CN111737912B - MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network - Google Patents
MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network Download PDFInfo
- Publication number
- CN111737912B CN111737912B CN202010543078.3A CN202010543078A CN111737912B CN 111737912 B CN111737912 B CN 111737912B CN 202010543078 A CN202010543078 A CN 202010543078A CN 111737912 B CN111737912 B CN 111737912B
- Authority
- CN
- China
- Prior art keywords
- data set
- mwhts
- neural network
- cloud
- deep neural
- 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.)
- Active
Links
- 238000004364 calculation method Methods 0.000 title claims abstract description 70
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 38
- 239000002184 metal Substances 0.000 title claims abstract description 6
- 238000012795 verification Methods 0.000 claims abstract description 75
- 238000004458 analytical method Methods 0.000 claims abstract description 46
- 230000005540 biological transmission Effects 0.000 claims abstract description 42
- 230000005855 radiation Effects 0.000 claims abstract description 42
- 238000003062 neural network model Methods 0.000 claims abstract description 41
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 238000003908 quality control method Methods 0.000 claims description 6
- 238000000034 method Methods 0.000 abstract description 6
- 230000003993 interaction Effects 0.000 abstract description 2
- 238000004088 simulation Methods 0.000 abstract description 2
- 238000001514 detection method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000004001 molecular interaction Effects 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004291 sulphur dioxide Substances 0.000 description 1
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
- 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
- G06N3/045—Combinations of networks
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Radiation Pyrometers (AREA)
Abstract
The MWHTS simulated bright temperature calculation method based on the deep neural network comprises the steps of establishing a matching data set of the MWHTS observed bright temperature and the climatology data set in space and time; dividing the matched data set into a clear sky data set, a cloud data set and a rain data set according to the cloud water content, and respectively forming a corresponding analysis data set and a corresponding verification data set; training a deep neural network model by using three analysis data sets, inputting atmospheric parameters in a corresponding verification data set into the trained deep neural network model, and calculating MWHTS (metal wrap through temperature) simulation bright temperature; the atmospheric parameters in the three verification data sets are input into a radiation transmission model to calculate the MWHTS simulated bright temperature, the calculation accuracy is compared with the MWHTS simulated bright temperature calculation accuracy based on the deep neural network, and the MWHTS channels with higher accuracy are selected to form the MWHTS simulated bright temperature calculation result. The method uses the deep neural network to model the interaction between microwaves and atmospheric molecules, so that the calculation accuracy is higher than that of the RTTOV of the business radiation transmission model, and the operation is simple and easy.
Description
Technical Field
The invention relates to a calculation method of MWHTS simulated bright temperature, belongs to the technical field of microwave remote sensing, and particularly relates to a calculation method of MWHTS simulated bright temperature based on a deep neural network.
Background
In the field of microwave remote sensing, problems to be processed can be classified into forward modeling and inversion. For remote sensing of satellite-borne microwaves, forward modeling is used for calculating the observed bright temperature of a satellite-borne microwave radiometer by modeling the radiation transmission process of microwaves in the atmosphere, and inversion is usually used for carrying out mathematical calculation on the inversion of a radiation transmission model, namely, the process of acquiring atmospheric parameters by utilizing the observed bright temperature of the satellite-borne microwave radiometer. The radiation transmission model of microwaves in the atmosphere is not only a theoretical basis for developing satellite payloads, but also a key point of detecting earth atmospheric temperature and humidity parameters, cloud and rain parameters and earth surface parameters by a satellite-borne microwave remote sensing instrument, and the calculation accuracy of the radiation transmission model on simulating the bright temperature is directly related to inversion application of satellite observation bright temperature.
At present, the radiation transmission model for business is developed based on the physical mechanism of the effects of emission, absorption, scattering and the like of microwaves in the atmosphere, wherein the RTTOV of the radiation transmission model is a typical representative of the radiation transmission model for business, and the RTTOV is widely applied to satellite data inversion and assimilation systems. The current lack of understanding of microwave and atmospheric molecular interactions is a major contributor to radiation transmission model errors. Especially under the cloud and rain atmospheric conditions, the nonlinearity of the radiation transmission equation is increased, and the modeling difficulty of the radiation transmission model on the cloud and rain scattering effect is high, so that the calculation accuracy of the simulation of the brightness and temperature is poor. However, aiming at the nonlinear relation between the atmospheric parameters and the observed bright temperature of the satellite-borne microwave radiometer, the strong nonlinear mapping capability of the deep neural network can provide a new thought for the calculation of the simulated bright temperature of the satellite-borne microwave radiometer.
The microwave humidity temperature detector (MWHTS) is an important load on a wind cloud No. three (FY-3) C star and a D star, is a microwave radiometer integrating a hygrometer and a thermometer at the first stage of the world, and is provided with eight temperature detection channels (channels 2-9), five humidity detection channels (channels 11-15) and two window channels (channels 1 and 10), so that the simultaneous detection of the atmospheric temperature and the water vapor parameters can be realized. The calculation of the MWHTS simulated bright temperature is very important for the application of the data in the fields of numerical weather forecast, current weather stability analysis, climate change research and the like.
Disclosure of Invention
In order to solve the technical problems, the invention provides the MWHTS simulated bright temperature calculation method based on the deep neural network, the interaction between microwaves and atmospheric molecules is modeled by using the deep neural network, the calculation accuracy higher than that of the RTTOV of the business radiation transmission model can be obtained, and the operation is simple and easy.
In order to achieve the technical purpose, the adopted technical scheme is as follows: the MWHTS simulated bright temperature calculating method based on the deep neural network comprises the following steps of:
step one: establishing a matching data set of the MWHTS observation brightness temperature and the atmospheric parameters of the climatology data set in time and space;
step two: dividing the matched data set established in the first step into a clear sky data set, a cloud data set and a rain data set according to the cloud water content, and respectively forming a corresponding clear sky analysis data set and a clear sky verification data set, a cloud analysis data set and a cloud verification data set, a rain analysis data set and a rain verification data set;
step three: respectively training a deep neural network model by utilizing the clear sky analysis data set, the cloud analysis data set and the rain analysis data set formed in the second step, respectively inputting atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rain verification data set into the trained corresponding deep neural network models, and calculating the MWHTS simulated bright temperature;
step four: and (3) respectively inputting the clear sky verification data set, the cloud verification data set and the atmospheric parameters in the rain verification data set into a radiation transmission model RTTOV to calculate the MWHTS simulated bright temperature, comparing the calculation precision of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV with the calculation precision of the MWHTS simulated bright temperature based on the deep neural network, and selecting the simulated bright temperature in the MWHTS channel with higher precision to form the calculation result of the MWHTS simulated bright temperature.
The first step specifically comprises: the first selected atmospheric parameters in the climatology dataset are: temperature profile, humidity profile, cloud water profile, surface temperature, surface humidity, surface pressure, 10m wind speed, and cloud water content; then matching the atmospheric parameters in the climatic data set with the MWHTS observation bright temperature in time and space according to a matching rule that the time error is less than 10 minutes and the longitude and latitude error is less than 0.1 degrees; and finally, performing quality control on the matching data, namely deleting the matching data with abnormal values in the MWHTS observation brightness and climatic data set, wherein the MWHTS brightness Wen Zhixiao is 180K or more than 310K, and the negative values in the climatic data set are all judged to be abnormal values, and the matching data after quality control form a matching data set.
The second step specifically comprises: in the matching data set formed in the step one, selecting matching data with a cloud water content value of zero as a clear sky data set, randomly selecting 80% of the matching data in the clear sky data set to form a clear sky analysis data set, and forming a clear sky verification data set by the remaining matching data; selecting matching data with cloud water content greater than zero and less than 0.5mm from the matching data set formed in the first step as a cloud data set, randomly selecting 80% of the matching data in the cloud data set to form a cloud analysis data set, and forming a cloud verification data set from the rest of the matching data; and in the matching data set formed in the step one, selecting matching data with cloud water content greater than or equal to 0.5mm as a rainy data set, randomly selecting 80% of the matching data in the rainy data set to form a rain analysis data set, and forming a rain verification data set on the rest matching data.
The third step specifically comprises: firstly, taking atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rain verification data set as input of a deep neural network model, taking MWHTS observation bright temperature as output of the deep neural network model, respectively training the deep neural network model by utilizing a clear sky analysis data set, a cloud analysis data set and a rain analysis data set to respectively obtain a clear sky deep neural network model, a cloud deep neural network model and a rain deep neural network model; then, respectively inputting the atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rainy verification data set into a corresponding clear sky deep neural network model, a cloud deep neural network model and a rainy deep neural network model to obtain corresponding MWHTS simulated bright temperatures under clear sky, cloud and rainy atmospheric conditions; and finally, respectively calculating root mean square errors between the MWHTS simulated bright temperatures and the observed bright temperatures under the atmospheric conditions of clear sky, clouds and rainy weather, and respectively serving as calculation precision of the MWHTS simulated bright temperatures based on the deep neural network under the atmospheric conditions of clear sky, clouds and rainy weather.
The fourth step specifically comprises: firstly, selecting a radiation transmission model RTTOV based on physical mechanism modeling as a radiation transmission model for calculating MWHTS (metal wrap through temperature) simulated bright temperature; then, respectively inputting atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rainy verification data set into a radiation transmission model RTTOV to calculate MWHTS simulated bright temperatures, and respectively obtaining the calculation accuracy of the MWHTS simulated bright temperatures based on the radiation transmission model RTTOV under clear sky, cloud and rainy atmospheric conditions; finally, the calculation precision of the MWHTS simulated bright temperature based on the deep neural network under the atmospheric conditions of clear sky, clouds and rains is respectively compared with the calculation precision of the MWHTS simulated bright temperature based on the RTTOV of the radiation transmission model under the corresponding atmospheric conditions, and the simulated bright temperature in the MWHTS channel with higher precision is selected to form the final result of the MWHTS simulated bright temperature.
The invention has the beneficial effects that: the invention aims to overcome the difficulty of inaccurate physical modeling when microwaves are transmitted in the atmosphere and improve the calculation precision of the MWHTS simulated bright temperature, adopts a deep neural network to model the nonlinear relation between atmospheric parameters and the MWHTS observed bright temperature, further compares the calculation precision of the MWHTS simulated bright temperature of a radiation transmission model RTTOV based on a physical method with the calculation precision of the MWHTS simulated bright temperature based on the deep neural network, and preferentially selects the simulated bright temperature with higher precision in the MWHTS channel to form the final result of the MWHTS simulated bright temperature. The method has higher MWHTS simulated bright temperature calculation precision, and is simple and easy to operate.
Drawings
FIG. 1 is a flow chart of a method for calculating the simulated bright temperature of the MWHTS based on the deep neural network;
fig. 2 is a graph comparing the calculation accuracy of MWHTS simulated bright temperature based on deep neural network with the calculation accuracy of MWHTS simulated bright temperature based on radiation transmission model RTTOV under clear air condition in practical example 1 of the present invention;
FIG. 3 is a graph showing the comparison of the calculation accuracy of the MWHTS simulated bright temperature based on the deep neural network and the calculation accuracy of the MWHTS simulated bright temperature based on the RTTOV in practical example 1 of the invention under the cloud atmosphere condition;
fig. 4 is a graph showing the comparison between the calculation accuracy of the MWHTS simulated bright temperature based on the deep neural network and the calculation accuracy of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV in practical example 1 of the present invention under the rainy atmosphere condition.
Detailed Description
The MWHTS simulated bright temperature calculating method based on the deep neural network comprises the following steps of:
step one: establishing a matching data set of the MWHTS observation brightness temperature and the atmospheric parameters of the climatology data set in time and space;
the first step specifically comprises: the first selected atmospheric parameters in the climatology dataset are: temperature profile, humidity profile, cloud water profile, surface temperature, surface humidity, surface pressure, 10m wind speed, and cloud water content; then matching the atmospheric parameters in the climatic data set with the MWHTS observation bright temperature in time and space according to a matching rule that the time error is less than 10 minutes and the longitude and latitude error is less than 0.1 degrees; and finally, performing quality control on the matching data, namely deleting the matching data with abnormal values in the MWHTS observation brightness and climatic data set, wherein the MWHTS brightness Wen Zhixiao is 180K or more than 310K, and the negative values in the climatic data set are all judged to be abnormal values, and the matching data after quality control form a matching data set.
Step two: dividing the matched data set established in the first step into a clear sky data set, a cloud data set and a rain data set according to the cloud water content, wherein the clear sky data set forms a corresponding clear sky analysis data set and a clear sky verification data set, the cloud data set forms a corresponding cloud analysis data set and a cloud verification data set, and the rain data set forms a corresponding rain analysis data set and a corresponding rain verification data set;
the second step specifically comprises: in the matching data set formed in the step one, selecting matching data with a cloud water content value of zero as a clear sky data set, randomly selecting 80% of the matching data in the clear sky data set to form a clear sky analysis data set, and forming a clear sky verification data set by the remaining matching data; selecting matching data with cloud water content greater than zero and less than 0.5mm from the matching data set formed in the first step as a cloud data set, randomly selecting 80% of the matching data in the cloud data set to form a cloud analysis data set, and forming a cloud verification data set from the rest of the matching data; and in the matching data set formed in the step one, selecting matching data with cloud water content greater than or equal to 0.5mm as a rainy data set, randomly selecting 80% of the matching data in the rainy data set to form a rain analysis data set, and forming a rain verification data set on the rest matching data.
Step three: respectively training a deep neural network model by utilizing the clear sky analysis data set, the cloud analysis data set and the rain analysis data set formed in the second step, respectively inputting atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rain verification data set into the trained corresponding deep neural network models, and calculating the MWHTS simulated bright temperature;
the third step specifically comprises: firstly, taking atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rain verification data set as input of a deep neural network model, taking MWHTS observation bright temperature as output of the deep neural network model, respectively training the deep neural network model by utilizing a clear sky analysis data set, a cloud analysis data set and a rain analysis data set to respectively obtain a clear sky deep neural network model, a cloud deep neural network model and a rain deep neural network model; then, respectively inputting the atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rainy verification data set into a corresponding clear sky deep neural network model, a cloud deep neural network model and a rainy deep neural network model to obtain corresponding MWHTS simulated bright temperatures under clear sky, cloud and rainy atmospheric conditions; and finally, respectively calculating root mean square errors between the MWHTS simulated bright temperatures and the observed bright temperatures under the atmospheric conditions of clear sky, clouds and rainy weather, and respectively serving as calculation precision of the MWHTS simulated bright temperatures based on the deep neural network under the atmospheric conditions of clear sky, clouds and rainy weather.
Step four: and (3) respectively inputting the clear sky verification data set, the cloud verification data set and the atmospheric parameters in the rain verification data set into a radiation transmission model RTTOV to calculate the MWHTS simulated bright temperature, comparing the calculation precision of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV with the calculation precision of the MWHTS simulated bright temperature based on the deep neural network, and selecting the simulated bright temperature in the MWHTS channel with higher precision to form the calculation result of the MWHTS simulated bright temperature.
The fourth step specifically comprises: firstly, selecting a radiation transmission model RTTOV based on physical mechanism modeling as a radiation transmission model for calculating MWHTS (metal wrap through temperature) simulated bright temperature; then, respectively inputting atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rainy verification data set into a radiation transmission model RTTOV to calculate MWHTS simulated bright temperatures, and respectively obtaining the calculation accuracy of the MWHTS simulated bright temperatures based on the radiation transmission model RTTOV under clear sky, cloud and rainy atmospheric conditions; finally, the calculation precision of the MWHTS simulated bright temperature based on the deep neural network under the atmospheric conditions of clear sky, clouds and rains is respectively compared with the calculation precision of the MWHTS simulated bright temperature based on the RTTOV of the radiation transmission model under the corresponding atmospheric conditions, and the simulated bright temperature in the MWHTS channel with higher precision is selected to form the final result of the MWHTS simulated bright temperature.
The invention will be further described with reference to examples and drawings, to which it should be noted that the examples do not limit the scope of the invention as claimed.
Example 1
The climatology data set is selected as an analysis data set ERA-Interim of the European middle weather forecast center (ECMWF), the ECMWF is used for analyzing the data set ERA-Interim to be in a time range from 9 months in 2018 to 8 months in 2019, the geographic range is (25 DEG N-45 DEG N,160 DEG E-220 DEG E), the data resolution is 0.5 DEG x 0.5 DEG, and the temperature profile, the humidity profile, the cloud water profile, the surface temperature, the surface humidity, the surface pressure, the wind speed of 10m and the cloud water content in the data set are used for calculating the simulated bright temperature. And establishing a matching data set (1060162 groups) according to a matching rule that the time error is less than 10 minutes and the longitude and latitude error is less than 0.1 degrees with FY-3D/MWHTS observation bright temperature. Selecting a clear sky data set from the matching data set according to the cloud water content of 0, and respectively establishing a clear sky analysis data set (13810 groups) and a clear sky verification data set (3453 groups); selecting a cloud data set according to the cloud water content of more than 0 and less than 0.5mm, and respectively establishing a cloud analysis data set (796457 group) and a cloud verification data set (199114 group); a rainy dataset was selected according to cloud water content greater than and equal to 0.5mm, and a rain analysis dataset (37862 group) and a rain verification dataset (9466 group) were formed, respectively.
And respectively using atmospheric parameters in the clear sky analysis data set, the cloud analysis data set and the rain analysis data set as input, and using the corresponding MWHTS observation bright temperature as output to train the deep neural network so as to obtain a clear sky deep neural network model, a cloud deep neural network model and a rain deep neural network model. And the atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rainy verification data set are respectively input into the corresponding clear sky deep neural network model, the cloud deep neural network model and the rainy deep neural network model, so that the calculation precision of the MWHTS simulated bright temperature and the calculation precision of the MWHTS simulated bright temperature under the conditions of clear sky, cloud and rainy atmosphere are respectively obtained.
And respectively inputting atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rainy verification data set into the RTTOV of the radiation transmission model, and calculating the MWHTS simulated bright temperature and the calculation accuracy of the MWHTS simulated bright temperature under the clear sky, cloud and rainy atmospheric conditions. The calculation accuracy of the MWHTS simulated bright temperature based on the deep neural network and the calculation accuracy of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV are respectively shown in the figures 2, 3 and 4 in comparison with the calculation accuracy under the clear air condition, the cloud air condition and the rainy air condition.
Under the condition of clear air atmosphere, as can be seen from fig. 2, the calculation accuracy of the MWHTS simulated bright temperature based on the deep neural network is better than that of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV in all 15 channels of the MWHTS, and particularly, the calculation accuracy of the MWHTS channels 1 to 6, 9, 10 and 15 is improved remarkably, and the maximum can be improved by 2.4K.
Under the cloud atmosphere condition, as can be seen from fig. 3, the calculation accuracy of the MWHTS simulated bright temperature based on the deep neural network is obviously improved in the other channels than the RTTOV except that the calculation accuracy of the MWHTS channels 11 and 13 is slightly higher than the calculation accuracy of the RTTOV of the radiation transmission model.
Under rainy atmospheric conditions, it can be seen from fig. 4 that the calculation accuracy of the deep neural network-based calculation method is significantly improved in the window channels 1 and 10 compared to RTTOV, while the calculation accuracy in the channels 7, 8, 9 and 15 is significantly improved, whereas the calculation accuracy in the channels 4 and 5 is not as good as RTTOV.
In summary, when the conditions of clear sky and cloudy atmosphere are met, the calculation result of the MWHTS based on the deep neural network is selected as the simulated bright temperature of the MWHTS; in rainy atmospheric conditions, MWHTS channels 4 and 5 select the calculation of simulated bright temperatures using the radiation transmission model RTTOV, while the remaining channels use the calculation of simulated bright temperatures of MWHTS based on deep neural networks.
Claims (5)
1. The MWHTS simulated bright temperature calculation method based on the deep neural network is characterized by comprising the following steps of:
step one: establishing a matching data set of the MWHTS observation brightness temperature and the atmospheric parameters of the climatology data set in time and space;
step two: dividing the matched data set established in the first step into a clear sky data set, a cloud data set and a rain data set according to the cloud water content, and respectively forming a corresponding clear sky analysis data set and a clear sky verification data set, a cloud analysis data set and a cloud verification data set, a rain analysis data set and a rain verification data set;
step three: respectively training a deep neural network model by utilizing the clear sky analysis data set, the cloud analysis data set and the rain analysis data set formed in the second step, respectively inputting atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rain verification data set into the trained corresponding deep neural network models, and calculating the MWHTS simulated bright temperature;
step four: and (3) respectively inputting the clear sky verification data set, the cloud verification data set and the atmospheric parameters in the rain verification data set into a radiation transmission model RTTOV to calculate the MWHTS simulated bright temperature, comparing the calculation precision of the MWHTS simulated bright temperature based on the radiation transmission model RTTOV with the calculation precision of the MWHTS simulated bright temperature based on the deep neural network, and selecting the simulated bright temperature in the MWHTS channel with higher precision to form the calculation result of the MWHTS simulated bright temperature.
2. The MWHTS simulated bright temperature computing method based on deep neural network of claim 1, characterized in that the step one specifically includes:
the first selected atmospheric parameters in the climatology dataset are: temperature profile, humidity profile, cloud water profile, surface temperature, surface humidity, surface pressure, 10m wind speed, and cloud water content; then matching the atmospheric parameters in the climatic data set with the MWHTS observation bright temperature in time and space according to a matching rule that the time error is less than 10 minutes and the longitude and latitude error is less than 0.1 degrees; and finally, performing quality control on the matching data, namely deleting the matching data with abnormal values in the MWHTS observation brightness and climatic data set, wherein the MWHTS brightness Wen Zhixiao is 180K or more than 310K, and the negative values in the climatic data set are all judged to be abnormal values, and the matching data after quality control form a matching data set.
3. The MWHTS simulated bright temperature computing method based on deep neural network of claim 1, characterized in that the second step specifically comprises:
in the matching data set formed in the step one, selecting matching data with a cloud water content value of zero as a clear sky data set, randomly selecting 80% of the matching data in the clear sky data set to form a clear sky analysis data set, and forming a clear sky verification data set by the remaining matching data; selecting matching data with cloud water content greater than zero and less than 0.5mm from the matching data set formed in the first step as a cloud data set, randomly selecting 80% of the matching data in the cloud data set to form a cloud analysis data set, and forming a cloud verification data set from the rest of the matching data; and in the matching data set formed in the step one, selecting matching data with cloud water content greater than or equal to 0.5mm as a rainy data set, randomly selecting 80% of the matching data in the rainy data set to form a rain analysis data set, and forming a rain verification data set on the rest matching data.
4. The MWHTS simulated bright temperature computing method based on deep neural network of claim 1, characterized in that the third step specifically includes:
firstly, taking atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rain verification data set as input of a deep neural network model, taking MWHTS observation bright temperature as output of the deep neural network model, respectively training the deep neural network model by utilizing a clear sky analysis data set, a cloud analysis data set and a rain analysis data set to respectively obtain a clear sky deep neural network model, a cloud deep neural network model and a rain deep neural network model; then, respectively inputting the atmospheric parameters in the clear sky verification data set, the cloud verification data set and the rainy verification data set into a corresponding clear sky deep neural network model, a cloud deep neural network model and a rainy deep neural network model to obtain corresponding MWHTS simulated bright temperatures under clear sky, cloud and rainy atmospheric conditions; and finally, respectively calculating root mean square errors between the MWHTS simulated bright temperatures and the observed bright temperatures under the atmospheric conditions of clear sky, clouds and rainy weather, and respectively serving as calculation precision of the MWHTS simulated bright temperatures based on the deep neural network under the atmospheric conditions of clear sky, clouds and rainy weather.
5. The MWHTS simulated bright temperature computing method based on deep neural network of claim 1, characterized in that the fourth step specifically includes:
firstly, selecting a radiation transmission model RTTOV based on physical mechanism modeling as a radiation transmission model for calculating MWHTS (metal wrap through temperature) simulated bright temperature; then, respectively inputting atmospheric parameters in a clear sky verification data set, a cloud verification data set and a rainy verification data set into a radiation transmission model RTTOV to calculate MWHTS simulated bright temperatures, and respectively obtaining the calculation accuracy of the MWHTS simulated bright temperatures based on the radiation transmission model RTTOV under clear sky, cloud and rainy atmospheric conditions; finally, the calculation precision of the MWHTS simulated bright temperature based on the deep neural network under the atmospheric conditions of clear sky, clouds and rains is respectively compared with the calculation precision of the MWHTS simulated bright temperature based on the RTTOV of the radiation transmission model under the corresponding atmospheric conditions, and the simulated bright temperature in the MWHTS channel with higher precision is selected to form the final result of the MWHTS simulated bright temperature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010543078.3A CN111737912B (en) | 2020-06-15 | 2020-06-15 | MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010543078.3A CN111737912B (en) | 2020-06-15 | 2020-06-15 | MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111737912A CN111737912A (en) | 2020-10-02 |
| CN111737912B true CN111737912B (en) | 2024-01-23 |
Family
ID=72649196
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010543078.3A Active CN111737912B (en) | 2020-06-15 | 2020-06-15 | MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111737912B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329334B (en) * | 2020-10-15 | 2024-01-23 | 洛阳师范学院 | MWHTS and MWTS-II fusion inversion sea surface air pressure method based on simulated bright temperature |
| CN112345151B (en) * | 2020-10-15 | 2022-01-07 | 洛阳师范学院 | Sensitivity test method of MWTS-II to sea surface air pressure based on natural atmosphere |
| CN113311510B (en) * | 2021-05-11 | 2023-07-25 | 洛阳师范学院 | MWHTS (metal wrap through) observation bright temperature classification method based on simulated bright temperature |
| CN113311509B (en) * | 2021-05-11 | 2023-07-25 | 洛阳师范学院 | Sensitivity test method of MWHTS (metal wrap through) to sea surface air pressure based on neural network |
| CN113435119B (en) * | 2021-06-29 | 2022-08-05 | 北京华云星地通科技有限公司 | Global ocean surface brightness temperature determination method and system |
| CN114252834B (en) * | 2021-12-23 | 2022-08-12 | 国家卫星海洋应用中心 | Satellite-borne microwave radiometer external calibration method and device based on ocean target |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108919151A (en) * | 2018-04-03 | 2018-11-30 | 西安空间无线电技术研究所 | A kind of microwave radiometer cross polarization error correcting method |
| CN109580003A (en) * | 2018-12-18 | 2019-04-05 | 成都信息工程大学 | A kind of stationary weather satellite Thermal Infrared Data estimation Air Close To The Earth Surface temperature methods |
| CN110826693A (en) * | 2019-10-29 | 2020-02-21 | 华中科技大学 | Three-dimensional atmospheric temperature profile inversion method and system based on DenseNet convolutional neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003036330A2 (en) * | 2001-10-22 | 2003-05-01 | Coppola Emery A Jr | Neural network based predication and optimization for groundwater / surface water system |
-
2020
- 2020-06-15 CN CN202010543078.3A patent/CN111737912B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108919151A (en) * | 2018-04-03 | 2018-11-30 | 西安空间无线电技术研究所 | A kind of microwave radiometer cross polarization error correcting method |
| CN109580003A (en) * | 2018-12-18 | 2019-04-05 | 成都信息工程大学 | A kind of stationary weather satellite Thermal Infrared Data estimation Air Close To The Earth Surface temperature methods |
| CN110826693A (en) * | 2019-10-29 | 2020-02-21 | 华中科技大学 | Three-dimensional atmospheric temperature profile inversion method and system based on DenseNet convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111737912A (en) | 2020-10-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111737912B (en) | MWHTS (metal wrap through) simulated bright temperature calculation method based on deep neural network | |
| CN111737913B (en) | MWHTS clear sky observation bright temperature selection method based on cloud water content inversion | |
| CN109580003B (en) | Method for estimating near-ground atmospheric temperature by thermal infrared data of stationary meteorological satellite | |
| CN108875254B (en) | One-dimensional variation inversion method of atmospheric temperature wet profile | |
| CN110174359B (en) | Aviation hyperspectral image soil heavy metal concentration assessment method based on Gaussian process regression | |
| CN111579504B (en) | Atmospheric pollution component vertical distribution inversion method based on optical remote sensing | |
| CN103413036B (en) | Continuous forest fire weather level forecasting method | |
| CN110595968B (en) | PM2.5 concentration estimation method based on geostationary orbit satellite | |
| CN112329334B (en) | MWHTS and MWTS-II fusion inversion sea surface air pressure method based on simulated bright temperature | |
| CN112163375A (en) | Long-time sequence near-surface ozone inversion method based on neural network | |
| CN111737641B (en) | MWHTS channel weight function calculation method based on neural network | |
| CN113108918B (en) | Method for inverting air temperature by using thermal infrared remote sensing data of polar-orbit meteorological satellite | |
| CN115062527B (en) | Geostationary satellite sea temperature inversion method and system based on deep learning | |
| CN117933095B (en) | Earth surface emissivity real-time inversion and assimilation method based on machine learning | |
| CN114004163A (en) | PM2.5 inversion method based on MODIS and long-and-short-term memory network model | |
| CN114880933A (en) | Atmospheric temperature and humidity profile inversion method and system for non-exploration-site foundation microwave radiometer based on reanalysis data | |
| CN115859789A (en) | Method for improving inversion accuracy of polar atmosphere temperature profile | |
| CN113111529B (en) | Infrared bright temperature simulation method integrating numerical mode and satellite microwave cloud inversion data | |
| CN113552080B (en) | Real-time inversion algorithm for ultra-spectrum remote sensing earth surface atmosphere relative humidity | |
| CN112345151B (en) | Sensitivity test method of MWTS-II to sea surface air pressure based on natural atmosphere | |
| CN110968926A (en) | Method for predicting atmospheric parameters based on improved background error covariance matrix | |
| CN113311509B (en) | Sensitivity test method of MWHTS (metal wrap through) to sea surface air pressure based on neural network | |
| CN116699729A (en) | Method and device for directly homogenizing foundation microwave radiometer data | |
| CN114757103B (en) | Earth surface O based on space-time light gradient intensifier 3 Concentration estimation method | |
| Salazar et al. | Estimation of daily average values of the Ångström turbidity coefficient β using a Corrected Yang Hybrid Model |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |