CN110082777A - A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network - Google Patents
A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network Download PDFInfo
- Publication number
- CN110082777A CN110082777A CN201910333252.9A CN201910333252A CN110082777A CN 110082777 A CN110082777 A CN 110082777A CN 201910333252 A CN201910333252 A CN 201910333252A CN 110082777 A CN110082777 A CN 110082777A
- Authority
- CN
- China
- Prior art keywords
- satellite
- fine particle
- polarization
- optical thickness
- particle aerosol
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
- G01B11/06—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness for measuring thickness ; e.g. of sheet material
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Electromagnetism (AREA)
- Geophysics And Detection Of Objects (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of fine particle aerosol optical thickness inversion methods based on polarization satellite remote sensing and neural network, including polarization satellite data 1) is carried out optimization pretreatment, keep each physical quantity more related to the characterization of fine particle aerosol optical thickness；2) the fine particle aerosol optical thickness for optimizing pretreated polarization satellite data and ground actual measurement is carried out to the matching on space-time；3) using the optimization of successful match polarization satellite data as input, the training that fine particle aerosol optical thickness carries out neural network as desired output is surveyed on ground.Complicated road radiation transmission process internalization in the weight and threshold value of each node, is reached best inverting state by the automatic study of network inside the neural network by this method, at the regional level in inverting apply than traditional inversion method more rapidly it is convenient accurately.
Description
Technical field
The present invention relates to a kind of fine particle aerosol optical thickness inverting sides based on polarization satellite remote sensing and neural network
Method belongs to the atmospheric remote sensing field in satellite remote sensing.
Background technique
With the propulsion of China's process of industrialization, more serious air pollution be increasingly becoming endanger people's living standard with
The key factor of health, the pollutant component of such as PM2.5 etc and other substances have collectively constituted the aerosol in atmosphere.Gas is molten
Colloidality matter is complicated, and source multiplicity, the origin cause of formation is different, can be divided into corase particles and fine particle by particle size.In some industrial cities,
Artificial pollution factor contributes to most fine particle aerosol accounting, accurately monitors in real time and classification energy to aerosol
Enough help government in time, science make a policy, especially the inverting of fine particle aerosol to industrial pollution supervision have very
Important meaning.The inversion method of aerosol mainly passes through the non-inclined remote sensing building look-up table such as MODIS satellite at present, this is first
Need to remove the information from ground, and these terrestrial informations have accounted for 95% or more ratio, therefore inverting in non-inclined remote sensing
Process is sufficiently complex.However fine particle aerosol has very strong sensibility to polarised light, and the light of land scattering is difficult to occur
Polarization signal is simultaneously transferred to satellite, and by this property, using satellite, fast accurate inverting fine particle aerosol optical thickness has
Scientific feasibility.
Satellite remote sensing using space, temporal extensive and long sequence as advantage, can with convenient real-time offer aerosol, cloud,
The information such as wind speed, this is the advantage that ground based detection and aerospace detection cannot compare.Satellite sounding polarization signal is utilized earliest
It is French POLDER system satellite, the series is retired in 2013.It is carried on No. five satellites of high score that China emits in May, 2018
The spaceborne multi-angle of first of China polarizes earth observation load, this will inject new power to probe into polarization remote sensing, however
Applicability is there are still space is improved at home for the algorithm of existing polarization remote sensing inverting fine particle aerosol, effectively quickly and accurately
The importance of inverting country fine particle aerosol parameter is self-evident.
Summary of the invention
To solve the above-mentioned problems, the present invention realizes a kind of fine particle gas based on polarization satellite remote sensing and neural network
Colloidal sol optical thickness inversion method.
To achieve the goals above, the technical scheme adopted by the invention is that:
A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network, including,
1) polarization satellite data is subjected to optimization pretreatment；
Pretreatment is intended to increase the linearly related degree of each physical quantity and fine particle aerosol optical thickness characterization, and the degree of correlation is highest
For optimal pretreatment；
2) pretreated polarization satellite data will be optimized and ground actual measurement fine particle aerosol optical thickness carries out time-space registration;
Satellite remote sensing and ground station actual measurement are run alone, the spatial character passed by using satellite and aerosol transmission when
Between characteristic carry out mean match and screening；
3) using the optimization of successful match polarization satellite data and website actual measurement fine particle aerosol thickness data as inputting and
Desired output carries out the training of neural network, which can be used to inverting fine particle aerosol optical thickness.
Selected pretreatment mode are as follows: two components of Q, U in the Stokes coefficient that satellite measures will be polarized and be scaled total polarization apparently
Reflectivity；The zenith angle of satellite, azimuth and solar zenith angle, four, azimuth radian are scaled light in atmosphere outer layer-
The distance of this radiation transmission of atmosphere-ground-atmosphere-satellite characterizes.Optimize the polarization apparent reflectance of data prediction
Reduction formula is,
Wherein, be to polarize the both direction component that linearly polarizes of Stokes,For polarization intensity.
The light transmission path for optimizing data prediction is characterized as,
;;
Wherein、The respectively zenith angle of the sun and satellite sensor,、The respectively sun and biography
The azimuth of sensor, the characterized transmission process of solar radiation of above three amount.
The time response of the spatial character and aerosol transmission passed by using satellite carries out mean match and screening: aerosol transmission
Speed is widely considered to be 50km/h, satellite is passed by the measured data mean value that the total ground 0.5h front and back 15min measures, with earth station
Satellite pixel data Corresponding matching in point surrounding 25km, reduces asynchronous bring error by space-time conversion.
The satellite of successful match optimization data are trained with measured data as input and desired output, desired output and reality
It is success that border, which exports related coefficient to be more than or equal to 0.9, and polarization data that can be new to network inputs carries out fine particle aerosol light
Learn the prediction and inverting of thickness.
Figure of description
Fig. 1 is the invention patent technology path.
Fig. 2 is that inverting value and measured value of the invention patent in CHINESE REGION compare.
Fig. 3 is that inverting value and measured value of the France PARASOL satellite official products in CHINESE REGION compare.
Fig. 4 is that inverting value and measured value of the MODIS official products in the U.S. in CHINESE REGION compare.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Implement to be only used for clearly illustrating below of the invention
Technical solution, and the scope of the present invention cannot be limited with this.
As shown in Figure 1, a kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network, packet
Include following steps.
Step 1, polarization satellite data pretreatment.
Pretreatment is intended to increase the linearly related degree of each physical quantity and fine particle aerosol optical thickness characterization, and the degree of correlation is highest
For optimal pretreatment.
Selected pretreatment mode is that will polarize two components of Q, U in three wave band Stokes coefficients that satellite measures to be scaled satellite
The corresponding total polarization apparent reflectance of each detecting band；To measure the satellite at moment zenith angle, azimuth and solar zenith angle,
Four, azimuth radian is converted to light in this biography of atmosphere outer layer-atmosphere-ground-atmosphere-satellite by trigonometric function
Distance characterization during defeated.
Optimize the polarization apparent reflectance reduction formula of data prediction are as follows:
Wherein、It is the both direction component for polarizing Stokes and linearly polarizing,For polarization intensity.
Optimize the light transmission path characterization of data prediction are as follows:
; ;
Wherein、The respectively zenith angle of the sun and satellite sensor,、The respectively sun and sensing
The azimuth of device, the characterized transmission process of solar radiation of above three amount.
Step 2, pretreated polarization satellite data will be optimized and ground actual measurement fine particle aerosol optical thickness carries out space-time
Matching.
The time response of the spatial character and aerosol transmission passed by using satellite carries out mean match and screening: aerosol transmission
Speed educational circles is widely considered to be 50km/h, and satellite is passed by front and back 15min, the measured data mean value that the total ground 0.5h measures, with
Satellite pixel data Corresponding matching around ground station in 25km reduces the asynchronous bring error of space-time.
Step 3, using the satellite polarization data of successful match as network inputs, website surveys fine particle aerosol thickness as the phase
It hopes and exports training network.
The atmospheric polarization apparent reflectance and three characterized satellites and the sun of selection tri- wave bands of 490nm, 670nm, 865nm
Positional relationship and the data of road radiation transmission process are inputted as a set of sample, and the fine particle aerosol optics that station probes obtain is thick
Degree is corresponding output.
It is trained by several samples, inverting comparison is carried out to test set after training, if output data and truthful data
Coefficient R > 0.9 then can be used the network to carry out more fine particle aerosol invertings.The calculation method of related coefficient are as follows:
Wherein,、Respectively measured data and neural network forecast data,ForWithCovariance,For
Variance,ForVariance.
Fig. 2 is the invention patent inversion result comparison diagram, and transverse and longitudinal coordinate is respectively true value, inverting value.Fig. 4 is MODIS official
The comparison diagram of fine particle aerosol product and true value can be seen that the fine particle gas for CHINESE REGION according to related coefficient etc.
Colloidal sol optical thickness inverting, the inversion result of the invention patent are substantially better than conventional MODIS inversion method.Fig. 3 is PARASOL
The inverting value and measured value of satellite official products compares, it can be seen that in CHINESE REGION, the result is better than MODIS product, but is lower than
The method of the invention patent.
The above is only a preferred embodiment of the present invention, it is noted that those skilled in the art are come
It says, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations are also answered
It is considered as protection scope of the present invention.
Claims (6)
1. a kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network, feature exist
In:
1) polarization satellite data is subjected to optimization pretreatment:
Pretreatment is intended to increase each physical quantity by converting every original satellite data and fine particle aerosol optical thickness characterizes
Linearly related degree, linear degree it is highest be optimal pretreatment；
2) pretreated polarization satellite data will be optimized and ground actual measurement fine particle aerosol optical thickness carries out time-space registration:
Satellite observation and ground station actual measurement are run alone, are passed by the spatial character and aerosol transmission of data using satellite
Time response carry out mean match and screening；
3) using the optimization of successful match polarization satellite data and website actual measurement fine particle aerosol thickness data as inputting and
Desired output carries out the training of neural network, and the successful network of training can be used to inverting fine particle aerosol optical thickness.
2. a kind of fine particle aerosol optical thickness based on polarization satellite remote sensing and neural network according to claim 1
Inversion method, it is characterised in that: selected pretreatment mode is that will polarize two components of Q, U in the Stokes coefficient that satellite measures to change
Calculating is total polarization apparent reflectance；Zenith angle, azimuth and solar zenith angle, four, the azimuth of the satellite at moment will be measured
Radian is converted to light in atmosphere outer layer-this transmission process Road of atmosphere-ground-atmosphere-satellite by trigonometric function
Diameter characterization.
3. a kind of fine particle aerosol optical thickness based on polarization satellite remote sensing and neural network according to claim 2
Inversion method, it is characterised in that: the polarization apparent reflectance reduction formula for optimizing data prediction is,
Wherein、It is the both direction component for polarizing Stokes and linearly polarizing,For polarization intensity.
4. a kind of fine particle aerosol optical thickness based on polarization satellite remote sensing and neural network according to claim 2
Inversion method, it is characterised in that: the light transmission path for optimizing data prediction is characterized as,
; ;
Wherein、The respectively zenith angle of the sun and satellite sensor,、The respectively sun and sensing
The azimuth of device, the characterized transmission process of solar radiation of above three amount.
5. a kind of fine particle aerosol optical thickness based on polarization satellite remote sensing and neural network according to claim 1
Inversion method, it is characterised in that: the time response of the spatial character and aerosol transmission passed by using satellite carries out mean match
With screening: aerosol transmission speed is considered 50km/h, and satellite is passed by front and back 15min, the actual measurement number that the total ground 0.5h measures
According to the satellite pixel data Corresponding matching in 25km around mean value, with ground station, the asynchronous bring error of space-time is reduced.
6. a kind of fine particle aerosol optical thickness based on polarization satellite remote sensing and neural network according to claim 1
Inversion method, it is characterised in that: carried out using the satellite of successful match optimization data and measured data as input and desired output
Training, it is success that desired output and reality output related coefficient, which are more than or equal to 0.9, polarization data that can be new to network inputs into
The prediction and inverting of row fine particle aerosol optical thickness.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910333252.9A CN110082777A (en) | 2019-04-24 | 2019-04-24 | A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910333252.9A CN110082777A (en) | 2019-04-24 | 2019-04-24 | A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110082777A true CN110082777A (en) | 2019-08-02 |
Family
ID=67416377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910333252.9A Pending CN110082777A (en) | 2019-04-24 | 2019-04-24 | A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110082777A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111191380A (en) * | 2020-01-08 | 2020-05-22 | 北京大学 | Atmospheric aerosol optical thickness estimation method and device based on measurement data of foundation spectrometer |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103927455A (en) * | 2014-04-24 | 2014-07-16 | 中国科学院遥感与数字地球研究所 | Land aerosol optical property retrieval method based on Gaofen-1 satellite |
| CN105023043A (en) * | 2015-07-23 | 2015-11-04 | 杭州师范大学 | AOD-based PM2.5 inversion model for Hangzhou region |
| CN107066786A (en) * | 2016-11-22 | 2017-08-18 | 深圳职业技术学院 | Aerosol optical depth inversion algorithm based on neutral net |
| CN108896450A (en) * | 2018-05-14 | 2018-11-27 | 中国科学院合肥物质科学研究院 | The atmospheric aerosol inversion method combined based on multiple angle multiple-pass polarization information with depth learning technology |
-
2019
- 2019-04-24 CN CN201910333252.9A patent/CN110082777A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103927455A (en) * | 2014-04-24 | 2014-07-16 | 中国科学院遥感与数字地球研究所 | Land aerosol optical property retrieval method based on Gaofen-1 satellite |
| CN105023043A (en) * | 2015-07-23 | 2015-11-04 | 杭州师范大学 | AOD-based PM2.5 inversion model for Hangzhou region |
| CN107066786A (en) * | 2016-11-22 | 2017-08-18 | 深圳职业技术学院 | Aerosol optical depth inversion algorithm based on neutral net |
| CN108896450A (en) * | 2018-05-14 | 2018-11-27 | 中国科学院合肥物质科学研究院 | The atmospheric aerosol inversion method combined based on multiple angle multiple-pass polarization information with depth learning technology |
Non-Patent Citations (2)
| Title |
|---|
| 汪立宏等: "《生物医学光学原理和成像》", 31 January 2017 * |
| 涂济民: "《太阳能系统工程-技术创新、产业整合的探索与实践》", 31 August 2015 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111191380A (en) * | 2020-01-08 | 2020-05-22 | 北京大学 | Atmospheric aerosol optical thickness estimation method and device based on measurement data of foundation spectrometer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Haeffelin et al. | Evaluation of mixing-height retrievals from automatic profiling lidars and ceilometers in view of future integrated networks in Europe | |
| Hilburn et al. | Development and interpretation of a neural-network-based synthetic radar reflectivity estimator using GOES-R satellite observations | |
| CN109814175B (en) | Strong convection monitoring method based on satellite and application thereof | |
| Reuter et al. | A joint effort to deliver satellite retrieved atmospheric CO 2 concentrations for surface flux inversions: the ensemble median algorithm EMMA | |
| CN106407656A (en) | Retrieval method for aerosol optical thickness based on high resolution satellite image data | |
| CN111753439B (en) | Aerosol optical thickness inversion method of multi-angle polarized satellite sensor | |
| CN110031916A (en) | Rainfall intensity measurement method based on satellite-ground link attenuation effect | |
| CN110595968B (en) | PM2.5 concentration estimation method based on geostationary orbit satellite | |
| CN111581884A (en) | Satellite-borne multi-resolution target optical radiation image simulation method | |
| Ma et al. | Detecting the ocean surface from the raw data of the MABEL photon-counting lidar | |
| Oue et al. | Estimation of cloud fraction profile in shallow convection using a scanning cloud radar | |
| CN105678236A (en) | Land vegetation canopy polarization reflection modeling method | |
| Rietveld et al. | The tromsø dynasonde | |
| CN112749494A (en) | Method for acquiring dynamic accumulated snow depth | |
| Savelyev et al. | On direct passive microwave remote sensing of sea spray aerosol production | |
| CN114139566A (en) | Method for improving accuracy of measurement based on machine learning weighted average fusion feature extraction | |
| CN110991087A (en) | Wind field inversion method and system based on multi-incidence-angle networking SAR satellite data | |
| CN110082777A (en) | A kind of fine particle aerosol optical thickness inversion method based on polarization satellite remote sensing and neural network | |
| Liu et al. | Graphics algorithm for deriving atmospheric boundary layer heights from CALIPSO data | |
| Qin et al. | Sea surface wind speed retrieval from Sentinel-1 HH polarization data using conventional and neural network methods | |
| Zhigang et al. | Retrieving atmospheric temperature profiles from AMSU-A data with neural networks | |
| Liu et al. | A novel algorithm for detecting center of tropical cyclone in satellite infrared images | |
| Bi et al. | Fast CO2 retrieval using a semi-physical statistical model for the high-resolution spectrometer on the Fengyun-3D satellite | |
| Hanabusa et al. | Deep-learning-based calibration in contrast source inversion based microwave subsurface imaging | |
| Reeder et al. | The detection of flow asymmetries in the tropical cyclone environment |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190802 |
|
| WD01 | Invention patent application deemed withdrawn after publication |