CN116756560A - Dust aerosol recognition method, system, model training method, medium and equipment - Google Patents
Dust aerosol recognition method, system, model training method, medium and equipment Download PDFInfo
- Publication number
- CN116756560A CN116756560A CN202310547308.7A CN202310547308A CN116756560A CN 116756560 A CN116756560 A CN 116756560A CN 202310547308 A CN202310547308 A CN 202310547308A CN 116756560 A CN116756560 A CN 116756560A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- dust
- aerosol
- clear sky
- 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
Links
- 239000000428 dust Substances 0.000 title claims abstract description 135
- 239000000443 aerosol Substances 0.000 title claims abstract description 124
- 238000012549 training Methods 0.000 title claims abstract description 106
- 238000000034 method Methods 0.000 title claims abstract description 62
- 239000004576 sand Substances 0.000 claims abstract description 56
- 238000004088 simulation Methods 0.000 claims abstract description 24
- 238000009529 body temperature measurement Methods 0.000 claims abstract description 21
- 238000004458 analytical method Methods 0.000 claims description 19
- 238000003384 imaging method Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 2
- 238000005282 brightening Methods 0.000 claims 2
- 230000008569 process Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 12
- 230000003287 optical effect Effects 0.000 description 10
- 230000000694 effects Effects 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 239000008264 cloud Substances 0.000 description 3
- 241000208818 Helianthus Species 0.000 description 2
- 235000003222 Helianthus annuus Nutrition 0.000 description 2
- CBENFWSGALASAD-UHFFFAOYSA-N Ozone Chemical compound [O-][O+]=O CBENFWSGALASAD-UHFFFAOYSA-N 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001747 exhibiting effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention provides a sand aerosol recognition method, a system, a model training method, a medium and equipment, wherein the training method of the sand aerosol recognition model comprises the steps of obtaining training thermal infrared appearance brightness temperature measurement data, training simulation clear sky brightness temperature data and training ground surface information data; acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data; and training a dust aerosol recognition model based on the training thermal infrared appearance brightness temperature data, the training simulation clear sky brightness temperature data, the training ground surface information data, the clear sky tag, the cloud tag and the dust tag. According to the invention, the influence of cloud and ground surface can be eliminated, the thin sand aerosol can be accurately identified under the all-weather condition, an additional cloud detection program is not needed as pretreatment, the steps of the sand identification process are reduced, the precision of sand identification is improved, and the distinction of clear sky, cloud and sand is effectively realized.
Description
Technical Field
The invention relates to an aerosol recognition method, in particular to a sand aerosol recognition method, a sand aerosol recognition system, a model training method, a medium and equipment.
Background
The dust aerosol is a natural source aerosol, and can influence the global climate and ecosystem by influencing the atmospheric radiation balance, causing cloud pollution and generating geographic biochemical effect in the transmission process. Dust aerosols typically occur in arid and semiarid regions, and are more common in asia in spring. Satellite observation enables dynamic monitoring of dust with large scale and high space-time resolution. Since the dust aerosol has similar spectral signals to other aerosols, clouds, bright ground surfaces (barren lands or deserts), identification of the dust aerosol is extremely challenging. In the former studies, many dust recognition algorithms were proposed, and the population was divided into a physics-based algorithm and a machine learning-based algorithm.
Physical-based dust identification algorithms typically use the bright temperature difference of a thermal infrared channel or a specific dust index to identify dust, such as dust exhibiting negative values at bright temperature differences of 10 μm and 12 μm. A classical physical-based dust identification algorithm uses a 10 μm and 11 μm bright temperature difference to discriminate dust, considering that the bright temperature difference of dust under the channel combination is usually 1-5K, while cloud and clear sky are negative. However, the setting of such a threshold is not deterministic and may vary from one atmosphere to another and from one surface type to another, and the threshold is not generally fixed and requires empirical determination. Other physical methods also typically require threshold settings and are more difficult to adapt to a variety of complex climatic geographic conditions.
The algorithm based on machine learning is more flexible, the recognition effects of different machine learning models are compared in the prior research, and the result is generally superior to that of a physical algorithm. However, most of these machine learning algorithms require a visible light band as an input, and cannot be used at night, which limits all-weather recognition of dust and sand. Second, these algorithms do not effectively address the difficulties in dust identification (confusing features), which are still affected by these climatic geographic features. In addition, some algorithms use aerosol optical thickness AOD products as labels to model, which can introduce errors to the identification of dust due to other aerosol effects.
Disclosure of Invention
In view of the above-mentioned drawbacks of the prior art, the present invention aims to provide a method, a system, a model training method, a medium and a device for recognizing dust aerosol, which are used for solving the technical problem that the prior art lacks a method for precisely and automatically recognizing dust aerosol under all-weather conditions.
To achieve the above and other related objects, a first aspect of the present invention provides a training method for a dust aerosol recognition model, including obtaining training thermal infrared appearance brightness temperature data, training simulation clear sky brightness temperature data, and training ground surface information data; acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data; and training a dust aerosol recognition model based on the training thermal infrared appearance brightness temperature data, the training simulation clear sky brightness temperature data, the training ground surface information data, the clear sky tag, the cloud tag and the dust tag.
In an embodiment of the first aspect, the implementation manner of obtaining training thermal infrared appearance brightness temperature measurement data, training simulation clear sky brightness temperature data and training ground surface information data includes obtaining the training thermal infrared appearance brightness temperature measurement data based on satellite observation data; and acquiring the training simulation clear sky bright temperature data and the training ground surface information data based on satellite observation data, medium resolution imaging spectrometer data and atmospheric analysis data.
In an embodiment of the first aspect, the acquiring the training simulation clear sky light temperature data based on satellite observation data, medium resolution imaging spectrometer data and atmospheric analysis data includes: acquiring observation angle data based on satellite observation data; acquiring surface emissivity data based on the medium resolution imaging spectrometer data; acquiring atmospheric profile data, surface temperature and surface pressure data based on the atmospheric analysis data; and acquiring the training simulation clear sky bright temperature data based on the atmospheric profile data, the ground surface temperature, the ground surface pressure data, the observation angle data and the ground surface emissivity data.
In an embodiment of the first aspect, the clear sky, cloud and dust labels correspond to cloud-aerosol lidar and infrared detector satellite data products.
In an embodiment of the first aspect, the acquiring clear sky labels, cloud labels, and dust labels corresponding to the training thermal infrared appearance brightness temperature data, the training simulated clear sky brightness temperature data, and the training ground surface information data includes: establishing corresponding relations among the satellite observation data, the medium resolution imaging spectrometer data, the atmospheric analysis data, the cloud-aerosol laser radar and infrared detector satellite data products based on a space-time matching principle; and acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data based on the corresponding relation.
In an embodiment of the first aspect, the cloud-aerosol lidar and infrared detector satellite data products include a vertical feature distribution data product and an aerosol profile data product.
The second aspect of the invention provides a method for identifying dust aerosol, which comprises the steps of obtaining thermal infrared appearance brightness temperature measurement data, simulated clear sky brightness temperature data and ground surface information data in an area to be identified; inputting the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data in the region to be identified into the dust aerosol identification model of the first aspect; and acquiring the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
The third aspect of the invention provides a recognition system of dust aerosol, which comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring thermal infrared appearance brightness temperature data, simulated clear sky brightness temperature data and ground surface information data in a region to be recognized; the identification module is used for inputting the thermal infrared appearance brightness temperature measurement data, the simulated clear sky brightness temperature data and the ground surface information data in the region to be identified into the dust aerosol identification model of the first aspect; and the second acquisition module is used for acquiring the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
A fourth aspect of the invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of training a recognition model of an aerosol of dust as described in the first aspect of the invention and/or the method of recognizing an aerosol of dust as described in the second aspect of the invention.
A fifth aspect of the present invention provides an electronic device, comprising: a memory storing a computer program; and the processor is in communication connection with the memory, and executes the training method of the dust aerosol recognition model according to the first aspect of the invention and/or the recognition method of the dust aerosol according to the second aspect of the invention when the computer program is called.
As described above, the method, the system, the model training method, the medium and the equipment for recognizing the dust aerosol provided by the embodiment of the invention have the following beneficial effects: the method can solve the problem of influence caused by complicated climate and geographic factors, eliminate the influence of cloud and ground surface, extract the information in the atmosphere under the all-weather condition, accurately identify the thin sand aerosol and realize the range detection of the sand area. Meanwhile, the method does not need an additional cloud detection program as pretreatment, reduces the steps of the sand and dust identification process, obviously improves the precision of sand and dust identification, and effectively realizes the distinction of clear sky, cloud and sand and dust.
Drawings
Fig. 1 is a schematic flow chart of a training method of a dust aerosol recognition model according to an embodiment of the invention.
Fig. 2 is a schematic flow chart of a training method of a sand aerosol recognition model according to an embodiment of the invention.
Fig. 3 is a schematic diagram showing a comparison of daytime recognition results of a sand aerosol recognition model and a vertical cross-section of CALIPSO according to an embodiment of the present invention.
Fig. 4 is a schematic diagram showing a comparison of daytime recognition results of a sand aerosol recognition model and a vertical cross-section of CALIPSO according to an embodiment of the present invention.
Fig. 5 is a schematic diagram showing a comparison of night recognition results of a sand aerosol recognition model with a vertical cross-sectional view of CALIPSO according to an embodiment of the present invention.
Fig. 6 is a schematic diagram showing a comparison of night recognition results of a sand aerosol recognition model with a vertical cross-sectional view of CALIPSO according to an embodiment of the present invention.
Fig. 7 is a schematic diagram showing a daytime recognition result and a daytime observation result of CALIPSO of a sand dust product according to an embodiment of the present invention.
Fig. 8 is a schematic diagram showing a daytime recognition result of the sand aerosol recognition model according to an embodiment of the present invention.
Fig. 9 is a schematic diagram showing night recognition results and night observation results of CALIPSO of a sand dust product according to an embodiment of the present invention.
Fig. 10 is a schematic diagram showing night recognition results of a sand aerosol recognition model according to an embodiment of the present invention.
Fig. 11 is a flow chart of a method for identifying dust aerosols according to an embodiment of the invention.
Fig. 12 is a schematic structural diagram of a system for recognizing a dust aerosol according to an embodiment of the present invention.
Fig. 13 is a schematic structural diagram of an electronic device according to an embodiment of the invention.
FIG. 14 is a diagram of observations in an embodiment of the present invention.
Description of element reference numerals
20. First acquisition module
30. Identification module
40. Second acquisition module
90. Electronic equipment
901. Memory device
902. Processor and method for controlling the same
903. Display device
S1 to S3 steps
S21 to S22 steps
S4 to S6 steps
Detailed Description
Other advantages and effects of the present invention will become apparent to those skilled in the art from the following disclosure, which describes the embodiments of the present invention with reference to specific examples. The invention may be practiced or carried out in other embodiments that depart from the specific details, and the details of the present description may be modified or varied from the spirit and scope of the present invention. It should be noted that the following embodiments and features in the embodiments may be combined with each other without conflict.
It should be noted that the illustrations provided in the following embodiments merely illustrate the basic concept of the present invention by way of illustration, and only the components related to the present invention are shown in the illustrations, not according to the number, shape and size of the components in actual implementation, and the form, number and proportion of each component in actual implementation may be arbitrarily changed, and the layout of the components may be more complex. Moreover, relational terms such as "first," "second," and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The invention provides a sand aerosol identification method, a sand aerosol identification system, a model training method, a medium and equipment, which can eliminate the influence of cloud and ground surface, extract information in the atmosphere under the condition of all weather, accurately identify the thin sand aerosol and realize the range detection of a sand area. Meanwhile, the method does not need an additional cloud detection program as pretreatment, reduces the steps of the sand and dust identification process, obviously improves the precision of sand and dust identification, and effectively realizes the distinction of clear sky, cloud and sand and dust.
The method, the system, the model training method, the medium and the equipment for recognizing the dust aerosol provided by the invention are described below by means of specific embodiments and with reference to the accompanying drawings.
As shown in fig. 1, in an embodiment, the training method of the dust aerosol recognition model of the present invention includes steps S1 to S4:
s1: and acquiring training thermal infrared appearance brightness temperature measurement data, training simulation clear sky brightness temperature data and training ground surface information data.
And acquiring the training thermal infrared appearance brightness temperature measurement data based on satellite observation data.
Specifically, in one embodiment, training thermal infrared appearance photometric temperature data is obtained based on Japanese sunflower No. 8 satellite thermal infrared 9 th-16 th channel photometric temperature.
And acquiring the training simulation clear sky bright temperature data and the ground surface information data based on satellite observation data, medium resolution imaging spectrometer data and atmospheric analysis data.
Specifically, in one embodiment, the simulated clear sky bright temperature data is calculated from a fast radiation transmission mode ERTM. The model uses as input the atmospheric profile (absolute humidity, temperature and ozone fraction), surface temperature, surface pressure data, japanese sunflower satellite No. 8 (AHI) observation angle data, and surface emissivity product data of the medium resolution imaging spectrometer data product (MODIS product) from ERA-5 analysis data, simulating the AHI observation bright temperature in cloudless conditions.
Specifically, the training surface information data is obtained based on medium resolution imaging spectrometer data and atmospheric analysis data.
In one embodiment, ERA-5 is used to analyze the data for surface pressure, surface temperature, and type of land cover for the MODIS product.
S2: and acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data.
Specifically, clear sky, cloud and dust labels correspond to cloud-aerosol lidar and infrared detector satellite data products (CALIPSO products). In the CALIPSO secondary product, the vertical feature distribution data product can be used for judging the type of cloud/aerosol, and the aerosol profile data product can be used for judging the optical thickness of the aerosol.
Thus, clear sky, cloud and dust labels are respectively associated with the vertical feature profile data product and the aerosol profile data product. The cloud pixels with the clear sky type of aerosol optical thickness less than or equal to 0.1, the cloud pixels with the cloud optical thickness less than or equal to 0.1 and the dust pixels with the cloud and aerosol optical thickness greater than or equal to 0.1 can be obtained.
Specifically, as shown in fig. 2, step S2 includes step S21 and step S22:
s21: and establishing corresponding relations among the satellite observation data, the medium-resolution imaging spectrometer data, the atmospheric analysis data, the cloud-aerosol laser radar and infrared detector satellite data products based on a space-time matching principle.
Specifically, the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data are obtained through the step S1 based on satellite observation data, medium resolution imaging spectrometer data and atmosphere analysis data, and the three types of labels are obtained based on CALIPSO secondary products. Therefore, in order to establish the relationship between the three kinds of labels and the training thermal infrared appearance brightness temperature data, the training simulation clear sky brightness temperature data and the training ground surface information data, step S21 firstly establishes the corresponding relationship between the satellite observation data, the medium resolution imaging spectrometer data, the atmosphere analysis data and the vertical feature distribution data product and the aerosol profile data product based on the space-time matching principle.
Specifically, the space-time matching principle is to establish a relationship of the nearest neighbor time, the nearest neighbor longitude and the nearest neighbor latitude. Step S21 will be further described below by way of an example.
First, step S21 determines the aerosol type determined by the CALIPSO vertical feature distribution data product at a certain time, a certain longitude, and a certain latitude, and the optical thickness obtained by the aerosol profile data product, so as to correspond to a clear sky label, a cloud label, or a dust label.
And secondly, step 21 acquires the most adjacent time, the most adjacent longitude, the 9 th to 16 th channel observation bright temperature of the AHI under the most adjacent dimension, the AHI observation angle data, the ground emissivity product data of the MODIS product, the ground coverage type product, the atmosphere profile (absolute humidity, temperature and ozone fraction) of ERA-5 analysis data, the ground surface temperature and the ground surface pressure data, thereby acquiring thermal infrared appearance bright temperature data, simulated clear sky bright temperature data and ground surface information data.
In step S21, the time and spatial resolution of the matching observation needs to be considered when performing space-time matching. When the AHI observation data, the MODIS data product and the ERA-5 data are matched with the label, the time difference and the position distance difference between the observation time and the observation position and the CAPIPSO observation should be smaller than the minimum time and the spatial resolution of the observation data. As shown in figure 14 of the drawings,
specifically, the AHI observed data has a temporal resolution of 10 minutes and a spatial resolution of 2km. When the AHI observation data is matched with the category labels, the time difference between the observation time of the AHI observation data and the time difference between the CAPIPSO observation is smaller than 5 minutes before and after, and the observed position distance difference is smaller than 2km in the longitude and latitude directions.
Specifically, the MODIS surface emissivity product adopted is an 8-balance average product, and the spatial resolution of the MODIS surface emissivity product is 0.05 degrees in terms of longitude and latitude. When MODIS surface emissivity data are matched with category labels, the time difference between the observation time of the surface emissivity data and the time difference between the observation time of the CAPIPSO is smaller than eight days before and after, and the observed position distance difference is smaller than 0.05 degrees in the longitude and latitude directions.
Specifically, the MODIS land cover type product is annual data, and the spatial resolution of the MODIS land cover type product is 500m. When the MODIS land cover type product is matched with the category label, the time difference between the observation time of the land cover type and the time difference between the observation time of the CALIPSO is smaller than one year before and after, and the observed position distance difference is smaller than 500m in the longitude and latitude directions.
Specifically, ERA-5 analyzes the data with a time resolution of 1 hour, wherein the spatial resolution of the atmospheric profile data is 0.25 degrees in terms of longitude and latitude, the spatial resolution of the surface temperature and pressure data is 0.25 degrees in the ocean surface, and the spatial resolution is higher in the land surface, which is 0.1 degrees. When ERA-5 analysis data are matched with category labels, the time difference between the observation time and the time difference between CALIPSO observation is smaller than 30 minutes before and after, wherein the distance difference between the observation positions of the atmospheric profile data is smaller than 0.25 degrees in the longitude and latitude directions, the distance difference between the observation positions of the earth surface temperature and the earth surface pressure data on the ocean surface is smaller than 0.25 degrees in the longitude and latitude directions, and the distance difference between the observation positions of the earth surface temperature and the earth surface pressure data on the land surface is smaller than 0.1 degrees in the longitude and latitude directions.
S22: and acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data based on the corresponding relation.
Specifically, the foregoing step S21 has established the correspondence between satellite observation data, medium resolution imaging spectrometer data, atmospheric analysis data, and vertical feature distribution data products, aerosol profile data products. Therefore, step S22 establishes the correspondence between the nearest neighbor time, the nearest neighbor longitude, the thermal infrared appearance brightness temperature data in the nearest neighbor dimension, the simulated clear sky brightness temperature data, the surface information data and the three kinds of tags.
S3: and training a dust aerosol recognition model based on the training thermal infrared appearance brightness temperature data, the training simulation clear sky brightness temperature data, the training ground surface information data, the clear sky tag, the cloud tag and the dust tag.
Specifically, in an embodiment, a section from 2017 to 3 to 5 months in 2020 is selected, based on step S2, the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data, the ground surface information data, the clear sky label, the cloud label and the dust label are matched, a machine learning data set is made, and based on the data set, a dust aerosol recognition model is trained, and a relation between input and the label is established. Wherein 10% of the data was taken as verification, the optimal performance was obtained by adjusting the model parameters, and the model performance was evaluated on an independent test set (2021, 3 months).
As shown in fig. 3, 4, 5 and 6, the dust aerosol recognition model has higher accuracy for recognition of dust aerosols on the test set. Fig. 3 and 4 show the daytime case classification results (2021, 3, 23, and 2021, 3, 28, respectively) of the dust aerosol recognition model on the CALIPSO trace, and fig. 5 and 6 show the nighttime case results (2021, 3, 11, and 16, respectively).
It can be seen that the sand aerosol recognition model constructed by the invention has equivalent performance in the daytime and at night, can effectively distinguish cloud and sand, and has good effect on recognition of thin sand.
To further verify the recognition accuracy of the dust aerosol recognition model, the model was further applied to the dust aerosol range detection of the region, and the results are shown in fig. 7, 8, 9, and 10.
Wherein, fig. 7 is a result of observing the product of the sand and dust and CALIPSO in day 3 month of 2021, the yellow dotted line range in the figure is the sand and dust range identified by the product of the sand and dust, and the white point is the pixel of the CALIPSO observing the sand and dust. Fig. 8 shows the recognition result of the dust aerosol recognition model of the present invention in day 3 of 2021 and day 30. Fig. 9 shows the results of the inspection of the sand and dust product and CALIPSO during the day and night of 3 months of 2021, the range of yellow dotted line is the range of sand and dust identified by the sand and dust product, and the white point is the pixel of CALIPSO that observed sand and dust. FIG. 10 shows the recognition result of the dust aerosol recognition model of the present invention at night, 3 months and 15 days of 2021.
It can be seen that the sand aerosol recognition model can more clearly show the result of sand recognition, clearly showing clear sky, clouds and ranges of sand, compared to the sand product. And its detected dust area is typically larger than the dust product, more closely resembling the CALIPSO observations. This is because algorithms based on the bright temperature of the thermal infrared channel applied to sand products are insensitive to thin sand and do not accurately identify small optical thickness sand aerosols. The dust aerosol recognition algorithm provided by the invention is more sensitive to thin dust due to the introduction of clear sky bright temperature and other information, so that thinner dust can be better recognized, and the dust recognition accuracy is obviously improved.
When constructing the data set, the strict quality control is also performed when training the dust aerosol recognition model. The method for performing quality control may refer to the existing method, and will not be described herein.
As shown in fig. 11, the present invention further provides a method for identifying dust aerosol, which includes steps S4 to S6:
s4: and acquiring the thermal infrared appearance brightness temperature measurement data, the simulated clear sky brightness temperature data and the ground surface information data of the area to be identified.
S5: and inputting the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data in the region to be identified into a sand dust aerosol identification model.
S6: and acquiring the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
Specifically, in an embodiment, thermal infrared bright temperature measurement data, simulated clear sky bright temperature data, ground pressure, ground temperature and ground coverage type in the area to be identified can be obtained based on the AHI satellite observation data, the MODIS data product and the ERA-5 analysis data, and the information is input into a sand aerosol identification model, wherein the model automatically identifies the sand aerosol in the area to be identified.
As shown in fig. 12, the present invention further provides a recognition system for dust aerosol, which includes a first acquisition module 20, a recognition module 30, and a second acquisition module 40.
The first acquiring module 20 is configured to acquire bright temperature data, bright temperature data of simulated clear sky, and ground surface information data of the thermal infrared appearance in the area to be identified.
The identification module 30 is configured to input the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data in the to-be-identified area into a dust aerosol identification model.
The second obtaining module 40 is configured to obtain the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
Specifically, in one embodiment, the first acquiring module 20 acquires the bright temperature data of the thermal infrared appearance, the simulated clear sky bright temperature data, the ground pressure, the ground surface temperature and the ground coverage type in the area to be identified based on the AHI satellite observation data, the MODIS data product and the ERA-5 analysis data, the identifying module 30 inputs the information into the dust aerosol identifying model, the model automatically identifies the dust aerosol in the area to be identified, and the second acquiring module 40 acquires the identifying result output by the model.
The present invention also provides a computer-readable storage medium having a computer program stored thereon. The computer program when executed by the processor realizes the training method of the dust aerosol recognition model provided in the embodiment of the invention and/or realizes the recognition method of the dust aerosol provided in the embodiment of the invention.
Any combination of one or more storage media may be employed in the present invention. The storage medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a RAM, a ROM, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The invention further provides electronic equipment. Fig. 13 is a schematic structural diagram of an electronic device 90 according to an embodiment of the invention. As shown in fig. 4, the electronic device 90 in this embodiment includes a memory 901 and a processor 902.
The memory 901 is for storing a computer program; preferably, the memory 901 includes: various media capable of storing program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
In particular, memory 901 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) and/or cache memory. Electronic device 90 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. Memory 901 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of embodiments of the present invention.
The processor 902 is connected to the memory 901, and is configured to execute a computer program stored in the memory 901, so that the electronic device 90 performs the training method of the recognition model of the dust aerosol provided in the embodiment of the present invention, and/or implements the recognition method of the dust aerosol provided in the embodiment of the present invention.
Preferably, the processor 902 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), etc.; but also digital signal processors (Digital Signal Processor, DSP for short), application specific integrated circuits (Application Specific Integrated Circuit, ASIC for short), field programmable gate arrays (Field Programmable Gate Array, FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
Preferably, the electronic device 90 in this embodiment may further comprise a display 903. A display 903 is communicatively coupled to the memory 901 and the processor 902 for displaying a training method of the dust aerosol recognition model and/or an associated GUI interactive interface of the recognition method of the dust aerosol.
The training method of the dust aerosol recognition model and/or the protection scope of the dust aerosol recognition method are not limited to the execution sequence of the steps listed in the embodiment, and all the schemes realized by adding or removing steps and replacing steps according to the prior art by the principle of the invention are included in the protection scope of the invention.
In summary, the embodiments of the present invention provide a method, a system, a model training method, a medium, and a device for identifying a dust aerosol, which can solve the influence caused by complex climate and geographic factors, exclude the influence of cloud and earth surface, extract information in the atmosphere under all weather conditions, accurately identify a thin dust aerosol, and realize the range detection of a dust region. Meanwhile, the method does not need an additional cloud detection program as pretreatment, reduces the steps of the sand and dust identification process, obviously improves the precision of sand and dust identification, and effectively realizes the distinction of clear sky, cloud and sand and dust.
The above embodiments are merely illustrative of the principles of the present invention and its effectiveness, and are not intended to limit the invention. Modifications and variations may be made to the above-described embodiments by those skilled in the art without departing from the spirit and scope of the invention. Accordingly, it is intended that all equivalent modifications and variations of the invention be covered by the claims, which are within the ordinary skill of the art, be within the spirit and scope of the present disclosure.
Claims (10)
1. A method of training a dust aerosol recognition model, comprising:
acquiring training thermal infrared appearance brightness temperature measurement data, training simulation clear sky brightness temperature data and training ground surface information data;
acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data;
and training a dust aerosol recognition model based on the training thermal infrared appearance brightness temperature data, the training simulation clear sky brightness temperature data, the training ground surface information data, the clear sky tag, the cloud tag and the dust tag.
2. The training method of the dust aerosol recognition model according to claim 1, wherein the implementation manners of obtaining training thermal infrared appearance brightness temperature measurement data, training simulation clear sky brightness temperature data and training ground surface information data comprise:
acquiring the training thermal infrared appearance brightness temperature measurement data based on satellite observation data;
and acquiring the training simulation clear sky bright temperature data and the training ground surface information data based on satellite observation data, medium resolution imaging spectrometer data and atmospheric analysis data.
3. The method of training a dust aerosol recognition model of claim 2, wherein the acquiring the training simulated clear sky light temperature data based on satellite observation data, medium resolution imaging spectrometer data, and atmospheric analysis data comprises: acquiring observation angle data based on satellite observation data;
acquiring surface emissivity data based on the medium resolution imaging spectrometer data;
acquiring atmospheric profile data, surface temperature and surface pressure data based on the atmospheric analysis data;
and acquiring the training simulation clear sky bright temperature data based on the atmospheric profile data, the ground surface temperature, the ground surface pressure data, the observation angle data and the ground surface emissivity data.
4. The method of training a dust aerosol recognition model of claim 2, wherein the clear sky, cloud and dust tags correspond to cloud-aerosol lidar and infrared detector satellite data products.
5. The method of claim 4, wherein the acquiring clear sky labels, cloud labels, and dust labels corresponding to the training thermal infrared appearance brightening temperature data, the training simulated clear sky brightening temperature data, and the training surface information data comprises:
establishing corresponding relations among the satellite observation data, the medium resolution imaging spectrometer data, the atmospheric analysis data, the cloud-aerosol laser radar and infrared detector satellite data products based on a space-time matching principle;
and acquiring clear sky labels, cloud labels and dust labels corresponding to the training thermal infrared appearance brightness temperature measurement data, the training simulation clear sky brightness temperature data and the training ground surface information data based on the corresponding relation.
6. A method of training a dust aerosol recognition model according to claim 4, wherein the cloud-aerosol lidar and infrared detector satellite data products comprise a vertical feature distribution data product and an aerosol profile data product.
7. A method of identifying an aerosol of dust, comprising:
acquiring thermal infrared appearance brightness temperature measurement data, simulated clear sky brightness temperature data and ground surface information data in an area to be identified;
inputting the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data in the region to be identified into the sand dust aerosol identification model according to any one of claims 1 to 6;
and acquiring the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
8. A system for identifying an aerosol of dust and sand, comprising:
the first acquisition module is used for acquiring the thermal infrared appearance brightness temperature measurement data, the simulated clear sky brightness temperature data and the ground surface information data in the area to be identified;
the recognition module is used for inputting the thermal infrared appearance brightness temperature data, the simulated clear sky brightness temperature data and the ground surface information data in the region to be recognized into the dust aerosol recognition model according to any one of claims 1 to 6;
and the second acquisition module is used for acquiring the recognition result of the dust aerosol in the region to be recognized, which is output by the dust aerosol recognition model.
9. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program, when executed by a processor, implements the training method of the dust aerosol recognition model of any one of claims 1 to 6 or the recognition method of the dust aerosol of claim 7.
10. An electronic device, the electronic device comprising:
a memory storing a computer program;
a processor, in communication with the memory, which when invoked executes the training method of the dust aerosol recognition model of any one of claims 1 to 6 or the recognition method of the dust aerosol of claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310547308.7A CN116756560A (en) | 2023-05-15 | 2023-05-15 | Dust aerosol recognition method, system, model training method, medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310547308.7A CN116756560A (en) | 2023-05-15 | 2023-05-15 | Dust aerosol recognition method, system, model training method, medium and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116756560A true CN116756560A (en) | 2023-09-15 |
Family
ID=87954158
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310547308.7A Pending CN116756560A (en) | 2023-05-15 | 2023-05-15 | Dust aerosol recognition method, system, model training method, medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116756560A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117725370A (en) * | 2024-02-18 | 2024-03-19 | 国家气象中心(中央气象台) | Minute-level sand and dust weather identification method, apparatus, equipment and medium |
-
2023
- 2023-05-15 CN CN202310547308.7A patent/CN116756560A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117725370A (en) * | 2024-02-18 | 2024-03-19 | 国家气象中心(中央气象台) | Minute-level sand and dust weather identification method, apparatus, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113705478B (en) | Mangrove single wood target detection method based on improved YOLOv5 | |
| CN109086668B (en) | Unmanned aerial vehicle remote sensing image road information extraction method based on multi-scale generation countermeasure network | |
| Klodt et al. | Field phenotyping of grapevine growth using dense stereo reconstruction | |
| CN111398176B (en) | Water body water color abnormity remote sensing identification method and device based on pixel scale characteristics | |
| Alvioli et al. | Topography-driven satellite imagery analysis for landslide mapping | |
| US20200242749A1 (en) | System and method for processing images of agricultural fields for remote phenotype measurement | |
| CN116756560A (en) | Dust aerosol recognition method, system, model training method, medium and equipment | |
| Varjo et al. | Image based visibility estimation during day and night | |
| CN114037907A (en) | Detection method and device for power transmission line, computer equipment and storage medium | |
| CN111553184A (en) | Small target detection method and device based on electronic purse net and electronic equipment | |
| US10453205B2 (en) | Hierarchical tiling method for identifying a type of surface in a digital image | |
| Liu et al. | Ground-based cloud classification using weighted local binary patterns | |
| Oga et al. | River state classification combining patch-based processing and CNN | |
| CN114550016B (en) | Unmanned aerial vehicle positioning method and system based on context information perception | |
| CN113553708B (en) | Method and device for tracing key influence factors of simulation model | |
| CN116091481A (en) | Spike counting method, device, equipment and storage medium | |
| Shao et al. | An explicit index for assessing the accuracy of cover-class areas | |
| CN113361405B (en) | Asian image recognition method and system based on yolo v3 | |
| CN114694128A (en) | Pointer instrument detection method and system based on abstract metric learning | |
| CN113989632A (en) | Bridge detection method and device for remote sensing image, electronic equipment and storage medium | |
| CN113360593A (en) | Sensor data processing method and device | |
| CN110895701B (en) | Forest fire online identification method and device based on CN and FHOG | |
| Rayegani et al. | A simple model for PIFs extraction at digital change detection approach | |
| CN116563841B (en) | Detection method and detection device for power distribution network equipment identification plate and electronic equipment | |
| CN109145945A (en) | A kind of hyperspectral image classification method that non local weighting joint sparse indicates |
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 |