[0120] By combining these two approaches, a refined intensity correction scheme for near-term (0-2h) forecast cloud images across the entire region is formed. This effectively reduces the systematic error of the original forecast and significantly improves the fit between the forecast cloud images and the actual atmospheric conditions.

[0121] like Figure 6 As shown, some embodiments of this application provide a structural schematic diagram of an intensity correction device for stationary infrared forecast cloud images. It should be understood that this device is related to... Figure 1 The method executed in the middle corresponds to the steps involved in the aforementioned method. The specific functions and effects of the device can be found in the description above. To avoid repetition, detailed descriptions are omitted here.

[0122] The intensity correction device for stationary infrared forecast cloud images includes: The acquisition unit 610 is used to acquire, in a spatially one-to-one correspondence, the original forecast cloud image, the original measured cloud image, the original total cloud cover forecast image, the clear sky background brightness temperature field, the forecast cloud image to be corrected, and the future total cloud cover forecast image; wherein, the original forecast cloud image and the original total cloud cover forecast image are forecast data at the historical target time, the original measured cloud image is measured data at the historical target time, and the forecast cloud image to be corrected and the future total cloud cover forecast image are forecast data at the future target time; The first identification unit 620 is used to identify cloud areas in the original measured cloud map based on the brightness temperature field of the clear sky background, and obtain the first cloud area identification result. The second identification unit 630 is used to identify the cloud areas of the original forecast cloud map based on the first preset cloud cover threshold and the original total cloud cover forecast map, and obtain the second cloud area identification result. The third identification unit 640 is used to identify the cloud areas of the forecast cloud map to be corrected based on the second preset cloud cover threshold and the future total cloud cover forecast map, and obtain the third cloud area identification result. The intensity correction unit 650 is used to correct the intensity of the forecast cloud map to be corrected based on the first cloud area identification result, the second cloud area identification result, and the clear sky background brightness temperature field, so as to obtain the target forecast cloud map.

[0123] In some embodiments, the strength correction unit 650 includes: The model building subunit 651 is used to build a grading correction model based on the first cloud area identification result, the second cloud area identification result and the preset grading interval. The grading correction subunit 652 is used to perform grading correction on the forecast cloud map to be corrected according to the grading correction model, so as to obtain the grading correction cloud map. Clear sky correction subunit 653 is used to perform clear sky correction on the graded correction cloud map based on the clear sky background brightness temperature field and the third cloud area identification results to obtain the target forecast cloud map.

[0124] In some embodiments, the grading correction subunit 651 includes: The filtering module is used to filter out a set of pixel samples for tiered statistics from the original forecast cloud map based on the first cloud area identification result and the second cloud area identification result; the pixel samples in the set of pixel samples are simultaneously marked as cloud areas or clear sky areas in the first cloud area identification result and the second cloud area identification result. The first calculation module is used to calculate the grading error based on the pixel sample set and the preset grading interval, and obtain the grading error corresponding to each of the multiple gradings; wherein, the grading correction model includes multiple gradings and the grading error corresponding to each of the multiple gradings.

[0125] In some embodiments, the grading correction subunit 652 includes: The first determining module is used to determine multiple correction errors that correspond one-to-one with multiple pixels to be corrected in the forecast cloud image to be corrected, based on the tiered correction model. The second calculation module is also used to calculate multiple corrected brightness temperature values ​​that correspond one-to-one with all the multiple pixels to be corrected, given the predicted brightness temperature values ​​and correction errors that have a one-to-one correspondence among the multiple pixels to be corrected. The update module is used to update the brightness temperature values ​​of the forecast cloud map to be corrected based on all corrected brightness temperature values, so as to obtain the graded corrected cloud map.

[0126] In some embodiments, the first calculation module is specifically used to obtain the predicted brightness temperature values ​​of all pixel samples in the original predicted cloud image and the measured brightness temperature values ​​in the original measured cloud image; based on the predicted brightness temperature values ​​of all pixel samples, divide all pixel samples into multiple corresponding sub-sub ...

[0127] In some embodiments, the clear sky correction subunit 653 includes: The second determination module is used to determine multiple clear sky area pixels in the forecast cloud image to be corrected, as well as multiple corrected brightness temperature values ​​corresponding to the multiple clear sky area pixels, based on the third cloud area identification result. The acquisition module is used to acquire multiple clear sky background field brightness temperature values ​​that correspond one-to-one with multiple clear sky area pixels based on the clear sky background brightness temperature field. The third calculation module is used to calculate the absolute difference between multiple corrected brightness temperature values ​​and multiple clear sky background field brightness temperature values, and to obtain multiple absolute differences corresponding to multiple clear sky area pixels. The second determining module is used to determine the clear-sky area pixels corresponding to the absolute difference that is less than or equal to the preset difference as pixels to be processed in the graded correction cloud map; The replacement module is used to replace the corrected brightness temperature value of the pixel to be processed in the graded corrected cloud map with the corresponding clear sky background field brightness temperature value, so as to obtain the clear sky corrected target forecast cloud map.

[0128] In some embodiments, the first identification unit 620 includes: The first acquisition subunit 621 is used to acquire pixel information of all measured pixels in the original measured cloud image; wherein, each pixel information includes a pixel position and a measured brightness temperature value with a one-to-one correspondence. The first acquisition subunit 621 is also used to acquire the brightness temperature value of the clear sky background field of all grid cells in the clear sky background brightness temperature field; wherein, all clear sky background field brightness temperature values ​​correspond one-to-one with all cell positions; The calculation subunit 622 is used to perform one-to-one difference calculation based on all measured brightness temperature values ​​and all clear sky background field brightness temperature values ​​to obtain multiple brightness temperature difference values. The first marking subunit 623 is used to mark all measured pixels according to multiple brightness temperature difference values ​​and a preset cloud detection threshold to obtain measured pixels in the cloud area and measured pixels in the clear sky area. The first processing subunit 624 is used to perform binarization processing on the labeling results of the measured pixels in the cloud area and the measured pixels in the clear sky area to obtain the first cloud area recognition result.

[0129] In some embodiments, the second identification unit 630 includes: The second acquisition subunit 631 is used to acquire the total cloud cover forecast data of the original forecast cloud map based on the original total cloud cover forecast map. Subunit 632 is defined to determine the cloud cover value of all forecast pixels in the original forecast cloud map based on the total cloud cover forecast data. The second marking subunit 633 is used to mark all forecast pixels according to the cloud cover value of all forecast pixels and the first preset cloud cover threshold to obtain cloud area forecast pixels and clear sky area forecast pixels. The second processing subunit 634 is used to perform binarization processing on the labeling results of cloud area forecast pixels and clear sky area forecast pixels to obtain the second cloud area identification result.

[0130] like Figure 7 As shown, this application provides an electronic device 700, which includes a processor 701 and a memory 702. The processor 701 and the memory 702 are interconnected and communicate with each other through a communication bus 703 and / or other forms of connection mechanism (not shown). The memory 702 stores a computer program that can be executed by the processor 701. When the computing device is running, the processor 701 executes the computer program to perform the method in any of the aforementioned optional implementations.

[0131] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method in any of the aforementioned optional implementations.

[0132] The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0133] This application provides a computer program product, which includes a computer program that, when run by a processor, executes the method in any of the aforementioned optional implementations.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for intensity correction of stationary infrared forecast cloud images, characterized in that, include: Acquire spatially corresponding original forecast cloud images, original measured cloud images, original total cloud cover forecast images, clear sky background brightness temperature field, forecast cloud images to be corrected, and future total cloud cover forecast images; wherein, the original forecast cloud images and the original total cloud cover forecast images are forecast data at historical target times, the original measured cloud images are measured data at the historical target times, and the forecast cloud images to be corrected and the future total cloud cover forecast images are forecast data at future target times; Based on the clear sky background brightness temperature field, cloud areas are identified in the original measured cloud map to obtain the first cloud area identification result; The cloud regions in the original forecast cloud map are identified based on the first preset cloud cover threshold and the original total cloud cover forecast map to obtain the second cloud region identification result; The cloud regions of the forecast cloud map to be corrected are identified based on the second preset cloud cover threshold and the forecast map of total future cloud cover, and the third cloud region identification result is obtained. The intensity of the forecast cloud map to be corrected is corrected based on the first cloud area identification result, the second cloud area identification result, the third cloud area identification result, and the clear sky background brightness temperature field to obtain the target forecast cloud map.

2. The intensity correction method for stationary infrared forecast cloud images according to claim 1, characterized in that, The step of performing intensity correction on the forecast cloud image to be corrected based on the first cloud region identification result, the second cloud region identification result, the third cloud region identification result, and the clear sky background brightness temperature field to obtain the target forecast cloud image includes: A grading correction model is established based on the identification results of the first cloud area, the identification results of the second cloud area, and the preset grading interval; The predicted cloud map to be corrected is corrected according to the tiered correction model to obtain the tiered corrected cloud map. Based on the clear sky background brightness temperature field and the third cloud area identification result, the graded correction cloud map is corrected for clear sky conditions to obtain the target forecast cloud map.

3. The intensity correction method for stationary infrared forecast cloud images according to claim 2, characterized in that, The step of establishing a tiered correction model based on the first cloud region identification result, the second cloud region identification result, and a preset tiering interval includes: Based on the first cloud area identification result and the second cloud area identification result, a set of pixel samples for tiered statistics is selected from the original forecast cloud map; the pixel samples in the set of pixel samples are simultaneously marked as cloud areas or clear sky areas in the first cloud area identification result and the second cloud area identification result. The grading error is calculated based on the pixel sample set and the preset grading interval to obtain the grading error corresponding to each of the multiple gradings. The tiered correction model includes multiple tiers and the tiered errors corresponding to each of the multiple tiers.

4. The intensity correction method for stationary infrared forecast cloud images according to claim 2, characterized in that, The step of performing tiered correction on the forecast cloud map to be corrected according to the tiered correction model to obtain a tiered corrected cloud map includes: Based on the tiered correction model, multiple correction errors are determined that correspond one-to-one with the multiple pixels to be corrected in the forecast cloud image to be corrected; For the predicted brightness temperature values ​​and correction errors that have a one-to-one correspondence among the plurality of pixels to be corrected, calculate a plurality of corrected brightness temperature values ​​that correspond one-to-one with all the plurality of pixels to be corrected. The brightness temperature values ​​of the forecast cloud map to be corrected are updated based on all corrected brightness temperature values ​​to obtain graded corrected cloud maps.

5. The intensity correction method for stationary infrared forecast cloud images according to claim 3, characterized in that, The step of calculating the grading error based on the pixel sample set and the preset grading interval to obtain the grading error corresponding to each of the multiple grading intervals includes: Obtain the predicted brightness temperature values ​​of all pixel samples in the original predicted cloud image and the measured brightness temperature values ​​in the original measured cloud image; Based on the predicted brightness temperature values ​​of all pixel samples, all pixel samples are divided into multiple corresponding sub-sub ... Calculate the average error between the predicted brightness temperature value and the measured brightness temperature value of all pixel samples in each graded sample set to obtain the graded error corresponding to each of the multiple grades.

6. The intensity correction method for stationary infrared forecast cloud images according to claim 2, characterized in that, The step of performing clear-sky correction on the graded corrected cloud map based on the clear-sky background brightness temperature field and the third cloud region identification result to obtain the target forecast cloud map includes: Based on the third cloud region identification result, determine multiple clear sky region pixels in the forecast cloud image to be corrected, and multiple corrected brightness temperature values ​​corresponding to the multiple clear sky region pixels. Based on the clear sky background brightness temperature field, obtain multiple clear sky background field brightness temperature values ​​that correspond one-to-one with the multiple clear sky area pixels; The absolute difference between the plurality of corrected brightness temperature values ​​and the plurality of clear sky background field brightness temperature values ​​is calculated to obtain the plurality of absolute differences corresponding to the plurality of clear sky area pixels. In the graded correction cloud map, the clear-sky area pixels corresponding to the absolute difference value that is less than or equal to the preset difference value are determined as pixels to be processed; The corrected brightness temperature value of the pixel to be processed in the graded corrected cloud map is replaced with the corresponding clear sky background field brightness temperature value to obtain the clear sky corrected target forecast cloud map.

7. The intensity correction method for stationary infrared forecast cloud images according to claim 1, characterized in that, The step of identifying cloud regions based on the clear sky background brightness temperature field of the original measured cloud image to obtain the first cloud region identification result includes: Obtain pixel information of all measured pixels in the original measured cloud image; wherein, each pixel information includes a pixel position and a measured brightness temperature value with a one-to-one correspondence. Obtain the brightness temperature value of the clear sky background field for all grid cells in the clear sky background brightness temperature field; wherein, each clear sky background field brightness temperature value corresponds one-to-one with each cell position; Multiple brightness temperature difference values ​​are obtained by calculating the difference between all measured brightness temperature values ​​and all clear sky background field brightness temperature values; All measured pixels are marked according to the multiple brightness temperature difference values ​​and the preset cloud detection threshold to obtain measured pixels in the cloud area and measured pixels in the clear sky area. The labeling results of the measured pixels in the cloud area and the measured pixels in the clear sky area are binarized to obtain the first cloud area identification result.

8. The intensity correction method for stationary infrared forecast cloud images according to claim 1, characterized in that, The step of identifying cloud regions in the original forecast cloud map based on a first preset cloud cover threshold and the original total cloud cover forecast map to obtain a second cloud region identification result includes: Obtain the total cloud cover forecast data of the original forecast cloud map based on the original total cloud cover forecast map; Based on the total cloud cover forecast data, determine the cloud cover value of all forecast pixels in the original forecast cloud map; All forecast pixels are labeled based on the cloud cover value and the first preset cloud cover threshold to obtain cloud area forecast pixels and clear sky area forecast pixels. The labeling results of the cloud area forecast pixels and the clear sky area forecast pixels are binarized to obtain the second cloud area identification result.

9. An intensity correction device for stationary infrared forecast cloud images, characterized in that, The intensity correction device for stationary infrared forecast cloud images includes: The acquisition unit is used to acquire, in a spatially one-to-one correspondence, the original forecast cloud image, the original measured cloud image, the original total cloud cover forecast image, the clear sky background brightness temperature field, the forecast cloud image to be corrected, and the future total cloud cover forecast image; wherein, the original forecast cloud image and the original total cloud cover forecast image are forecast data at the historical target time, the original measured cloud image is measured data at the historical target time, and the forecast cloud image to be corrected and the future total cloud cover forecast image are forecast data at the future target time; The first identification unit is used to identify cloud areas in the original measured cloud map based on the clear sky background brightness temperature field, and obtain the first cloud area identification result. The second identification unit is used to identify the cloud areas of the original forecast cloud map based on the first preset cloud cover threshold and the original total cloud cover forecast map, and obtain the second cloud area identification result. The third identification unit is used to identify the cloud areas of the forecast cloud map to be corrected based on the second preset cloud cover threshold and the future total cloud cover forecast map, and obtain the third cloud area identification result. An intensity correction unit is used to correct the intensity of the forecast cloud map to be corrected based on the first cloud area identification result, the second cloud area identification result, the third cloud area identification result, and the clear sky background brightness temperature field, so as to obtain the target forecast cloud map.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor running the computer program to cause the electronic device to perform the intensity correction method for stationary infrared forecast cloud images as described in any one of claims 1 to 7.

Images