CN116384241A - Method for constructing rapid radiation transmission model based on deep neural network - Google Patents

Method for constructing rapid radiation transmission model based on deep neural network Download PDF

Info

Publication number
CN116384241A
CN116384241A CN202310337331.3A CN202310337331A CN116384241A CN 116384241 A CN116384241 A CN 116384241A CN 202310337331 A CN202310337331 A CN 202310337331A CN 116384241 A CN116384241 A CN 116384241A
Authority
CN
China
Prior art keywords
neural network
angle
radiation transmission
deep neural
transmission model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310337331.3A
Other languages
Chinese (zh)
Inventor
王红梅
王雅鹏
张欣欣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nantong University
Original Assignee
Nantong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nantong University filed Critical Nantong University
Priority to CN202310337331.3A priority Critical patent/CN116384241A/en
Publication of CN116384241A publication Critical patent/CN116384241A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
    • G01N21/33—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using ultraviolet light
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Geometry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computer Hardware Design (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Photometry And Measurement Of Optical Pulse Characteristics (AREA)

Abstract

The invention belongs to the technical field of satellite remote sensing, and particularly relates to a method for constructing a rapid radiation transmission model based on a deep neural network. The invention comprises the following steps: step 101, simulating a near-edge ultraviolet spectrum data set based on an atmospheric radiation transmission model; step 102, training the near-edge ultraviolet spectrum data set serving as a training sample through a neural network to obtain a deep learning model; and 103, extracting longitude and latitude, time, solar zenith angle, observation angle, azimuth angle and height cutting sequence of a first-class product of the satellite limb load in real time, and calculating to obtain solar radiation with different height cutting received by a limb detector by adopting the deep learning model. The method for constructing the clear sky adjacent side ultraviolet rapid radiation transmission model based on the deep neural network can calculate solar radiation under the adjacent side observation under the clear sky condition by adopting the method of the deep neural network, and has the advantages of high calculation speed, strong universality, no influence of an operating system and high portability.

Description

Method for constructing rapid radiation transmission model based on deep neural network
Technical Field
The invention belongs to the technical field of satellite remote sensing, and particularly relates to a method for constructing a rapid radiation transmission model based on a deep neural network.
Background
In the process of inverting the atmospheric components by the critical edges, the atmospheric radiation transmission model plays a key role. There are methods for calculating the model of atmospheric radiation transmission based on line-by-line integration and look-up tables. For example, the progressive integration model is based on a progressive integration forward radiation transmission model SCIATRAN, and the progressive integration needs to read spectral line parameters of an atmospheric high-resolution molecular absorption database, such as HITRAN, and calculate atmospheric absorption, emission and reflection spectrums of different spectral bands according to the input atmospheric parameters and surface parameters. The look-up table method has been widely used with an interpretable physical mechanism for atmospheric radiation transport and with reasonable computational efficiency. The disadvantage is that the computational efficiency is very low when the input parameters are too many. The critical edge inversion requires the use of an accurate and CPU efficient forward model to simulate the radiation transmission through the atmosphere and take into account the spectral and spatial characteristics (dispersion, slit functions) of the instrument.
However, the current radiation transmission model has lower efficiency in the edge inversion, so that the current radiation transmission model needs to be improved, thereby improving the inversion speed. Deep learning is to learn the internal law and the representation level of sample data to obtain a nonlinear relation between output and input; and combining the deep neural network with the forward radiation transmission model to construct a rapid forward model, thereby improving the business operation capability.
Disclosure of Invention
The invention aims at: aiming at the defects that an ultraviolet limb radiation transmission model is long in time consumption, poor in universality and unsuitable for service when solar radiation is calculated, the method for constructing the rapid radiation transmission model based on the deep neural network is simpler, more convenient, faster and more accurate.
In order to achieve the aim of the invention, the technical scheme adopted by the invention is as follows:
a method for constructing a fast radiation transmission model based on a deep neural network comprises the following steps:
step 101, simulating a near-edge ultraviolet spectrum data set based on an atmospheric radiation transmission model; the near-edge ultraviolet spectrum data set comprises the following data: longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmospheric background library, height cutting sequence and solar radiation received by a limb detector at different heights;
step 102, training the near-edge ultraviolet spectrum data set serving as a training sample through a neural network to obtain a deep learning model; the device comprises a longitudinal and latitudinal direction, a date, a solar zenith angle, an observation angle, an azimuth angle, an earth surface reflectivity, an angle of view, an instrument slit function, an atmospheric background library and a height cutting sequence, which are used as input layers of a neural network, and solar radiation with different height cutting received by a boundary detector is used as output layers of the neural network;
and 103, extracting longitude and latitude, time, solar zenith angle, observation angle, azimuth angle and height cutting sequence of a first-class product of the satellite limb load in real time, and calculating to obtain solar radiation with different height cutting received by a limb detector by adopting the deep learning model.
In a further preferred embodiment of the present invention, in step 101, the atmospheric radiation transmission model is a SCIATRAN forward radiation transmission model; the atmosphere background library adopts the American standard atmosphere; the software used for simulating the atmospheric radiation transmission model is version SCIATRAN4.5.7; the standard atmosphere in the United states is divided into five atmosphere modes, namely tropical atmosphere, northern hemisphere middle latitude, northern hemisphere polar region, southern hemisphere middle latitude and southern hemisphere polar region.
Further as a preferred technical scheme of the present invention, in the step 103, the ERA5 re-analysis data and the atmospheric profile of the AURA MLS secondary product are extracted as a real-time atmospheric background library, and the MODIS secondary surface reflectivity product is extracted as a real-time surface reflectivity.
Further as a preferable technical scheme of the invention, the near-edge ultraviolet spectrum range is 300nm-500nm, and the spectrum interval is 0.4nm; the value range of the cut-up sequence is 5-65km, and the set interval is 100m.
Further as a preferable technical scheme of the invention, the neural network comprises 1 input layer, 1 output layer and 3 hidden layers; the 3 hidden layers contain 256, 128, 64 neurons, respectively.
Further as a preferred technical scheme of the invention, the longitude and latitude of the near-edge ultraviolet spectrum data set takes the latitude interval 90 DEG S-90 DEG N as the value of 5 DEG, and respectively takes any one of 90 DEG N, 85 DEG N, 80 DEG N, 75 DEG N, 70 DEG N, 65 DEG N, 60 DEG N, 55 DEG N, 50 DEG N, 45 DEG N, 40 DEG N, 35 DEG N, 30 DEG N, 25 DEG N, 20 DEG N, 15 DEG N, 10 DEG N, 5 DEG N, 0 DEG S, 5 DEG S, 10 DEG S, 15 DEG S, 20 DEG S, 25 DEG S, 30 DEG S, 35 DEG S, 40 DEG S, 45 DEG S, 50 DEG S, 55 DEG S, 60 DEG S, 65 DEG S, 70 DEG S, 75 DEG S, 80 DEG S, 85 DEG S and 90 DEG S; taking the longitude interval 180 DEG W-180 DEG E as a value at intervals of 30 DEG, and taking any one of 180 DEG E, 150 DEG E, 120 DEG E, 90 DEG E, 60 DEG E, 30 DEG E, 0 DEG, 30 DEG W, 60 DEG W, 120 DEG W, 150 DEG W and 180 DEG W respectively.
Further, as a preferable embodiment of the present invention, the solar zenith angles are respectively any one of 0 °,5 °, 10 °, 15 °, 20 °, 25 °, 30 °, 35 °, 40 °, 45 °, 50 °, 55 °, 60 °, 70 °, 80 ° and 90 ° at intervals of 5 °; the observation angle is calculated and set according to the positions of longitude and latitude, and the setting interval is 2 degrees; the angle of view is set according to the instrument, the setting interval is 1, and the earth's surface reflectivity value is any one of 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and 1.0.
Further, as a preferable technical scheme of the invention, a pytorch framework is adopted to perform deep neural network learning.
Compared with the prior art, the method for constructing the rapid radiation transmission model based on the deep neural network has the following technical effects:
the method for constructing the clear sky adjacent side ultraviolet rapid radiation transmission model based on the deep neural network can calculate solar radiation under the adjacent side observation under the clear sky condition by adopting the method of the deep neural network, and has the advantages of high calculation speed, strong universality, no influence of an operating system and high portability.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention;
FIG. 2 is a schematic diagram of training sample generation using SCITRA in accordance with the present invention;
FIG. 3 is a schematic diagram of model training using a deep neural network in accordance with the present invention;
fig. 4 is a schematic diagram of calculating an ultraviolet limb spectrum by using actual extraction longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity and cutting height sequence, and inputting field angle of limb load and instrument slit function.
Detailed Description
The invention is further explained in the following detailed description with reference to the drawings so that those skilled in the art can more fully understand the invention and can practice it, but the invention is explained below by way of example only and not by way of limitation.
As shown in fig. 1, a method for constructing a fast radiation transmission model based on a deep neural network includes the following steps:
step 101, simulating a near-edge ultraviolet spectrum data set based on an atmospheric radiation transmission model; the near-edge ultraviolet spectrum data set comprises the following data: longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmospheric background library, height cutting sequence and solar radiation received by a limb detector at different heights;
step 102, training the near-edge ultraviolet spectrum data set serving as a training sample through a neural network to obtain a deep learning model; the device comprises a longitudinal and latitudinal direction, a date, a solar zenith angle, an observation angle, an azimuth angle, an earth surface reflectivity, an angle of view, an instrument slit function, an atmospheric background library and a height cutting sequence, which are used as input layers of a neural network, and solar radiation with different height cutting received by a boundary detector is used as output layers of the neural network;
and 103, extracting longitude and latitude, time, solar zenith angle, observation angle, azimuth angle and height cutting sequence of a first-class product of the satellite limb load in real time, and calculating to obtain solar radiation with different height cutting received by a limb detector by adopting the deep learning model.
In step 101, the atmospheric radiation transmission model is a SCIATRAN forward radiation transmission model; the atmosphere background library adopts the American standard atmosphere; the software used for simulating the atmospheric radiation transmission model is version SCIATRAN4.5.7; the standard atmosphere in the United states is divided into five atmosphere modes, namely tropical atmosphere, northern hemisphere middle latitude, northern hemisphere polar region, southern hemisphere middle latitude and southern hemisphere polar region.
In step 103, ERA5 analysis data and the atmospheric profile of the AURAMLS secondary product are extracted as a real-time atmospheric background library, and the MODIS secondary surface reflectivity product is extracted as a real-time surface reflectivity.
The near-edge ultraviolet spectrum range is 300nm-500nm, and the spectrum interval is 0.4nm; the value range of the cut-up sequence is 5-65km, and the set interval is 100m. The neural network comprises 1 input layer, 1 output layer and 3 hidden layers; the 3 hidden layers contain 256, 128, 64 neurons, respectively.
Taking the latitude and longitude of the clinical ultraviolet spectrum data set, taking the latitude interval 90 DEG S-90 DEG N as the interval of 5 DEG, and taking any one of 90 DEG N, 85 DEG N, 80 DEG N, 75 DEG N, 70 DEG N, 65 DEG N, 60 DEG N, 55 DEG N, 50 DEG N, 45 DEG N, 40 DEG N, 35 DEG N, 30 DEG N, 25 DEG N, 20 DEG N, 15 DEG N, 10 DEG N, 5 DEG N, 0 DEG, 5 DEG S, 10 DEG S, 15 DEG S, 20 DEG S, 25 DEG S, 30 DEG S, 35 DEG S, 40 DEG S, 45 DEG S, 50 DEG S, 55 DEG S, 60 DEG S, 65 DEG S, 70 DEG S, 75 DEG S, 80 DEG S, 85 DEG S and 90 DEG S; taking the longitude interval 180 DEG W-180 DEG E as a value at intervals of 30 DEG, and taking any one of 180 DEG E, 150 DEG E, 120 DEG E, 90 DEG E, 60 DEG E, 30 DEG E, 0 DEG, 30 DEG W, 60 DEG W, 120 DEG W, 150 DEG W and 180 DEG W respectively.
The solar zenith angles are respectively any one of 0 degree, 5 degree, 10 degree, 15 degree, 20 degree, 25 degree, 30 degree, 35 degree, 40 degree, 45 degree, 50 degree, 55 degree, 60 degree, 70 degree, 80 degree and 90 degree with 5 degree as interval; the observation angle is calculated and set according to the positions of longitude and latitude, and the setting interval is 2 degrees; the angle of view is set according to the instrument, the setting interval is 1, and the earth's surface reflectivity value is any one of 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and 1.0.
Deep neural network learning is performed using a pytorch framework.
In specific implementation, referring to fig. 2, a schematic diagram of training sample generation by SCIATRAN according to an embodiment of the present invention specifically includes the following steps:
step 201, longitude, latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmosphere background library, cut-up sequence and wave band range are used as input data of a forward radiation transmission model; firstly, positioning a product on a monthly scale according to a time dimension; positioning the space dimension on a theodolite grid, wherein the latitude is at intervals of 5 degrees, and the longitude is at intervals of 30 degrees; the solar zenith angle is divided into 0-90 degrees at 5-degree intervals, and the observation angles are set at 2 degrees; the view angle is set according to the instrument, and the surface reflectivity is any one of 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and 1.0; the value range of the cut-up sequence is 5-65km, and the interval is set to be 100m; the atmosphere background library is the American standard atmosphere and is divided into 5 atmosphere modes of a tropical region, a northern hemisphere middle latitude, a northern hemisphere polar region, a southern hemisphere middle latitude and a southern hemisphere polar region; the wave band range is 300-500 nm, and the spectrum interval is 0.4nm;
step 202, according to the input, using a radiation transmission model SCITRA and a programming tool matlab, circularly calling the SCITRA model at the matlab, and modifying parameter files of the SCITRA model, wherein the parameter files comprise control.inp, control_geom.inp, control_ac.inp, control_la.inp, and section.inp;
programming a program for modifying each control file of an atmospheric radiation transmission model SCITRA by using matlab software, wherein the program can repeatedly modify longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, atmospheric background library and cut-up sequence; inputting the modified longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmospheric background library and cut-height sequence into each configuration file of an atmospheric radiation transmission model to obtain simulated spectrum values of different cut heights in an adjacent ultraviolet band, wherein the number of simulated spectrums is 402;
and 203, outputting and generating solar radiation intensity data under different atmospheric background conditions, different earth surface reflectivities, different solar zenith angles, different areas and different cut heights.
Referring to fig. 3, the implementation steps of the model training method using the deep neural network are as follows:
taking longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmospheric background library, height cutting sequence and solar radiation with different heights as input layers; and training through a neural network to obtain a deep learning model, wherein solar radiation with different heights is used as an output layer of the neural network.
Referring to fig. 4, the invention uses the actual extraction longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity and cut-up sequence, inputs the field angle of the critical load and instrument slit function, calculates the schematic diagram of the ultraviolet critical spectrum, and the implementation steps are as follows:
step 301, extracting and analyzing ERA5 products and secondary products of AURAMLS according to longitude and latitude and date of detection data;
step 302, extracting an MODIS surface reflectivity product according to longitude and latitude and the date of detection data;
step 303, extracting satellite load ultraviolet near-edge detection data according to longitude and latitude and the date of the detection data;
step 304, according to steps 301-303, obtaining solar zenith angle, azimuth angle, view angle, observation angle, cut-up sequence, earth surface reflectivity, atmospheric absorption profile and slit function of input satellite load, as input data of a deep neural network model, and according to the deep neural network model, calculating solar spectrum at the satellite entrance pupil.
The invention utilizes a Deep Neural Network (DNN) method to construct a complex nonlinear model between the input and the output of a radiation transmission model; based on the critical radiation transmission simulation function of the radiation transmission model SCITRA, forward radiation transmission simulation is carried out at a height cutting interval of 150m under the condition of different solar zenith angles, earth surface reflectivities and instrument slit functions and under the condition of the American standard atmosphere, and ultraviolet critical radiation spectrums under different conditions are obtained; the input and output of the radiation transmission model are training data, model parameter training is carried out, and a DNN model is constructed, wherein the input data comprise longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, satellite azimuth angle, earth surface reflectivity, instrument slit function, atmosphere background library and cut-height sequence which are used as the input layer of the neural network, and solar radiation spectrums on different cut heights received by the edge detector are used as the output layer of the neural network.
While the foregoing is directed to embodiments of the present invention, other and further details of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.

Claims (8)

1. The method for constructing the fast radiation transmission model based on the deep neural network is characterized by comprising the following steps of:
step 101, simulating a near-edge ultraviolet spectrum data set based on an atmospheric radiation transmission model; the near-edge ultraviolet spectrum data set comprises the following data: longitude and latitude, date, solar zenith angle, observation angle, azimuth angle, earth surface reflectivity, field angle, instrument slit function, atmospheric background library, height cutting sequence and solar radiation received by a limb detector at different heights;
step 102, training the near-edge ultraviolet spectrum data set serving as a training sample through a neural network to obtain a deep learning model; the device comprises a longitudinal and latitudinal direction, a date, a solar zenith angle, an observation angle, an azimuth angle, an earth surface reflectivity, an angle of view, an instrument slit function, an atmospheric background library and a height cutting sequence, which are used as input layers of a neural network, and solar radiation with different height cutting received by a boundary detector is used as output layers of the neural network;
and 103, extracting longitude and latitude, time, solar zenith angle, observation angle, azimuth angle and height cutting sequence of a first-class product of the satellite limb load in real time, and calculating to obtain solar radiation with different height cutting received by a limb detector by adopting the deep learning model.
2. The method for constructing a fast radiation transmission model based on a deep neural network according to claim 1, wherein in the step 101, an atmospheric radiation transmission model is a SCIATRAN forward radiation transmission model; the atmosphere background library adopts the American standard atmosphere.
3. The method for constructing a fast radiation transmission model based on a deep neural network according to claim 1, wherein in the step 103, the ERA5 analysis data and the atmospheric profile of the AURAMLS secondary product are extracted as a real-time atmospheric background library, and the MODIS secondary surface reflectivity product is extracted as a real-time surface reflectivity.
4. The method for constructing a fast radiation transmission model based on a deep neural network according to claim 1, wherein the near-edge ultraviolet spectrum range is 300nm-500nm, and the spectrum interval is 0.4nm; the value range of the cut-up sequence is 5-65km, and the set interval is 100m.
5. A method of constructing a fast radiation delivery model based on a deep neural network according to any one of claims 1-4, wherein the neural network comprises 1 input layer, 1 output layer, 3 hidden layers; the 3 hidden layers contain 256, 128, 64 neurons, respectively.
6. The method for constructing a fast radiation transmission model based on a deep neural network according to claim 1, wherein the longitude and latitude of the near-edge ultraviolet spectrum dataset take values of 90 ° S-90 ° N at intervals of 5 ° and take any one of 90 ° N, 85 ° N, 80 ° N, 75 ° N, 70 ° N, 65 ° N, 60 ° N, 55 ° N, 50 ° N, 45 ° N, 40 ° N, 35 ° N, 30 ° N, 25 ° N, 20 ° N, 15 ° N, 10 ° N, 5 ° N, 0 °,5 ° S, 10 ° S, 15 ° S, 20 ° S, 25 ° S, 30 ° S, 35 ° S, 40 ° S, 45 ° S, 50 ° S, 55 ° S, 60 ° S, 65 ° S, 70 ° S, 75 ° S, 80 ° S, 85 ° S, 90 ° S; taking the longitude interval 180 DEG W-180 DEG E as a value at intervals of 30 DEG, and taking any one of 180 DEG E, 150 DEG E, 120 DEG E, 90 DEG E, 60 DEG E, 30 DEG E, 0 DEG, 30 DEG W, 60 DEG W, 120 DEG W, 150 DEG W and 180 DEG W respectively.
7. The method for constructing a fast radiation transmission model based on a deep neural network according to claim 6, wherein the solar zenith angles are respectively any one of 0 °,5 °, 10 °, 15 °, 20 °, 25 °, 30 °, 35 °, 40 °, 45 °, 50 °, 55 °, 60 °, 70 °, 80 ° and 90 ° at intervals of 5 °; the observation angle is calculated and set according to the positions of longitude and latitude, and the setting interval is 2 degrees; the angle of view is set according to the instrument, the setting interval is 1, and the earth's surface reflectivity value is any one of 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and 1.0.
8. The method for constructing a fast radiation delivery model based on a deep neural network of claim 5,
deep neural network learning is performed using a pytorch framework.
CN202310337331.3A 2023-03-31 2023-03-31 Method for constructing rapid radiation transmission model based on deep neural network Pending CN116384241A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310337331.3A CN116384241A (en) 2023-03-31 2023-03-31 Method for constructing rapid radiation transmission model based on deep neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310337331.3A CN116384241A (en) 2023-03-31 2023-03-31 Method for constructing rapid radiation transmission model based on deep neural network

Publications (1)

Publication Number Publication Date
CN116384241A true CN116384241A (en) 2023-07-04

Family

ID=86962893

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310337331.3A Pending CN116384241A (en) 2023-03-31 2023-03-31 Method for constructing rapid radiation transmission model based on deep neural network

Country Status (1)

Country Link
CN (1) CN116384241A (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110175375A (en) * 2019-05-13 2019-08-27 中国科学院遥感与数字地球研究所 A kind of earth's surface Calculation method for solar radiation based on deep learning
CN115438566A (en) * 2022-07-27 2022-12-06 南京航空航天大学 Atmospheric radiation transmission model simulation method based on full-connection and RNN neural network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110175375A (en) * 2019-05-13 2019-08-27 中国科学院遥感与数字地球研究所 A kind of earth's surface Calculation method for solar radiation based on deep learning
CN115438566A (en) * 2022-07-27 2022-12-06 南京航空航天大学 Atmospheric radiation transmission model simulation method based on full-connection and RNN neural network

Similar Documents

Publication Publication Date Title
Karger et al. CHELSA-TraCE21k–high-resolution (1 km) downscaled transient temperature and precipitation data since the Last Glacial Maximum
Viatte et al. Methane emissions from dairies in the Los Angeles Basin
CN110174359B (en) Aviation hyperspectral image soil heavy metal concentration assessment method based on Gaussian process regression
Pereira et al. Development of an ANN based corrective algorithm of the operational ECMWF global horizontal irradiation forecasts
CN105988146A (en) Application data processing method of spaceborne microwave radiometer
CN109387487A (en) Short-wave infrared high-spectral data atmospheric methane fast inversion method under the conditions of cirrus
CN102507586B (en) Remote sensing monitoring method for carbon emission
CN111579504A (en) Atmospheric pollution component vertical distribution inversion method based on optical remote sensing
CN111257241A (en) Atmospheric carbon dioxide concentration inversion algorithm based on DEEI (DeEI)
CN112800603A (en) Atmospheric environment data assimilation method based on set optimal interpolation algorithm
CN105784624A (en) Retrieval method and device for water vapor profile
Sadeghi et al. Contributions of meteorology to ozone variations: Application of deep learning and the Kolmogorov-Zurbenko filter
CN116297288B (en) Rapid remote sensing inversion method and system for atmospheric methane dry air mixing ratio
CN103743679B (en) A kind of method accelerating greenhouse gases vertical column retrieving concentration speed
Bao et al. Assessing the impact of Chinese FY-3/MERSI AOD data assimilation on air quality forecasts: Sand dust events in northeast China
Lin et al. A coupled experiment with LICOM2 as the ocean component of CESM1
CN111879915A (en) High-resolution monthly soil salinity monitoring method and system for coastal wetland
Lu et al. Prediction of diffuse solar radiation by integrating radiative transfer model and machine-learning techniques
Morales et al. Controlled-release experiment to investigate uncertainties in UAV-based emission quantification for methane point sources
Joyce et al. Using a deep neural network to detect methane point sources and quantify emissions from PRISMA hyperspectral satellite images
CN113740263A (en) Aerosol optical thickness inversion method and atmospheric particulate matter remote sensing inversion method
CN117523406A (en) Clamping correction method for ocean water color and temperature scanner and computer readable medium
CN116384241A (en) Method for constructing rapid radiation transmission model based on deep neural network
WO2023229705A1 (en) System and method for automatically estimating gas emission parameters
CN116167003A (en) Near-ground artificial source nitrogen dioxide high-definition product estimation method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination