Atmospheric radiation prediction method and device

Abstract

The embodiments of this specification provide an atmospheric radiation prediction method and device, wherein the atmospheric radiation prediction method is applied to a radiation prediction model, the radiation prediction model is obtained by training based on atmospheric data of a first number of layers, the radiation prediction model includes a forward network and a backward network, and the method includes: obtaining atmospheric data of a second number of layers, sequentially inputting atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, and inputting atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; determining target radiation information of the corresponding atmosphere according to the first radiation information and the second radiation information of the atmospheric data of each layer. By using the atmospheric data of the second number of layers, the target radiation information of the second number of layers is obtained, which can be used for different vertical layers of numerical weather forecasts and has stronger generalization ability.

Description

Technical Field

[0001] The embodiments of the present specification relate to the field of computer technology, and in particular to a method for predicting atmospheric radiation. Background Art

[0002] The atmospheric radiation transfer parameterization scheme in the numerical weather prediction model is the most time-consuming of all physical process parameterization schemes. Therefore, in order to speed up the calculation of the numerical weather prediction model, the time step of the radiation transfer scheme is usually larger than the time step of the numerical weather prediction model. At present, there are radiation transfer schemes based on fully connected neural network models, which are only applicable to inputs of a specified number of vertical layers, and the generalization ability of the model is weak. Summary of the invention

[0003] In view of this, an embodiment of this specification provides an atmospheric radiation prediction method. One or more embodiments of this specification also relate to an atmospheric radiation prediction device, a computing device, a computer-readable storage medium and a computer program to solve the technical defects existing in the prior art.

[0004] According to a first aspect of an embodiment of this specification, an atmospheric radiation prediction method is provided, which is applied to a radiation prediction model, wherein the radiation prediction model is trained based on atmospheric layer data of a first number of layers, and the radiation prediction model includes a forward network and a backward network, and the method includes:

[0005] Acquire atmospheric data of a second number of layers, sequentially input atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, and

[0006] Inputting the atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information;

[0007] The target radiation information of the corresponding atmospheric layer is determined according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

[0008] According to a second aspect of an embodiment of this specification, an atmospheric radiation prediction device is provided, which is applied to a radiation prediction model, wherein the radiation prediction model is trained based on atmospheric layer data of a first number of layers, the radiation prediction model includes a forward network and a backward network, and the device includes:

[0009] a data input module configured to obtain atmospheric data of a second number of layers, and sequentially input the atmospheric data of each layer of the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information; and

[0010] Inputting the atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information;

[0011] The information determination module is configured to determine the target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

[0012] According to a third aspect of an embodiment of this specification, a computing device is provided, including:

[0013] Memory and processor;

[0014] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the above-mentioned atmospheric radiation prediction method are implemented.

[0015] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned atmospheric radiation prediction method are implemented.

[0016] According to a fifth aspect of the embodiments of this specification, a computer program is provided, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned atmospheric radiation prediction method.

[0017] The embodiments of this specification provide an atmospheric radiation prediction method and device, wherein the atmospheric radiation prediction method is applied to a radiation prediction model, the radiation prediction model is obtained based on training of atmospheric data of a first number layer, the radiation prediction model includes a forward network and a backward network, and the method includes: obtaining atmospheric data of a second number layer, sequentially inputting atmospheric data of each layer in the atmospheric data of the second number layer into the forward network to obtain corresponding first radiation information, and inputting atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; determining target radiation information of the corresponding atmosphere according to the first radiation information and the second radiation information of the atmospheric data of each layer. By inputting atmospheric data of the second number layer into the radiation prediction model obtained by training based on atmospheric data of the first number layer, target radiation information corresponding to the atmosphere of the second number layer is obtained, which can be used for different vertical stratifications of numerical weather forecasts and has stronger generalization ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of an atmospheric radiation prediction method provided by an embodiment of this specification;

[0019] Figure 2 is a schematic diagram of an atmospheric radiation prediction method provided by an embodiment of this specification;

[0020] Figure 3is a processing flow chart of an atmospheric radiation prediction method provided by an embodiment of this specification;

[0021] Figure 4 It is a structural schematic diagram of an atmospheric radiation prediction device provided by an embodiment of this specification;

[0022] Figure 5 It is a structural block diagram of a computing device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0023] Many specific details are described in the following description to facilitate a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of this specification, so this specification is not limited to the specific implementation disclosed below.

[0024] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a", "said" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0025] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0026] First, the terms involved in one or more embodiments of this specification are explained.

[0027] Numerical weather forecast model: refers to a method of predicting the atmospheric motion state and weather phenomena in a certain period of time in the future by performing numerical calculations on large computers under certain initial and boundary conditions based on the actual atmospheric conditions to solve the fluid mechanics and thermodynamics equations that describe the weather evolution process.

[0028] Physical process parameterization scheme: A scheme used to describe atmospheric physical processes that cannot be explicitly resolved by spatial grids under the conditions of current numerical model spatial resolution.

[0029] Atmospheric radiative transfer model: used to calculate radiative transfer in the atmosphere.

[0030] Atmospheric profile refers to the distribution data of oxygen, water vapor and other trace gases in the atmosphere at different altitudes.

[0031] Sequence to sequence model: A sequence-to-sequence conversion model framework, which is used in scenarios such as machine translation and automatic answering.

[0032] Transform model: is a classic NLP model.

[0033] The full name of LSTM is Long Short-Term Memory, which is a type of RNN (Recurrent Neural Network).

[0034] BiLSTM is the abbreviation of Bi-directional Long Short-Term Memory, which is a combination of forward LSTM and backward LSTM.

[0035] Data assimilation: It is a process of combining data from different sources through a series of processing and adjustments to make them available for comprehensive use. In numerical weather forecasting, data assimilation was initially considered to be a process of analyzing and processing observational data distributed in space and time to provide an initial field for numerical forecasting.

[0036] Accompanying mode: refers to a group of moving objects moving together within a limited range for at least w duration. This kind of movement mode is called accompanying mode.

[0037] In this specification, an atmospheric radiation prediction method is provided. This specification also relates to an atmospheric radiation prediction device, a computing device, and a computer-readable storage medium, which are described in detail one by one in the following embodiments.

[0038] See also Figure 1 , Figure 1 A flowchart of an atmospheric radiation prediction method provided according to an embodiment of the present specification is shown. The method is applied to a radiation prediction model, and the radiation prediction model is trained based on atmospheric layer data of a first number of layers. The radiation prediction model includes a forward network and a backward network. The method specifically includes the following steps.

[0039] Step 102: Acquire atmospheric data of a second number of layers, and sequentially input atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, and

[0040] The atmospheric data of each layer is input into the backward network in reverse order to obtain the corresponding second radiation information.

[0041] Among them, the radiation prediction model can be a sequence to sequence model, or a transform model, for example: a BiLSTM model; the forward network or the backward network is the forward network or the backward network in the structure of the radiation prediction model, for example, the forward network is a forward LSTM or a backward LSTM; the second number of layers can be any number of layers, for example, 10 layers; the atmospheric layer data can be the atmospheric profile in the glossary, for example, temperature, air pressure, humidity, liquid water content, water vapor, cloud water, cloud ice, ozone, carbon dioxide, methane, nitrous oxide, oxygen, carbon tetrachloride, chlorofluorocarbons, etc.; the radiation information can be long-wave or short-wave radiation information, for example, radiation flux.

[0042] In practical applications, the radiation prediction model can be a BiLSTM model, which includes a forward LSTM and a backward LSTM. The model can use the atmospheric data of the first number of layers, such as 10 layers, during training. When in use, the atmospheric data of the second number of layers (for example, 20 layers) are passed through the forward network to obtain the first radiation information. The first radiation information can be the radiation flux of long waves in each atmospheric layer, that is, the radiation of sunlight reflected from the ground to high altitude. The atmospheric data of the second number of layers are passed through the backward network to obtain the second radiation information. The second radiation information can be the radiation flux of short waves in each atmospheric layer, that is, the radiation of sunlight reaching the ground from high altitude.

[0043] For example, the radiation prediction model uses 10 layers of atmospheric data during training; 20 layers of atmospheric data are obtained, the 20 layers of atmospheric data are passed through a forward network to obtain the first radiation information of the 20 layers of atmospheric data, and the 20 layers of atmospheric data are passed through a backward network to obtain the second radiation information of the 20 layers of atmospheric data.

[0044] When the atmospheric layer data of the second number of layers are passed through the forward network or the backward network, the input order is determined according to the order of the atmospheric layers, and the specific implementation is as follows.

[0045] The step of sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain the corresponding first radiation information comprises:

[0046] sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network, and

[0047] The first initial parameters of each layer are input into the forward network to obtain corresponding first radiation information and first transfer parameters.

[0048] In practical applications, the forward LSTM or backward LSTM will generate different parameters for downward transmission according to the input order, so it is necessary to simulate the order of illumination for input. In other words, the forward LSTM has a first initial parameter and the backward LSTM has a second initial parameter. Each time a layer of atmospheric data is input, the first initial parameter or the second initial parameter will also change during the processing.

[0049] For example, the radiation prediction model uses 10 layers of atmospheric data during training; 20 layers of atmospheric data are obtained, and these 20 layers of atmospheric data represent the atmospheric data from the ground to 100 kilometers in the sky, that is, 0-5 kilometers are the atmospheric data of the first layer, 5-10 kilometers are the atmospheric data of the second layer, ..., 95-100 kilometers are the atmospheric data of the 20th layer, and these 20 layers of atmospheric data are sequentially passed through the forward network from the atmospheric data of the first layer and the first initial parameters of the first layer to the first initial parameters of the atmosphere of the 20th layer and the atmospheric data of the 20th layer to obtain the first radiation information of the atmospheric data of the 20 layers (radiation information of sunlight reflected from the ground to the sky) and the first transfer parameters of the 20 layers.

[0050] Specifically, the atmospheric data may be input in a cyclic manner, and the specific implementation is described as follows.

[0051] The step of sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain the corresponding first radiation information comprises:

[0052] Inputting the first initial parameter of the i-th layer and the atmospheric data of the i-th layer in the atmospheric data of the second number of layers into the forward network, obtaining the first radiation information of the i-th layer and the first transfer parameter of the i-th layer, wherein the initial value of i is 1, and i is less than or equal to the second number;

[0053] The transfer parameters of the i-th layer are used as the first initial parameters of the i+1-th layer, and i is incremented by 1.

[0054] In practical applications, except for the initial parameters of the first layer which are pre-set, the initial parameters of other layers are the transfer parameters obtained after the calculation of the previous layer is completed.

[0055] For example, see Figure 2, the first initial parameter of the first layer of the forward network is h0. By inputting h0 and the atmosphere data xt of the first layer into the forward LSTM, the first radiation information yta1 of the first layer and the first transfer parameter h1 can be obtained. The first transfer parameter h1 is used as the first initial parameter of the second layer; then the first initial parameter h1 of the second layer and the atmosphere data xt+1 of the second layer are input into the forward LSTM to obtain the first radiation information yta2 of the second layer and the first transfer parameter h2, ..., the first initial parameter hT-1 of the Tth layer and the atmosphere data xt+T of the Tth layer are input into the forward LSTM to obtain the first radiation information ytaT of the Tth layer and the first transfer parameter hT.

[0056] It should be noted that the embodiments of this specification do not limit the method of sequentially inputting the atmospheric data of each layer into the forward network, as long as the function can be achieved.

[0057] Accordingly, the input method of the backward network is described as follows.

[0058] The step of inputting the atmospheric data of each layer into the backward network in reverse order to obtain the corresponding second radiation information comprises:

[0059] Inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the backward network in reverse order, and

[0060] The second initial parameters of each layer are input into the backward network to obtain corresponding second radiation information and second transfer parameters.

[0061] For example, the radiation prediction model uses 10 layers of atmospheric data during training; 20 layers of atmospheric data are obtained, and these 20 layers of atmospheric data represent the atmospheric data from the ground to 100 kilometers in the sky, that is, 0-5 kilometers are the atmospheric data of the first layer, 5-10 kilometers are the atmospheric data of the second layer, ..., 95-100 kilometers are the atmospheric data of the 20th layer, and these 20 layers of atmospheric data are sequentially passed through the backward network from the atmospheric data of the 20th layer and the second initial parameters of the 20th layer to the second initial parameters of the first layer and the atmospheric data of the first layer to obtain the second radiation information of the 20 layers of atmospheric data (radiation information of sunlight from high altitude to the ground) and the second transfer parameters of the 20 layers.

[0062] Specifically, the step of inputting the atmospheric data of each layer into the backward network in reverse order to obtain the corresponding second radiation information includes:

[0063] Inputting the second initial parameter of the jth layer and the atmospheric data of the jth layer among the atmospheric data of the second number of layers into the forward network, obtaining the second radiation information of the jth layer and the second transfer parameter of the jth layer, wherein the initial value of j is equal to the second number, and j is greater than or equal to 1;

[0064] The transfer parameters of the jth layer are used as the second initial parameters of the j+1th layer, and j is incremented by 1.

[0065] For example, see Figure 2 , the second initial parameter of the Tth layer of the backward network is h0'. By inputting h0' and the atmospheric data xt+T of the Tth layer into the backward LSTM, the second radiation information ytbT of the Tth layer and the second transfer parameter h1' can be obtained. The second transfer parameter h1' is used as the second initial parameter of the T-1th layer; then the second initial parameter h1' of the T-1th layer and the atmospheric data xt+T-1 of the T-1th layer are input into the backward LSTM to obtain the second radiation information ytbT-1 of the T-1th layer and the second transfer parameter h2', ..., the second initial parameter hT-1' of the first layer and the atmospheric data xt of the first layer are input into the backward LSTM to obtain the second radiation information ytb1 of the first layer and the second transfer parameter hT'.

[0066] It should be noted that the embodiment of this specification does not limit the method of inputting the atmospheric data of each layer into the backward network in reverse order, as long as the function can be achieved.

[0067] The embodiments of this specification can be used for vertical stratification of different atmospheric layers and have stronger generalization capabilities.

[0068] Step 104: Determine target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

[0069] The target radiation information may be information obtained by superimposing all radiation information, for example, information obtained by superimposing the first radiation information and the second radiation information.

[0070] In practical applications, the calculated first radiation information and the second radiation information are superimposed to obtain the target radiation information. That is to say, the target radiation information of each layer includes the radiation information of sunlight from high altitude to the ground, and the radiation information of sunlight reflected from the ground to high altitude.

[0071] Specifically, determining the target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer includes:

[0072] The first radiation information and the second radiation information of the atmospheric data of each layer are calculated according to a preset calculation rule to obtain the target radiation information of each layer.

[0073] Among them, the preset calculation rules can be addition, subtraction and other rules.

[0074] Using the above example, see Figure 2 , add the first radiation information yta1 and the second radiation information ytb1 of the first layer to obtain yt, add the first radiation information yta2 and the second radiation information ytb2 of the second layer to obtain yt+1,..., add the first radiation information ytaT and the second radiation information ytbT of the Tth layer to obtain yt+T.

[0075] Furthermore, the temperature change rate may also be determined according to the target radiation information of each layer, and the specific implementation method is as follows.

[0076] The atmospheric radiation prediction method of the embodiment of this specification also includes:

[0077] The temperature change rate is determined based on the target radiation information of the atmospheric data of all layers.

[0078] In practical applications, other related functions can be added to the model, for example, directly outputting the temperature change rate, thereby improving the usability of the model.

[0079] Furthermore, after receiving the atmospheric data, the order of the atmospheric data can be determined according to the carried sequence identifier, so as to facilitate the subsequent input of the atmospheric data into the forward network and the backward network according to the order. The specific implementation method is described as follows.

[0080] Specifically, after acquiring the atmospheric data of the second number layer, the method further includes:

[0081] Obtaining a sequence identifier carried by the atmospheric data of each layer in the atmospheric data of the second number of layers;

[0082] The order of the atmospheric layer data of the second number of layers is determined according to the sequence identifier.

[0083] The sequence identifier may be a text, number or other identifier, for example, 1, 2, 3, etc.

[0084] For example, if the sequence identifier of a certain atmospheric layer data received is 1, it is determined that the atmospheric layer data is the atmospheric layer data of the first layer.

[0085] Furthermore, deep learning models are used to accelerate the atmospheric radiation transfer parameterization scheme in the numerical weather forecast model, reduce the atmospheric radiation transfer calculation time, and increase the frequency of radiation scheme calls, thereby improving forecast accuracy.

[0086] Specifically, after determining the target radiation information of the corresponding atmospheric layer, the method further includes:

[0087] Acquire atmospheric dynamic information, and predict weather information based on the target radiation information and the atmospheric dynamic information.

[0088] In practical applications, the atmospheric radiation transfer parameterization scheme in the numerical weather forecast model is the most time-consuming of all physical process parameterization schemes. Therefore, in order to speed up the calculation of the numerical weather forecast model, the time step of the radiation transfer scheme is usually larger than the time step of the numerical weather forecast model. In the presence of clouds, especially when the state of the clouds changes rapidly, a shorter radiation transfer scheme step is conducive to more accurate radiation calculations. Replacing the radiation transfer scheme with a deep learning model can more efficiently calculate atmospheric radiation, thereby shortening the radiation transfer scheme time step and obtaining a more accurate forecast.

[0089] Furthermore, the atmospheric radiation prediction method of the embodiment of this specification also includes:

[0090] The atmospheric data of the corresponding layer is determined according to the target radiation information.

[0091] Specifically, determining the atmospheric data of the corresponding layer according to the target radiation information includes:

[0092] The target radiation information is derived to obtain atmospheric data of the corresponding layer.

[0093] In practical applications, the automatic differentiation capability of deep learning is used to obtain the adjoint mode corresponding to the radiation transfer, which can be used to improve the accuracy of data assimilation.

[0094] The embodiment of this specification provides an atmospheric radiation prediction method applied to a radiation prediction model, the radiation prediction model is obtained based on training of atmospheric data of a first number layer, the radiation prediction model includes a forward network and a backward network, the method includes: obtaining atmospheric data of a second number layer, sequentially inputting atmospheric data of each layer in the atmospheric data of the second number layer into the forward network to obtain corresponding first radiation information, and inputting atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; according to the first radiation information and the second radiation information of the atmospheric data of each layer, determining the target radiation information of the corresponding atmosphere. By inputting the atmospheric data of the second number layer into the radiation prediction model obtained by training based on the atmospheric data of the first number layer, the target radiation information corresponding to the atmospheric layer of the second number layer is obtained, which can be used for different vertical layers of numerical weather forecasts and has stronger generalization ability.

[0095] The following combination Figure 3Taking the application of the atmospheric radiation prediction method provided in this specification in a server as an example, the atmospheric radiation prediction method is further described. Figure 3 A processing flow chart of an atmospheric radiation prediction method provided by an embodiment of the present specification is shown, which specifically includes the following steps.

[0096] Step 302: The server obtains atmospheric data of a second number of layers.

[0097] Step 304: The server inputs the first initial parameter of the i-th layer and the atmospheric data of the i-th layer among the atmospheric data of the second number of layers into the forward network to obtain the first radiation information of the i-th layer and the first transfer parameter of the i-th layer, where the initial value of i is 1, and i is less than or equal to the second number.

[0098] Step 306: The server uses the transfer parameter of the i-th layer as the first initial parameter of the i+1-th layer, and increments i by 1.

[0099] Step 308: The server determines whether i is greater than the second number. If yes, execute step 310. If no, execute step 304.

[0100] Step 310: The server inputs the second initial parameter of the jth layer and the atmospheric data of the jth layer among the atmospheric data of the second number of layers into the forward network to obtain the second radiation information of the jth layer and the second transfer parameter of the jth layer, wherein the initial value of j is equal to the second number, and j is greater than or equal to 1.

[0101] Step 312: The server uses the transfer parameter of the jth layer as the second initial parameter of the j+1th layer, and increments j by 1.

[0102] Step 314: The server determines whether j is less than 1. If so, execute step 316. If not, execute step 310.

[0103] Step 316: The server calculates the first radiation information and the second radiation information of each layer of atmospheric data according to a preset calculation rule to obtain target radiation information of each layer.

[0104] The embodiment of this specification obtains the target radiation information corresponding to the atmosphere of the second number layer by inputting the atmospheric data of the second number layer into the radiation prediction model trained based on the atmospheric data of the first number layer. This can be used for different vertical stratifications of numerical weather forecasts and has stronger generalization ability.

[0105] Corresponding to the above method embodiment, this specification also provides an atmospheric radiation prediction device embodiment, Figure 4 FIG. 1 is a schematic diagram showing the structure of an atmospheric radiation prediction device provided by an embodiment of the present specification. Figure 4 As shown, the device is applied to a radiation prediction model, the radiation prediction model is obtained by training based on atmospheric data of a first number of layers, the radiation prediction model includes a forward network and a backward network, and the device includes:

[0106] The data input module 402 is configured to obtain atmospheric data of a second number of layers, and sequentially input the atmospheric data of each layer of the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, and

[0107] Inputting the atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information;

[0108] The information determination module 404 is configured to determine the target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

[0109] Furthermore, the data input module 402 is also configured to:

[0110] sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network, and

[0111] The first initial parameters of each layer are input into the forward network to obtain corresponding first radiation information and first transfer parameters.

[0112] Furthermore, the data input module 402 is also configured to:

[0113] Inputting the first initial parameter of the i-th layer and the atmospheric data of the i-th layer in the atmospheric data of the second number of layers into the forward network, obtaining the first radiation information of the i-th layer and the first transfer parameter of the i-th layer, wherein the initial value of i is 1, and i is less than or equal to the second number;

[0114] The transfer parameters of the i-th layer are used as the first initial parameters of the i+1-th layer, and i is incremented by 1.

[0115] Furthermore, the data input module 402 is also configured to:

[0116] Inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the backward network in reverse order, and

[0117] The second initial parameters of each layer are input into the backward network to obtain corresponding second radiation information and second transfer parameters.

[0118] Furthermore, the data input module 402 is also configured to:

[0119] Inputting the second initial parameter of the jth layer and the atmospheric data of the jth layer among the atmospheric data of the second number of layers into the forward network, obtaining the second radiation information of the jth layer and the second transfer parameter of the jth layer, wherein the initial value of j is equal to the second number, and j is greater than or equal to 1;

[0120] The transfer parameters of the jth layer are used as the second initial parameters of the j+1th layer, and j is incremented by 1.

[0121] Furthermore, the information determination module 404 is further configured to:

[0122] The first radiation information and the second radiation information of the atmospheric data of each layer are calculated according to a preset calculation rule to obtain the target radiation information of each layer.

[0123] Furthermore, the information determination module 404 is further configured to:

[0124] The temperature change rate is determined based on the target radiation information of the atmospheric data of all layers.

[0125] Furthermore, the data input module 402 is also configured to:

[0126] Obtaining a sequence identifier carried by the atmospheric data of each layer in the atmospheric data of the second number of layers;

[0127] The order of the atmospheric layer data of the second number of layers is determined according to the sequence identifier.

[0128] Furthermore, the information determination module 404 is further configured to:

[0129] Acquire atmospheric dynamic information, and predict weather information based on the target radiation information and the atmospheric dynamic information.

[0130] Furthermore, the information determination module 404 is further configured to:

[0131] The atmospheric data of the corresponding layer is determined according to the target radiation information.

[0132] Furthermore, the information determination module 404 is further configured to:

[0133] The target radiation information is derived to obtain atmospheric data of the corresponding layer.

[0134] The embodiment of the present specification provides an atmospheric radiation prediction device, which is applied to a radiation prediction model, wherein the radiation prediction model is obtained by training based on atmospheric data of a first number layer, and the radiation prediction model includes a forward network and a backward network, and the method includes: obtaining atmospheric data of a second number layer, sequentially inputting atmospheric data of each layer in the atmospheric data of the second number layer into the forward network to obtain corresponding first radiation information, and inputting atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; and determining target radiation information of the corresponding atmosphere according to the first radiation information and the second radiation information of the atmospheric data of each layer. By inputting atmospheric data of the second number layer into the radiation prediction model obtained by training based on atmospheric data of the first number layer, target radiation information corresponding to the atmosphere of the second number layer is obtained, which can be used for different vertical stratifications of numerical weather forecasts and has stronger generalization ability.

[0135] The above is a schematic scheme of an atmospheric radiation prediction device of this embodiment. It should be noted that the technical scheme of the atmospheric radiation prediction device and the technical scheme of the atmospheric radiation prediction method described above belong to the same concept, and the details not described in detail in the technical scheme of the atmospheric radiation prediction device can be referred to the description of the technical scheme of the atmospheric radiation prediction method described above.

[0136] Figure 5 The structure block diagram of a computing device 500 provided according to an embodiment of the present specification is shown. The components of the computing device 500 include but are not limited to a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and the database 550 is used to store data.

[0137] The computing device 500 also includes an access device 540 that enables the computing device 500 to communicate via one or more networks 560. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 540 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a World Wide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0138] In one embodiment of the present specification, the above components of the computing device 500 and Figure 5 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 5The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0139] The computing device 500 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or PC. The computing device 500 may also be a mobile or stationary server.

[0140] The processor 520 is used to execute the following computer executable instructions: when the computer executable instructions are executed by the processor, the steps of the above-mentioned atmospheric radiation prediction method are implemented.

[0141] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above atmospheric radiation prediction method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above atmospheric radiation prediction method.

[0142] An embodiment of the present specification also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned atmospheric radiation prediction method.

[0143] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the above-mentioned atmospheric radiation prediction method belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the above-mentioned atmospheric radiation prediction method.

[0144] An embodiment of the present specification also provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned atmospheric radiation prediction method.

[0145] The above is a schematic scheme of a computer program of this embodiment. It should be noted that the technical scheme of the computer program and the technical scheme of the above atmospheric radiation prediction method belong to the same concept, and the details not described in detail in the technical scheme of the computer program can be referred to the description of the technical scheme of the above atmospheric radiation prediction method.

[0146] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0147] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0148] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0149] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0150] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The optional embodiments do not describe all the details in detail, nor do they limit the invention to only the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that technicians in the relevant technical field can well understand and use this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. An atmospheric radiation prediction method, applied to a radiation prediction model, wherein the radiation prediction model is trained based on atmospheric layer data of a first number of layers, the radiation prediction model comprises a forward network and a backward network, and the method comprises: Obtaining atmospheric data of a second number of layers, and sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, wherein the atmospheric data refers to the distribution data of each gas in the atmosphere at different altitudes, and sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, including: sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network, and inputting the first initial parameter of each layer into the forward network to obtain corresponding first radiation information and first transfer parameter; or sequentially inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, including: inputting the first initial parameter of the i-th layer and the atmospheric data of the i-th layer in the atmospheric data of the second number of layers into the forward network to obtain first radiation information of the i-th layer and first transfer parameter of the i-th layer, wherein the initial value of i is 1, and i is less than or equal to the second number; using the transfer parameter of the i-th layer as the first initial parameter of the i+1-th layer, and increasing i by 1, and Inputting the atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; The target radiation information of the corresponding atmospheric layer is determined according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

2. The method according to claim 1, wherein the step of inputting the atmospheric data of each layer into the backward network in reverse order to obtain the corresponding second radiation information comprises: Inputting the atmospheric data of each layer in the atmospheric data of the second number of layers into the backward network in reverse order, and The second initial parameters of each layer are input into the backward network to obtain corresponding second radiation information and second transfer parameters.

3. The method according to claim 1, wherein the step of inputting the atmospheric data of each layer into the backward network in reverse order to obtain the corresponding second radiation information comprises: Inputting the second initial parameter of the jth layer and the atmospheric data of the jth layer among the atmospheric data of the second number of layers into the forward network, obtaining the second radiation information of the jth layer and the second transfer parameter of the jth layer, wherein the initial value of j is equal to the second number, and j is greater than or equal to 1; The transfer parameters of the jth layer are used as the second initial parameters of the j+1th layer, and j is incremented by 1.

4. The method according to claim 1, wherein determining the target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer comprises: The first radiation information and the second radiation information of the atmospheric data of each layer are calculated according to a preset calculation rule to obtain the target radiation information of each layer.

5. The method according to claim 1, further comprising: The temperature change rate is determined based on the target radiation information of the atmospheric data of all layers.

6. The method according to claim 1, after acquiring the atmospheric data of the second number layer, further comprises: Obtaining a sequence identifier carried by the atmospheric data of each layer in the atmospheric data of the second number of layers; The order of the atmospheric layer data of the second number of layers is determined according to the sequence identifier.

7. The method according to claim 1, after determining the target radiation information of the corresponding atmospheric layer, further comprises: Acquire atmospheric dynamic information, and predict weather information based on the target radiation information and the atmospheric dynamic information.

8. The method according to claim 1, further comprising: The atmospheric data of the corresponding layer is determined according to the target radiation information.

9. The method according to claim 8, wherein determining atmospheric data of a corresponding layer according to the target radiation information comprises: The target radiation information is derived to obtain atmospheric data of the corresponding layer.

10. An atmospheric radiation prediction device, applied to a radiation prediction model, wherein the radiation prediction model is trained based on atmospheric layer data of a first number of layers, the radiation prediction model comprises a forward network and a backward network, and the device comprises: a data input module configured to obtain atmospheric data of a second number of layers, and sequentially input the atmospheric data of each layer of the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, wherein the atmospheric data refers to distribution data of each gas in the atmosphere at different altitudes, and Inputting the atmospheric data of each layer into the backward network in reverse order to obtain corresponding second radiation information; The data input module is further configured to obtain atmospheric data of a second number of layers, sequentially input the atmospheric data of each layer in the atmospheric data of the second number of layers into the forward network to obtain corresponding first radiation information, and reversely input the atmospheric data of each layer into the backward network to obtain corresponding second radiation information; or, the data input module is further configured to input the first initial parameter of the i-th layer and the atmospheric data of the i-th layer in the atmospheric data of the second number of layers into the forward network to obtain the first radiation information of the i-th layer and the first transfer parameter of the i-th layer, wherein the initial value of i is 1, and i is less than or equal to the second number; use the transfer parameter of the i-th layer as the first initial parameter of the i+1-th layer, and increment i by 1; The information determination module is configured to determine the target radiation information of the corresponding atmospheric layer according to the first radiation information and the second radiation information of the atmospheric layer data of each layer.

11. A computing device comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the atmospheric radiation prediction method described in any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the atmospheric radiation prediction method according to any one of claims 1 to 9.

13. A computer program product, characterized in that The method comprises computer instructions which, when executed by a processor, implement the steps of the atmospheric radiation prediction method according to any one of claims 1 to 9.

Images