High-precision solar radiation estimation method based on multi-source satellite observation

Abstract

The invention relates to the technical field of meteorological satellite detection, and provides a multi-source satellite observation-based high-precision solar radiation estimation method, which comprises the following steps of: accurately inverting aerosol and cloud extinction coefficient profiles by using a multi-source satellite, ground-based observation data, a graph neural network model and a Transform model; acquiring an atmospheric temperature and humidity profile data set matched with satellite observation; the ground solar radiation is accurately calculated in combination with a three-dimensional radiation transmission model MCBRaT3D; finally, an integrated learning method is adopted to carry out error correction on a ground solar radiation calculation result, a high-precision and physically interpretable solar radiation product is generated, and therefore the problems that in a traditional method, precision is insufficient, data fusion is difficult, and model calculation efficiency is low are solved. And a new technical means and method are provided for the fields of solar energy resource evaluation, climate research, environment monitoring and the like.

Description

Technical Field

[0001] The present invention relates to the field of meteorological satellite detection technology, and in particular to a high-precision solar radiation estimation method based on multi-source satellite observation. Background Art

[0002] Solar radiation is a core driver of Earth's climate system and energy balance, and is crucial for solar resource assessment, climate change research, and environmental monitoring. Accurately acquiring ground-based solar radiation information is crucial for improving solar power generation, agricultural production, and climate models. However, due to the limited spatial distribution of ground-based observatories, traditional solar radiation measurement methods cannot meet the requirements of large-scale, high-resolution measurements.

[0003] Currently, methods for estimating solar radiation based on satellite remote sensing data have become a hot topic of research. These methods primarily utilize data from a single satellite, employing empirical or physical models to estimate ground-level solar radiation. However, single-satellite observations suffer from incomplete data, low spatial resolution, and insufficient temporal coverage, limiting the accuracy of these estimates. Furthermore, traditional methods suffer from inaccurate processing of atmospheric factors such as aerosols and clouds, making them inadequate for high-precision applications.

[0004] Aerosols and clouds are key factors influencing solar radiation transfer, with their vertical distribution and optical properties significantly influencing the radiation transfer process. However, existing solar radiation estimation methods typically use simplified atmospheric models that fail to fully account for the three-dimensional distribution and dynamic changes of aerosols and clouds, resulting in biased estimates. Furthermore, traditional radiation transfer models suffer from low computational efficiency under complex atmospheric conditions and are unable to meet the needs of large-scale, high-resolution radiation estimation.

[0005] With the advancement of remote sensing technology and big data analysis methods, the use of multi-source satellite observation data and advanced deep learning models is expected to improve the accuracy of solar radiation estimates. Multi-source satellite data can provide more comprehensive and detailed atmospheric information, including aerosol extinction coefficient profiles, cloud extinction coefficient profiles, and atmospheric temperature and humidity profiles. However, how to effectively integrate multi-source data to construct high-precision radiative transfer models remains a pressing technical challenge. Summary of the Invention

[0006] The present invention aims to provide a high-precision solar radiation estimation method based on multi-source satellite observations to solve the problems of insufficient accuracy, difficulty in data fusion and low model calculation efficiency existing in traditional methods.

[0007] In a first aspect, the present invention provides a high-precision solar radiation estimation method based on multi-source satellite observations, comprising: S1, collect multi-source satellite and ground-based observation data, and generate a matched data set through time and space matching technology; the multi-source satellite and ground-based observation data include aerosol extinction coefficient profile data, cloud extinction coefficient profile data, and multi-channel meteorological satellite observation data; S2, use a graph neural network model to train the multi-channel meteorological satellite observation data in the data set, retrieve aerosol extinction coefficient profiles, and obtain high-precision aerosol extinction coefficient profile data; S3, use a Transformer model to train the multi-channel meteorological satellite observation data in the data set, retrieve cloud extinction coefficient profiles, and obtain high-precision cloud extinction coefficient profile data; S4, collect atmospheric temperature and humidity profile data provided by a numerical weather prediction model, and generate an atmospheric temperature and humidity profile data set matched with satellite observation; S5, spatiotemporally fuse the aerosol extinction coefficient profile data obtained in step S2, the cloud extinction coefficient profile data obtained in step S3, and the atmospheric temperature and humidity profile data set obtained in step S4 to form a three-dimensional grid data set; S6, input the three-dimensional grid data set into a three-dimensional radiative transfer model MCBRaT3D to calculate ground solar radiation; S7, compare and analyze the ground solar radiation calculated in step S6 with ground-based measured solar radiation data, use an ensemble learning method to construct an error correction model, and perform error correction.

[0008] In some embodiments, in step S1, quality control is needed for the collected aerosol extinction coefficient profile data and cloud extinction coefficient profile data, and a statistical filtering and / or anomaly detection algorithm is used to remove outliers.

[0009] In some embodiments, in step S1, time and space matching uses a nearest neighbor interpolation method or a bilinear interpolation method, and takes the longitude and latitude of the meteorological satellite and the observation time as the reference to match the corresponding aerosol extinction coefficient profile data and cloud extinction coefficient profile data to generate a matched data set.

[0010] In some embodiments, in step S2, the relationship between the aerosol extinction coefficient and the multi-channel meteorological satellite observation data in the data set is represented as:

[0011] wherein, represents a graph neural network model, which captures spatial neighborhood information by constructing a graph structure of multi-channel meteorological satellite observation data, and learns the distribution characteristics of aerosol extinction coefficient.

[0012] ​In some embodiments, in step S3, the cloud extinction coefficient The relationship between the multi-channel meteorological satellite observation data in the data set is expressed as:

[0013] wherein, represents a Transformer model, which captures the spatial and temporal characteristics of the cloud extinction coefficient profile by constructing a transmission structure of satellite observation data and using a self-attention mechanism.

[0014] In some embodiments, in step S4, the atmospheric temperature and humidity profile data are time-interpolated and space-interpolated to generate an atmospheric temperature and humidity profile data set matching the satellite observation.

[0015] In some embodiments, the space interpolation adopts a bilinear interpolation method to calculate the atmospheric temperature and humidity at the target point according to the data and weights of the adjacent grid points, and the formula is as follows:

[0016] wherein, is the temperature of the adjacent grid point, is the weight inversely proportional to the distance.

[0017] In some embodiments, the time interpolation adopts a linear interpolation method to calculate the atmospheric temperature and humidity at the target time according to the data at the previous and subsequent time points:

[0018] wherein, and are the temperatures at the adjacent time points, and are the times corresponding to the adjacent time points.

[0019] In some embodiments, in step S6, the three-dimensional radiative transfer model MCBRaT3D considers the scattering, absorption, and multiple scattering processes of aerosols and clouds; the input parameters further include three-dimensional aerosol, cloud, and atmospheric temperature and humidity profiles, satellite observation angles, solar elevation angles, instrument response parameters, and surface parameters.

[0020] In some embodiments, in step S7, a gradient boosting decision tree and a random forest model are jointly used to construct an error correction model; and the error correction formula is expressed as:

[0021] wherein, is the corrected solar radiation value, is the ground solar radiation calculated in step S6, The error correction value predicted by the error correction model is a weighted combination of the error prediction results of the gradient boosting decision tree and the random forest model:

[0022] wherein, is the error prediction result of the gradient boosting decision tree, is the error prediction result of the random forest model, is the weight of the gradient boosting decision tree, is the weight of the random forest model, satisfying =1;Model training uses AERONET observation network solar radiation data as labels to determine the best model parameters and weights.

[0023] In a second aspect, the present application provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the at least one processor executes the method by executing the instructions stored in the memory.

[0024] In a third aspect, the present application provides a computer readable storage medium for storing instructions, when the instructions are executed, the method is realized.

[0025] In a fourth aspect, the present application provides a computer program product, when the computer program product is invoked by a computer, the computer executes the above-mentioned method.

[0026] In summary, due to the adoption of the above technical solutions, the present application has the following advantages: The present application can accurately retrieve the extinction coefficient profile of aerosols and clouds using multi-source satellite and ground-based observation data, without the need for other auxiliary parameters, and directly realizes high-precision estimation of solar radiation. In the process of establishing the estimation model, the present application considers the atmospheric radiation transfer model, fully considers the physical processes of atmospheric absorption, scattering, etc., and constructs a physically interpretable model. In addition, the present application combines atmospheric temperature and humidity environmental factors, aerosol and cloud optical properties in the estimation process, and uses advanced deep learning models (such as graph neural network models and Transformer models) and three-dimensional radiation transfer model MCBRaT3D, and adopts an ensemble learning method for error correction. Therefore, the obtained solar radiation product has high precision and physical interpretability, and can be more accurately applied to the fields of solar energy resource assessment, climate research and environmental monitoring, etc. BRIEF DESCRIPTION OF DRAWINGS ​

[0027] Figure 1 A flow chart of a high-precision solar radiation estimation method based on multi-source satellite observations provided by an embodiment of the present application.

[0028] Figure 2 A flow chart of a solar radiation data calculation method of a three-dimensional radiation transfer model MCBRaT3D in an embodiment of the present application.

[0029] Figure 3 A structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0030] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the following will be a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0031] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative labor are within the scope of protection of the present application.

[0032] The following will be a detailed description of the specific embodiments of the present application in conjunction with actual applications, and the ground-based solar radiation data of the FY-4A satellite, the CALIPSO spaceborne lidar, the CloudSat spaceborne cloud radar, the GFS-GRAPES regional numerical prediction model of China, and the AERONET observation network will be illustrated. As shown in FIG. 1, a high-precision solar radiation estimation method based on multi-source satellite observations provided by an embodiment of the present application includes the following steps: Figure 1 S1, collecting multi-source satellite and ground-based observation data, and generating a matched data set through time and space matching technology; the multi-source satellite and ground-based observation data include aerosol extinction coefficient profile data, cloud extinction coefficient profile data, and multi-channel meteorological satellite observation data.

[0033] In this embodiment, the aerosol extinction coefficient profile data is obtained using the CALIPSO spaceborne lidar, the cloud extinction coefficient profile data is obtained using the CloudSat spaceborne cloud radar, and the multi-channel meteorological satellite observation data is visible and infrared channel data observed by the FY-4A satellite. These data can be used to more accurately retrieve aerosol and cloud characteristics in the atmosphere.​

[0034] In this embodiment, quality control needs to be performed on the collected aerosol extinction coefficient profile data and cloud extinction coefficient profile data, and outliers need to be removed. Statistical filtering and / or outlier detection algorithms can be used to remove outliers. For example, a physically reasonable threshold range is set, and data exceeding the threshold is considered as an outlier and removed; or based on statistical methods, such as exceeding three standard deviations of the mean, which is considered as an outlier and removed.

[0035] In this embodiment, the nearest neighbor interpolation method or the bilinear interpolation method is used for time and space matching. The aerosol extinction coefficient profile data and the cloud extinction coefficient profile data corresponding to the geodetic coordinates and observation time of the stationary meteorological satellite (FY-4A satellite) are matched to generate a matched data set.

[0036] S2, using a graph neural network (GNN) model, training the multi-channel meteorological satellite observation data in the data set, inverting the aerosol extinction coefficient profile, and obtaining high-precision aerosol extinction coefficient profile data.

[0037] In this embodiment, the aerosol extinction coefficient and the multi-channel meteorological satellite observation data in the data set The relationship can be expressed as:

[0038] Among them, The graph neural network model is represented by constructing a graph structure of the multi-channel meteorological satellite observation data, capturing spatial neighborhood information, and learning the distribution characteristics of the aerosol extinction coefficient.

[0039] S3, using a Transformer model, training the multi-channel meteorological satellite observation data in the data set, inverting the cloud extinction coefficient profile, and obtaining high-precision cloud extinction coefficient profile data.

[0040] In this embodiment, the cloud extinction coefficient and the multi-channel meteorological satellite observation data in the data set The relationship can be expressed as:

[0041] Among them, The Transformer model is represented by constructing a transfer structure of the satellite observation data, using a self-attention mechanism, and capturing the spatiotemporal characteristics of the cloud extinction coefficient profile.

[0042] S4, collect the atmospheric temperature and humidity profile data provided by the GFS-GRAPES regional numerical weather prediction model, and perform time interpolation and space interpolation on the atmospheric temperature and humidity profile data according to the longitude and latitude of the satellite observation pixel and the observation time, to ensure the spatial continuity and time continuity of the data, and generate a set of atmospheric temperature and humidity profile data matched with the satellite observation.

[0043] In this embodiment, the space interpolation adopts a bilinear interpolation method, and the atmospheric temperature and humidity of the target point are calculated according to the data and weights of the adjacent grid points, and the formula is as follows:

[0044] wherein, is the temperature of the adjacent grid point, is the weight inversely proportional to the distance.

[0045] In this embodiment, the time interpolation adopts a linear interpolation method, and the atmospheric temperature and humidity of the target time are calculated according to the data of the previous and subsequent time points:

[0046] wherein, and are the temperatures of the adjacent time points, and are the times corresponding to the adjacent time points.

[0047] S5, spatio-temporally fuse the aerosol extinction coefficient profile data obtained in step S2, the cloud extinction coefficient profile data obtained in step S3, and the set of atmospheric temperature and humidity profile data obtained in step S4 to form a unified three-dimensional grid data set.

[0048] S6, input the three-dimensional grid data set into the three-dimensional radiation transfer model MCBRaT3D to calculate the ground solar radiation, including direct radiation, scattered radiation and total radiation.

[0049] In this embodiment, as shown in Figure 2 , the input parameters of the three-dimensional radiation transfer model MCBRaT3D further include three-dimensional aerosol, cloud and atmospheric temperature and humidity profiles, satellite observation angles (such as the observation angles of FY-4A), solar elevation angles, instrument response parameters (such as the spectral response function, field of view angle, etc. of FY-4A satellite), and ground surface parameters (such as ground surface albedo, terrain information, etc.).

[0050] The model first randomly emits photons according to the solar elevation angle and observation geometry, and then gradually tracks the propagation of the photons in the three-dimensional aerosol, cloud and temperature and humidity fields; each time a particle is encountered, the absorption or scattering is determined according to the scattering albedo, and the direction is updated according to the phase function until the photon reaches the ground or leaves the atmosphere. The photons reaching the ground are divided into direct and scattered shares, and if they encounter the ground, they are reflected according to the albedo. Finally, the direct radiation, the scattered radiation and the sum thereof (total radiation) are statistically summarized in each grid cell, and are weighted according to the spectral response of the satellite channel, so as to output a physically interpretable and controllable ground solar irradiance field for observation systems such as FY-4A.

[0051] The three-dimensional radiation transfer model MCBRaT3D considers the scattering, absorption and multiple scattering processes of aerosols and clouds, and can accurately simulate the ground solar radiation, thereby constructing a physically interpretable model.

[0052] S7, the ground solar radiation calculated in step S6 is compared and analyzed with the ground-based measured solar radiation data, an error correction model is constructed by using an ensemble learning method, error correction is performed, and finally a high-precision and physically interpretable solar radiation product is generated.

[0053] In this embodiment, the ground-based measured solar radiation data is selected from the solar radiation data of the AERONET observation network. By using an ensemble learning method, a gradient boosting decision tree (GBDT) and a random forest model are jointly used to construct an error correction model. The error correction formula is represented as:

[0054] wherein, is the corrected solar radiation value, is the ground solar radiation calculated in step S6, is an error correction value predicted by the error correction model, which is composed of the error prediction results of the gradient boosting decision tree and the random forest model by weighted combination:

[0055] wherein, is the error prediction result of the gradient boosting decision tree, is the error prediction result of the random forest model, is the weight of the gradient boosting decision tree, is the weight of the random forest model, satisfying + =1. The model training uses the solar radiation data of the AERONET observation network as labels to determine the best model parameters and weights, and finally generates a high-precision and physically interpretable solar radiation product.

[0056] Based on the same technical concept, the embodiment of the present application also provides an electronic device which can implement the high-precision solar radiation estimation method based on multi-source satellite observation provided by the above-mentioned embodiment of the present application. In an embodiment, the electronic device can be a server, a terminal device or other electronic device. As shown in Figure 3 , the electronic device can include: at least one processor, and a memory connected with the at least one processor, and the specific connection medium between the processor and the memory in the embodiment of the present application is not limited, Figure 3 for example, the connection between the processor and the memory through the bus is taken as an example. The bus is represented by a thick line in Figure 3 , and the connection mode between other components is only schematically illustrated and is not limited. The bus can be divided into an address bus, a data bus, a control bus, etc., for the convenience of representation, Figure 3 only one thick line is used in the embodiment of the present application, but it does not mean that there is only one bus or only one type of bus. Alternatively, the processor can also be called a controller, and the name is not limited.

[0057] In the embodiment of the present application, the memory stores instructions executable by the at least one processor, and the at least one processor can execute the high-precision solar radiation estimation method based on multi-source satellite observation discussed above by executing the instructions stored in the memory. The processor can implement Figure 3 the functions of various modules in the device shown in

[0058] Among them, the processor is the control center of the device, and can connect all parts of the control device through various interfaces and lines, and through running or executing the instructions stored in the memory and calling the data stored in the memory, the device can process data and various functions, thereby monitoring the whole device.

[0059] In an optional design, the processor can include one or more processing units, and the processor can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, the user interface and the application program, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor. In some embodiments, the processor and the memory can be implemented on the same chip, and in some embodiments, they can also be implemented on separate chips respectively.

[0060] The processor can be a general purpose processor, such as a CPU, a digital signal processor, an application specific integrated circuit, a field programmable gate array, or other programmable logic device, discrete gate or transistor logic, discrete hardware components, can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general purpose processor can be a microprocessor or any conventional processor, etc. The steps of a high-precision solar radiation estimation method based on multi-source satellite observation disclosed in combination with the embodiments of the present application can be directly embodied as hardware processor execution, or executed by a combination of hardware and software modules in the processor.

[0061] The memory is a non-volatile computer readable storage medium, which can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card type memory, random access memory (RAM), static random access memory (SRAM), programmable read only memory (PROM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), magnetic memory, magnetic disk, optical disk, etc. The memory is any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but is not limited to this. The memory in the embodiments of the present application can also be a circuit or any other device capable of realizing the storage function, used for storing program instructions and / or data.

[0062] By designing and programming the processor, the code corresponding to the high-precision solar radiation estimation method based on multi-source satellite observation introduced in the foregoing embodiments can be fixed in the chip, so that the chip can execute the steps of the method of the foregoing embodiments when running. How to design and program the processor is a technology known to those skilled in the art, which will not be described here.

[0063] Based on the same inventive concept, the embodiments of the present application also provide a storage medium storing computer instructions, when the computer instructions run on a computer, the computer executes the foregoing high-precision solar radiation estimation method based on multi-source satellite observation.

[0064] In some alternative embodiments, various aspects of the method for high-precision solar radiation estimation based on multi-source satellite observations can also be implemented as a program product in the form of a computer program or a plurality of computer programs including a plurality of program codes to be executed by a computer or a plurality of computers to provide the device with the functionality described above when the program product is run on the device.

[0065] It should be noted that, although several units or sub-units of the device are mentioned in the foregoing detailed description, such a division is merely exemplary and not mandatory. Indeed, according to an embodiment of the application, the features and functionalities of two or more units described above can be embodied in one unit. Conversely, the features and functionalities of one unit described above can be further divided into a plurality of units embodied. Moreover, although the operations of the method of the application are described in a particular order in the figures, this does not imply that the operations must be performed in that particular order, or that all of the illustrated operations must be performed to achieve the desired results. Additionally or alternatively, certain steps can be omitted, combined, performed in a different order, and / or split into multiple steps.

[0066] Those skilled in the art will appreciate that embodiments of the present application can be provided as methods, systems, or computer program products. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can be embodied in the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROMs, optical storage media, etc.) having computer usable program code embodied therein.

[0067] The present application is described with reference to the flowcharts and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions, which are executed via the processor of the computer or other programmable data processing apparatus, generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The means for carrying out the functions specified in one or more flows and / or blocks in the flowcharts and / or block diagrams. Figure 1 The means for carrying out the functions specified in one or more flows and / or blocks in the flowcharts and / or block diagrams.

[0068] Program code to implement the application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. The embodiments of the application covered herein are to be

[0069] In situations in which the remote computing device utilizes a network, the remote computing device can be connected to the user's computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN). Such networks are well known to those having ordinary skill in the art and therefore will not be discussed herein in more detail.

[0070] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0071] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0072] The above description is intended to be illustrative and not restrictive. Many other changes and modifications can occur to those skilled in the art once advised of the principles of the application. Any such changes and modifications including without limitation those relating to the methods of operation, methods of implementation, associated computing device details, and the like are intended to be included within the scope of the application as set forth in the claims below. Therefore, the true scope of the application is not to be limited to the description described above, but is to be accorded the full scope permissible pursuant to the patent laws.

Claims

1. A high-precision solar radiation estimation method based on multi-source satellite observations, characterized in that: include: S1. Collect multi-source satellite and ground-based observation data and generate matching datasets through temporal and spatial matching techniques. The multi-source satellite and ground-based observation data include aerosol extinction coefficient profile data, cloud extinction coefficient profile data, and multi-channel meteorological satellite observation data; S2. Using a graph neural network model, train the multi-channel meteorological satellite observation data in the dataset to invert the aerosol extinction coefficient profile and obtain high-precision aerosol extinction coefficient profile data; S3. Using the Transformer model, training the multi-channel meteorological satellite observation data in the dataset, inverting the cloud extinction coefficient profile, and obtaining high-precision cloud extinction coefficient profile data; S4. Collect atmospheric temperature and humidity profile data provided by numerical weather forecast models and generate atmospheric temperature and humidity profile datasets that match satellite observations; S5, performing spatiotemporal fusion on the aerosol extinction coefficient profile data obtained in step S2, the cloud extinction coefficient profile data obtained in step S3, and the atmospheric temperature and humidity profile data set obtained in step S4 to form a three-dimensional grid data set; S6. Inputting the three-dimensional grid data set into the three-dimensional radiation transfer model MCBRaT3D to calculate ground solar radiation; S7. Compare and analyze the ground solar radiation calculated in step S6 with the solar radiation data measured on the ground, and use an integrated learning method to build an error correction model to perform error correction.

2. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1 is characterized in that: In step S1, it is necessary to perform quality control on the collected aerosol extinction coefficient profile data and cloud extinction coefficient profile data, and use statistical filtering and / or anomaly detection algorithms to remove outliers.

3. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1 is characterized in that: In step S1, the time and space matching adopts the nearest neighbor interpolation method or the bilinear interpolation method, and the corresponding aerosol extinction coefficient profile data and cloud extinction coefficient profile data are matched based on the latitude and longitude and observation time of the meteorological satellite to generate a matching data set.

4. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1 is characterized in that: In step S2, the aerosol extinction coefficient Multi-channel meteorological satellite observation data in the dataset The relationship is expressed as: in, Representing a graph neural network model, it captures spatial neighborhood information by constructing a graph structure of multi-channel meteorological satellite observation data and learns the distribution characteristics of aerosol extinction coefficient.

5. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1 is characterized in that: In step S3, the cloud extinction coefficient Multi-channel meteorological satellite observation data in the dataset The relationship is expressed as: in, Represents the Transformer model, which captures the spatiotemporal characteristics of cloud extinction coefficient profiles by constructing a transmission structure for satellite observation data and using the self-attention mechanism.

6. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1 is characterized in that: In step S4, temporal interpolation and spatial interpolation are performed on the atmospheric temperature and humidity profile data to generate an atmospheric temperature and humidity profile dataset that matches the satellite observations.

7. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 6 is characterized in that: The spatial interpolation adopts bilinear interpolation method to calculate the atmospheric temperature and humidity of the target point based on the data and weights of the neighboring grid points. The formula is as follows: in, is the temperature of the neighboring grid point, is a weight inversely proportional to the distance.

8. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 6 is characterized in that: The time interpolation adopts linear interpolation method to calculate the atmospheric temperature and humidity at the target time based on the data of the previous and next time points: in, and is the temperature at adjacent time points, and is the time corresponding to adjacent time points.

9. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1, characterized in that: In step S6, the three-dimensional radiation transfer model MCBRaT3D takes into account the scattering, absorption and multiple scattering processes of aerosols and clouds; the input parameters also include three-dimensional aerosols, clouds and atmospheric temperature and humidity profiles, satellite observation angles, solar altitude angles, instrument response parameters, and surface parameters.

10. The high-precision solar radiation estimation method based on multi-source satellite observation according to claim 1, characterized in that: In step S7, the error correction model is constructed by combining the gradient boosting decision tree and the random forest model. The error correction formula is expressed as: in, is the corrected solar radiation value, is the ground solar radiation calculated in step S6, is the error correction value predicted by the error correction model, which is a weighted combination of the error prediction results of the gradient boosting decision tree and the random forest model: in, is the error prediction result of the gradient boosting decision tree, is the error prediction result of the random forest model, is the weight of the gradient boosting decision tree, is the weight of the random forest model, satisfying + =1; Model training uses solar radiation data from the AERONET observation network as labels to determine the optimal model parameters and weights.

Images