CN113849490A - Residual error correction method for satellite remote sensing reflectivity product data - Google Patents
Residual error correction method for satellite remote sensing reflectivity product data Download PDFInfo
- Publication number
- CN113849490A CN113849490A CN202111096999.0A CN202111096999A CN113849490A CN 113849490 A CN113849490 A CN 113849490A CN 202111096999 A CN202111096999 A CN 202111096999A CN 113849490 A CN113849490 A CN 113849490A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- data
- residual error
- sensing reflectivity
- satellite
- 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
- 238000002310 reflectometry Methods 0.000 title claims abstract description 94
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000012937 correction Methods 0.000 title claims description 24
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 29
- 230000003595 spectral effect Effects 0.000 claims abstract description 27
- 238000001228 spectrum Methods 0.000 claims abstract description 15
- 230000008859 change Effects 0.000 claims abstract description 8
- 238000013213 extrapolation Methods 0.000 claims description 18
- 238000003062 neural network model Methods 0.000 claims description 18
- 238000010521 absorption reaction Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 11
- 230000005540 biological transmission Effects 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 5
- 238000011065 in-situ storage Methods 0.000 claims description 5
- 239000002245 particle Substances 0.000 claims description 4
- 239000013618 particulate matter Substances 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000004519 manufacturing process Methods 0.000 abstract description 2
- 238000004422 calculation algorithm Methods 0.000 description 30
- 230000003287 optical effect Effects 0.000 description 6
- 238000013459 approach Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 239000013535 sea water Substances 0.000 description 3
- 241000195474 Sargassum Species 0.000 description 2
- ATNHDLDRLWWWCB-AENOIHSZSA-M chlorophyll a Chemical compound C1([C@@H](C(=O)OC)C(=O)C2=C3C)=C2N2C3=CC(C(CC)=C3C)=[N+]4C3=CC3=C(C=C)C(C)=C5N3[Mg-2]42[N+]2=C1[C@@H](CCC(=O)OC\C=C(/C)CCC[C@H](C)CCC[C@H](C)CCCC(C)C)[C@H](C)C2=C5 ATNHDLDRLWWWCB-AENOIHSZSA-M 0.000 description 2
- 229930002868 chlorophyll a Natural products 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000004907 flux Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000003702 image correction Methods 0.000 description 2
- 238000005259 measurement Methods 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
- 238000011160 research Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 235000002779 Morchella esculenta Nutrition 0.000 description 1
- 240000002769 Morchella esculenta Species 0.000 description 1
- CLOMYZFHNHFSIQ-UHFFFAOYSA-N clonixin Chemical compound CC1=C(Cl)C=CC=C1NC1=NC=CC=C1C(O)=O CLOMYZFHNHFSIQ-UHFFFAOYSA-N 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Computational Mathematics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention provides a method for correcting data residual errors of satellite remote sensing reflectivity products, which is used for quickly and effectively removing the data residual errors of the remote sensing reflectivity products and improving the data accuracy of ocean water color satellite products; a remote sensing reflectivity data residual wave spectrum relation reconstruction method is provided, and remote sensing reflectivity data residual wave spectrum characteristics are analyzed more simply and conveniently; by combining a neural network technology, a spectral slope coefficient method for extracting the residual error of the remote sensing reflectivity data from the satellite remote sensing reflectivity information is provided, and the time-space change characteristics of the spectral characteristics of the residual error of the satellite image remote sensing reflectivity data can be described more finely; the method is used for correcting the data residual error of the ocean water body remote sensing reflectivity product, can obtain the remote sensing reflectivity product with more reasonable numerical value and more stable and reliable space-time distribution, and supports the production of high-precision water color products.
Description
Technical Field
The invention belongs to the technical field of ocean optical remote sensing, and particularly relates to a method for correcting data residual errors of satellite remote sensing reflectivity products.
Background
The ocean optical remote sensing technology is a new detection technology, can detect the water color information of the sea surface through a detector carried on a remote sensing platform so as to obtain the ocean information, is widely applied to aspects of ocean ecological environment monitoring, ocean disaster influence monitoring and evaluation and the like, and greatly improves the capability of human beings for coping with natural disasters and global climate change. Since the first ocean water color satellite appeared in 1978, water color remote sensing has been developed in the aspects of ocean gas radiation transmission process, ocean organism optical mechanism, satellite product authenticity inspection and the like.
Remote sensing reflectivity data RrsIs an important foundation for ocean optics research and is also the key point for ocean water color remote sensing in China. However, the uncertainty of the noise, radiometric calibration and atmospheric correction in the sensor makes the satellite remote sensing reflectivity product full of noise, which seriously affects the quality of the water color product. In addition, any measurement lacking uncertainty information is an invalid observation, and the existing remote sensing reflectivity product has fresh uncertainty description information.
In the prior art, the star-to-earth method is a relatively old and mature method (Bailey and Werdell 2006; Clark et al 1997; Hooker and Maritorena 2000; McClain et al 1992; Moore et al 2009; O' Reilly and Werdell 2019; Zibordi et al 2004); there is also a strict match criterion (Bailey and Werdell 2006), which allows R to be observed by contrasting satellitesrsAnd measuring R in sitursResidual errors are easily obtained, and furthermore, the satellite RrsCan be used to determine ocean water type species, where residual data is calculated using a star-to-ground alignment algorithm, so that the residual distribution can be plotted dynamically at a global scale (Moore et al 2009). However, data coverage is a significant drawback to these methods, as the evaluation statistics are only valid if coverage is measured in the field.
ResearchersIt was found that when the total absorption coefficient (a (. lamda.))0) And the extrapolated index of backscattering coefficient (Y) can be determined from error-containing R at about 555nnrsWhen the data is accurately derived (chen et al 2014), the pixel level residuals can be resolved by the IOPs (intrinsic optical properties) data processing system (IDAS). One advantage of the clean water approach (clear water approach) over the star-to-ground algorithm is that it does not rely on site observation time and space coverage. Potential assumptions for this approach are (Hu et al 2012): (1) the main optical component should be homogeneous in a relatively small area; (2) the band-difference algorithm is almost insensitive to residual errors, while the band-ratio algorithm is very sensitive to these errors, so when inputting RrsBoth chlorophyll-a algorithms can converge to the same chlorophyll-a concentration without error (Hu et al 2013); however, the clean water method is suitable for some small areas without error pixels; but less effective for low quality data contaminated with a large amount of uncertainty.
In order to improve the spectral relationship (IDAS) based on the clean water body methodcw) The performance of the IDAS algorithm is that a more robust spectral relationship is researched when the IDAS algorithm is used so as to improve the R of the satellitersThe product is very important for providing uncertainty information of remote sensing reflectivity data and improving the quality of the remote sensing reflectivity data.
Disclosure of Invention
The invention provides a method for correcting data residual errors of satellite remote sensing reflectivity products, which aims to solve the problems of inaccurate marine data monitoring and low reliability in the field of marine remote sensing and aims to quickly and effectively remove the data residual errors of the remote sensing reflectivity products and improve the data precision of marine water color satellite products.
The invention is realized by adopting the following technical scheme: a satellite remote sensing reflectivity product data residual error correction method comprises the following steps:
step A, measuring remote sensing reflectivity data R by satellitersObtaining data residual error information of satellite remote sensing reflectivity products by comparing in-situ observation with satellite detection to form remote sensing reflectivity residual error delta RrsA data set;
step B, establishing remote sensing reflectivity residual error delta R by analyzing remote sensing reflectivity residual error spectrum relationrsA relation to the wavelength λ;
wherein, Δ RrsRepresenting remote sensing reflectivity data residual error, C is a positive constant, S represents the spectral slope coefficient of the residual error, lambda represents wavelength, lambda represents the spectral slope coefficient of the residual error0Represents a reference band; setting reference waveband remote sensing reflectivity residual error delta Rrs(λ0) Used for calculating the remote sensing reflectivity residual error delta R of other wave bandsrsA value of (d);
step C, remotely sensing reflectivity data R by using satellitersTaking a spectral slope coefficient S of a remote sensing reflectivity data residual as an output for input, establishing a spectral slope coefficient neural network model of the residual to calculate the value of S;
step D, simulating the remote sensing reflectivity residual error delta R by combining the remote sensing reflectivity data residual error probability distribution and the spectrum relation characteristic on the basis of the radiation transmission simulation datarsTo establish an absorption coefficient a (lambda)0) And a neural network model of the backscattering coefficient extrapolation index Y, and solving a (lambda)0) And Y;
step E, determining an extrapolation wave band lambda1Total backscattering coefficient b ofb(λ1) The extrapolated band is defined by a reference band λ0Extrapolation gives:
wherein, bbwKnown pure water backscattering coefficient, bbpIs the backscattering coefficient of the particulate matter; setting the back scattering coefficient b of the particles at the extrapolation wavebandbp(λ1) Then λ can be calculated1Total backscattering coefficient of (b)b(λ1);
Step F, using absorption coefficient a (lambda)0) General, aCoefficient of backscattering bbRemote sensing reflectivity data RrsAnd remote sensing reflectivity residual error Δ RrsAnd (3) calculating to obtain the final remote sensing reflectivity residual error through nonlinear iteration on the basis of the relation (1) of the wavelength lambda.
Further, the iterative process of step F is as follows:
step F1: setting an initial DeltaRrs(λ0) And bbp(λ0) A value of (d);
step F2: if it is the first iteration, use the initial Δ R of step F1rs(λ0) Substituting the S obtained by the calculation in the step C into a formula (1) to obtain a satellite visible light waveband remote sensing reflectivity data residual error, realizing the preliminary correction of the data residual error and updating the satellite remote sensing reflectivity data Rrs(ii) a If not, Δ R in step F5 is usedrs(λ0) Carrying out data residual error correction and updating satellite remote sensing reflectivity data Rrs;
Step F3: satellite remote sensing reflectivity data R obtained in step F2rsFor input, according to the absorption coefficient and backscattering coefficient extrapolation index neural network model constructed in the step D, a (lambda) is obtained0) And Y;
step F4: if it is the first iteration, use initial b of step F1bp(λ0) Value, in combination with equation (2) to solve bb(λ1) If not, using b in step F5bp(λ0) Calculation of bb(λ1) The information of (a);
step F5: combining a (lambda) in step F30) And Y information, while combining b of step F4b(λ1) Substituting the information into the formula (3) to change the formula (3) to contain only DeltaRrs(λ0) And bb(λ0) A system of equations for two variables, by solving the system of equations, Δ R is obtainedrs(λ0) And bb(λ0);
Wherein, bb=bbp+bbw,bbwIs a known value, g0And g1Is a known empirical coefficient;
step F6: and then calculating and obtaining data residual error information of a visible light wave band according to the spectrum relation of the formula (1), realizing remote sensing reflectivity data residual error correction, and updating the data residual error information for a (lambda)0) And the remote sensing reflectance calculated by Y;
step F7: Δ R obtained when two adjacent iterations are calculatedrs(λ0) And bb(λ0) Are all less than 10-7If so, ending the iteration and repeatedly executing the steps B-F; updating DeltaRrsAnd the correlation value is used for realizing the data residual error correction of the satellite remote sensing reflectivity product.
Compared with the prior art, the invention has the advantages and positive effects that:
the remote sensing reflectivity data residual wave spectrum relation reconstruction method provided by the scheme can analyze the remote sensing reflectivity data residual wave spectrum characteristics more simply and conveniently; by combining a neural network technology, a spectral slope coefficient method for extracting the residual error of the remote sensing reflectivity data from the satellite remote sensing reflectivity information is provided, and the time-space change characteristics of the spectral characteristics of the residual error of the satellite image remote sensing reflectivity data can be described more finely; the method is used for correcting the data residual error of the ocean water body remote sensing reflectivity product, can obtain the remote sensing reflectivity product with more reasonable numerical value and more stable and reliable space-time distribution, and supports the production of high-precision water color products.
Drawings
FIG. 1 is a diagram of a residual error correction method (IDAS) according to an embodiment of the present inventionnn) Schematic diagram of the principle;
FIG. 2 is a comparison of spectral slope coefficients derived using a neural network model according to an embodiment of the present invention;
FIG. 3 shows an embodiment of the present invention based on simulated data and measured data, where (a) - (c), (d) - (f) and (g) - (i) respectively contain errors RrsResidual error comparison schematic diagram;
FIG. 4 shows HY-1C secondary GAC R of Sargassum at 21 days 6 months in 2021rs(565) Model derived bbp(565): (a) and (b) two residual error correction algorithms IDAScwAnd IDASnnB of (a)bp(565) An image; (c) and (d) is bbp(565) IDAS after image correctioncwAnd IDASnnA CV value of (D); (e) and (f) is bbp(565) Histograms and CV value histograms of the original and corrected images;
FIG. 5 illustrates an embodiment of the present invention using (a) IDAScwAnd (b) IDASnnDaily average b after residual correction by the algorithmbp(565) And (4) data.
Detailed Description
In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be further described with reference to the accompanying drawings and examples. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those described herein, and thus, the present invention is not limited to the specific embodiments disclosed below.
The embodiment provides a residual error correction method for satellite remote sensing reflectivity product data, which is used for improving satellite remote sensing reflectivity RrsThe principle of the product is shown in figure 1, and the method specifically comprises the following steps:
step 1: measured satellite remote sensing reflectivity data RrsObtaining data residual error information of satellite remote sensing reflectivity products through in-situ observation and satellite detection comparison to form remote sensing reflectivity residual error delta RrsA data set;
specifically, the in-situ observation and satellite detection comparison are carried out by taking an NASA geosynchronous constraint rule as guidance, wherein the NASA geosynchronous constraint rule refers to accurate satellite sea color data published by the national aviation and space navigation administration (NASA) Marine biological processing group (OBPG), and the rule carries out data correction by comparing satellite data with field measurement data (refer to Sean W.Bailey, P.Jerney Werdell. Amulti-sensor address for the on-inhibition evaluation of environmental color satellite data products [ J ]. Remote Sensing of environmental aspects 102(2006) 12-23.).
Step 2: establishing remote sensing reflectivity residual error by analyzing remote sensing reflectivity residual error spectrum relation△RrsA relation to the wavelength λ;
satellite remote sensing reflectivity product RrsThe spectral relationship of the residual error in the visible light range from short wavelength to long wavelength satisfies monotonicity, and the spectral relationship is generally described by an exponential function, but the slope of the exponential function is larger than that of the long wavelength band in the short wavelength blue light band, which is unfavorable for accurate line fitting of the residual error, so that the scheme combines a linear adjustment factor in an exponential equation to simulate the natural spectral curve of the residual error, and adds a constant C to delta R to all visible light bands before regression analysisrsIn this embodiment, 565nn band is used as reference, i.e. reference band λ0Preferably 565nm, which is generally 400-570nm, and the data residual radiation transmission relation equation is expressed as the spectrum relation shown in the formula (1):
wherein, Δ RrsRepresenting the remote-sensing reflectance residual, C is a positive "constant", this example C is set to 0.04, Δ R from step 1rsThe data set is statistically estimated, S represents a spectral slope coefficient of a residual error, and lambda represents a wavelength; this embodiment sets Δ Rrs(565) Initial value of (a): 0.0001, preliminary calculation of Δ RrsIn subsequent iterative calculations, Δ R is continuously optimized by equation (1) and step 6-1rsAnd updated in the remote sensing reflectivity data Rrs;
And step 3: remote sensing of reflectivity data R by satellitersTaking a spectral slope coefficient S of a remote sensing reflectivity data residual as an output for input, establishing a spectral slope coefficient neural network model of the residual to calculate the value of S;
in particular by satellite remote sensing reflectivity data RrsFor inputting, obtaining the spectral slope coefficient S of the remote sensing reflectivity residual error of each pixel by utilizing the established spectral slope coefficient neural network model, wherein the neural network model is a neural network model comprising two hidden layers, and calculating the value of S by establishing the neural network model and the like to be a relatively mature technologyAlthough not explained in detail, the embodiment uses the satellite RrsInitial values in 4 common bands (443nm, 490nm, 565nm and 670nm) are used as inputs to obtain the value of S;
and 4, step 4: simulating satellite remote sensing reflectivity data containing data residual by combining the remote sensing reflectivity data residual probability distribution and the spectrum relation (formula (1)) characteristic on the basis of radiation transmission simulation data, establishing a neural network model of an absorption coefficient a and a backscattering coefficient extrapolation index Y, and solving a (lambda)0) And Y;
the simulated data is calculated by artificially adding residual data to R of the histogram distribution of extrapolation indexrsData obtained by commercial software of Hydrolight, which is a radiation transmission model written by Fortran language according to light and water to solve a radiation transmission equation, and the input of the radiation transmission equation comprises water absorption and scattering characteristics, water component concentrations and the like, and the output is RrsAnd (4) data. The correlation data are combined with R, 3866-3886 using a mature NQAA neural network Algorithm (Quasi-Analytical-based neural network Algorithm, NN-Quasi-Analytical Algorithm, Chen et al (2003), Impropening software data products for open environments with a scheme to correct the residual errors in removal sensing communication. journal of geographic Research-Oceann, 121,3866-3886.)rsTaking data as input, respectively solving a (lambda)0) And Y; the absorption coefficient a (lambda)0) In this example, the absorption coefficient at 565nm is calculated, and a (565) is solved; the extrapolation index Y of the backscattering coefficient is equal to bb(λ) related index.
And 5: determining an extrapolation band λ1The total backscattering coefficient of (a);
this example extrapolates the band λ1For example 670, the extrapolation band may be selected from the reference band λ0Extrapolation, for obtaining the final remote sensing reflection in cooperation with iterative calculationThe rate residual, then:
wherein, bbwIs the backscattering coefficient of pure water, bbw(670) Is the backscattering coefficient of pure water at 670nm and is a known parameter, bb(670) Total backscattering coefficient calculated for 670nm, bbp(565) Is the particle backscattering coefficient at 565nm, and b is set at the first iterationbp(565) Is 0.0001, and b, which is iteratively updated in step 6-1, is used in subsequent iterative calculationsbp(565) The value is obtained.
Here, it should be explained that:
coefficient of backscattering (b)b) Defining: the ratio of the radiation flux scattered by the beam to the incident radiation flux over an angle of 90-180 deg.
bb(λ)=bbp(λ)+bbw(λ)
Wherein: bb(λ): the total backscattering coefficient of seawater; bbp(λ): the backscattering coefficient of the particulate matter; bbw(λ): pure water backscattering coefficient. It is now generally accepted that the backscattering of pure water is constant and relatively small; the Morel article list gives the values of the backscattering coefficients for different bands of pure seawater, e.g. bbw(lambda) 0.0038m at 400nm-1And the particle size is reduced to 0.0035m at 700nm-1And the research result is used up to now, b in the schemebw(565) And bbw(670) All using correlation coefficient values (Morel, A. (1974). Optical properties of pure water and pure sea water&E.Steeman-Nielsen(Eds.),OpticalAspects of Oceanography(pp.1–24).New York:AcademicPress.)。
Step 6: with an absorption coefficient a (lambda)0) Total backscattering coefficient bbRemote sensing reflectivity RrsAnd calculating to obtain the remote sensing reflectivity data residual through nonlinear iteration on the basis of the data residual radiation transmission relation equation (1), and specifically comprising the following steps:
step 6-1: the following system of equations was constructed:
b is obtained by using satellite remote sensing reflectivity data as input and utilizing the constructed absorption coefficient and backscattering coefficient extrapolation index neural network modelbp(565) And Δ Rrs(565) The information of (a);
wherein R isrs(565) And Rrs(670) Is satellite reflectivity data at 565nm and 670nm, a (lambda)0) Is the absorption coefficient, a (565) is the absorption coefficient calculated at 565nm, a (670) approximates the absorption coefficient of pure water, bb(670) The total backscattering coefficient calculated for 670nm is calculated from equation (2); bb(565) Can be according to the relation bb=bbp+bbwIs calculated as where bbwThe values of (c) can be referred to in the relevant literature as known constants, see step 5; wherein b isbp(565) Initial value 0.0001, updated in subsequent iterations, g0And g1Is a known empirical coefficient, g0And g1Respectively taking the value of 0.089sr-1And 0.1245sr-1A (565) and the backscatter coefficient extrapolation index Y can be refined from R containing the residual errorrsDeriving data; calculating DeltaR according to the formulars(565) And Δ Rrs(670);Rrs、△Rrs(565)、△Rrs(670)、a(565)、bb(565)、bb(670) The value of Y is updated in subsequent iterative calculations;
step 6-2: substituting S obtained by calculation in the step B into the formula (1) to obtain the residual spectrum relation of the remote sensing reflectivity data of 565nm and 670nm wave bands and the a (56) in the step 6-15) And Y information, and substituting into formula (3) in combination with formula (2), wherein formula (3) is changed to contain only Δ Rrs(565) And bb(565) Obtaining a delta R by solving a nonlinear system of equations for two variablesrs(565) And bb(565);
Step 6-3: further combining the spectral relation of the formula (1), calculating to obtain data residual error information of a visible light wave band, realizing preliminary correction of remote sensing reflectivity data residual error, and updating the remote sensing reflectivity used for a (565) and Y calculation in the step 6-1;
step 6-4: Δ R obtained when two adjacent iterations are calculatedrsAnd bbAre all less than 10-7If so, ending the iteration and repeatedly executing the steps 2-6; updating DeltaRrsAnd the correlation value is used for realizing the data residual error correction of the satellite remote sensing reflectivity product.
In this embodiment, since the linear relationship of the S value on the left and right sides of the 565nm band is not consistent, two S values, S, need to be calculatedrg(corresponding to the red band) and Sbg(corresponding to the blue band), FIG. 2 shows the S derived from the neural network modelbgAnd SrgAll 50000 data were extracted using satellite images analyzed using the clean water method, in comparison to their known values, corresponding to the blue-green (a) and green-red (b) regions of known values, respectively. As a result, it was found that SbgThe values varied from-0.5292 to 1.0288, and the neural network model was from satellite RrsAn efficient method of deriving spectral slope coefficients from the data. It is particularly noted that the model-derived spectral slope coefficient is relatively consistent with the known value, and SbgAnd SrgThe corresponding measured coefficients are 0.62 and 0.63, respectively, and it can be seen that the use of the neural network model can account for the change in spectral slope coefficient in the global open ocean of over 62%.
FIG. 3 shows a residual spectrum correction algorithm based on simulated data and actual measured data; (a) each of (c), (d) - (f) and (g) - (i) contains an error RrsResidual error comparison; the results showed that even at 443nm, 490nm and 565nm bands, 13.98%, 13.86% and 36.95% random noise, IDAS, was increasednnInverted RrsAnd Δ RrsWith the known RrsAnd Δ RrsAnd (5) the consistency is achieved. For example, regression and algorithm derived Δ R are knownrsThe slope and deviation of the linear regression between is about 0.90 and<0.00002sr-1,R2greater than 0.74, RMSE (root mean square error) less than 0.0006sr-1The results show that using the IDASnn algorithm can cause the residual error of the three COCTS visible bands in the synthetic data to vary by more than 74%.
FIG. 4 shows HY-1C secondary GAC R of Sargassum at 21 days 6 months in 2021rs(565) Model derived bbp(565): FIG. 4(a) and FIG. 4(b) two residual error correction algorithms IDAScw(existing Algorithm) and IDASnn(Algorithm of the invention) bbp(565) Image (unified as 10)-3m-1) (ii) a FIG. 4(c) and FIG. 4(d) are bbp(565) IDAS after image correctioncwAnd IDASnnA CV value of (D); FIG. 4(e) and FIG. 4(f) are bbp(565) Histogram and CV value histogram of original and corrected image, white representing cloud layer, via IDASnnAfter the algorithm (i.e. the method of the present invention) is corrected, the spatial consistency and smoothness of the image are significantly improved (see fig. 4b and 4d), IDASnnAlgorithm generated Rrs(565) R of image displayrs(565) The distribution was tighter than the original data, whereas the CV values (fig. 4c, 4d and 4f) had been reduced from 0.2116 to 0.0751. The results show that IDASnnThe algorithm can improve the COCTS R in the blue water arearsThe CV value (coefficient of variation) is an index for evaluating the effect of the present technique, among the data quality of the image.
FIG. 5 uses (a) IDAScwAnd (b) IDASnnDaily average b after residual correction by the algorithmbp(565) Data received by the data of 54-56W, 24-26N HY-1C secondary GAC in North Atlantic circumfluence of 2019 and 2020. The circle represents the daily average bbp(565) The vertical line represents the average deviation of 3 × 3 pixels at the center of the position, representing the smoothness of the daily image, and it is clear that IDAScwB of the Algorithmbp(565) The product has higher time variability than the IDASnn algorithm. bbp(565) Annual average CV value of data from IDAScwAlgorithm 0.312 down to IDASnn0.158 of the algorithm, all these studies showed that with IDASnnImplementation of the Algorithm, poor nutrient WaterB of (a)bp(565) The data product is improved.
In addition, it should be noted that in the present scheme, besides calculation by constructing a neural network model, a genetic algorithm, a deep learning algorithm, and the like may also be adopted, and without much description, calculation may be performed by replacing with other algorithms on the premise of not departing from the core idea.
The above description is only a preferred embodiment of the present invention, and not intended to limit the present invention in other forms, and any person skilled in the art may apply the above modifications or changes to the equivalent embodiments with equivalent changes, without departing from the technical spirit of the present invention, and any simple modification, equivalent change and change made to the above embodiments according to the technical spirit of the present invention still belong to the protection scope of the technical spirit of the present invention.
Claims (2)
1. A satellite remote sensing reflectivity product data residual error correction method is characterized by comprising the following steps:
step A, measuring remote sensing reflectivity data R by satellitersObtaining data residual error information of satellite remote sensing reflectivity products by comparing in-situ observation with satellite detection to form remote sensing reflectivity residual error delta RrsA data set;
step B, establishing remote sensing reflectivity data residual error delta R by analyzing remote sensing reflectivity residual error spectrum relationrsA relation to the wavelength λ;
wherein, Δ RrsRepresenting remote sensing reflectivity data residual error, C is a positive constant, S represents the spectral slope coefficient of the residual error, lambda represents wavelength, lambda represents the spectral slope coefficient of the residual error0Represents a reference band; setting reference waveband remote sensing reflectivity residual error delta Rrs(λ0) For calculating the remote sensing reflectivity residual error Delta R of other wave bandsrsA value;
step C,Remote sensing of reflectivity data R by satellitersTaking a spectral slope coefficient S of a remote sensing reflectivity data residual as an output for input, establishing a spectral slope coefficient neural network model of the residual to calculate the value of S;
step D, simulating the remote sensing reflectivity residual error delta R by combining the remote sensing reflectivity data residual error probability distribution and the spectrum relation characteristic on the basis of the radiation transmission simulation datarsTo establish an absorption coefficient a (lambda)0) And a neural network model of the backscattering coefficient extrapolation index Y, and solving a (lambda)0) And Y;
step E, determining an extrapolation wave band lambda1Total backscattering coefficient b ofb(λ1) The extrapolated band is defined by a reference band λ0Extrapolation gives:
wherein, bbwKnown pure water backscattering coefficient, bbpIs the backscattering coefficient of the particulate matter; setting the back scattering coefficient b of the particles at the extrapolation wavebandbp(λ1) Calculating λ1Total backscattering coefficient of (b)b(λ1);
Step F, using absorption coefficient a (lambda)0) Total backscattering coefficient bbRemote sensing reflectivity data RrsAnd remote sensing reflectivity residual error Δ RrsAnd (3) calculating to obtain the final remote sensing reflectivity residual error through nonlinear iteration on the basis of the relation (1) of the wavelength lambda.
2. The residual error correction method for the satellite remote sensing reflectivity product data according to claim 1, characterized in that: the iterative process of the step F is as follows:
step F1: setting an initial DeltaRrs(λ0) And bbp(λ0) Value of (A)
Step F2: if it is the first iteration, use the initial Δ R of step F1rs(λ0) Obtained by calculating step CSubstituting S into the formula (1) to obtain the satellite visible light waveband remote sensing reflectivity data residual error, realizing the preliminary correction of the data residual error, and updating the satellite remote sensing reflectivity data Rrs(ii) a If not, Δ R in step F5 is usedrs(λ0) Carrying out data residual error correction and updating satellite remote sensing reflectivity data Rrs;
Step F3: satellite remote sensing reflectivity data R obtained in step F2rsFor input, according to the absorption coefficient and backscattering coefficient extrapolation index neural network model constructed in the step D, a (lambda) is obtained0) And Y;
step F4: if it is the first iteration, use initial b of step F1bp(λ0) Value, in combination with equation (2) to solve bb(λ1) If not, using b in step F5bp(λ0) Calculation of bb(λ1) The information of (a);
step F5: combining a (lambda) in step F30) And Y information, while combining b of step F4b(λ1) Substituting the information into the formula (3) to change the formula (3) to contain only DeltaRrs(λ0) And bb(λ0) A system of equations for two variables, by solving the system of equations, Δ R is obtainedrs(λ0) And bb(λ0);
Wherein, bb=bbp+bbw,bbwIs a known value, g0And g1Is a known empirical coefficient;
step F6: and then calculating and obtaining data residual error information of a visible light wave band according to the spectrum relation of the formula (1), realizing remote sensing reflectivity data residual error correction, and updating the data residual error information for a (lambda)0) And the remote sensing reflectance calculated by Y;
step F7: Δ R obtained when two adjacent iterations are calculatedrs(λ0) And bb(λ0) Are all less than 10-7If so, ending the iteration and repeatedly executing the steps B-F; updating DeltaRrsAnd the correlation value is used for realizing the data residual error correction of the satellite remote sensing reflectivity product.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111096999.0A CN113849490A (en) | 2021-09-18 | 2021-09-18 | Residual error correction method for satellite remote sensing reflectivity product data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111096999.0A CN113849490A (en) | 2021-09-18 | 2021-09-18 | Residual error correction method for satellite remote sensing reflectivity product data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113849490A true CN113849490A (en) | 2021-12-28 |
Family
ID=78974527
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111096999.0A Pending CN113849490A (en) | 2021-09-18 | 2021-09-18 | Residual error correction method for satellite remote sensing reflectivity product data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113849490A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050027475A1 (en) * | 2003-07-29 | 2005-02-03 | Elgar Desa | Method of determining the volume scattering function of ocean waters in the backward direction using a satellite ocean color sensor |
CN102914505A (en) * | 2012-09-27 | 2013-02-06 | 国家海洋局第二海洋研究所 | Method for distinguishing marine red tide algaes by using MODIS ocean color remote sensing data |
CN106442233A (en) * | 2016-09-21 | 2017-02-22 | 南京信息工程大学 | Novel mixed method for estimating particle size of seawater suspended particulate matter by means of satellite data |
CN109187385A (en) * | 2018-08-22 | 2019-01-11 | 河南科技大学 | A kind of method of high light spectrum image-forming identification microorganism |
CN111460681A (en) * | 2020-04-17 | 2020-07-28 | 南京信息工程大学 | Satellite remote sensing method for monitoring depth of true light layer of offshore water body |
-
2021
- 2021-09-18 CN CN202111096999.0A patent/CN113849490A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050027475A1 (en) * | 2003-07-29 | 2005-02-03 | Elgar Desa | Method of determining the volume scattering function of ocean waters in the backward direction using a satellite ocean color sensor |
CN102914505A (en) * | 2012-09-27 | 2013-02-06 | 国家海洋局第二海洋研究所 | Method for distinguishing marine red tide algaes by using MODIS ocean color remote sensing data |
CN106442233A (en) * | 2016-09-21 | 2017-02-22 | 南京信息工程大学 | Novel mixed method for estimating particle size of seawater suspended particulate matter by means of satellite data |
CN109187385A (en) * | 2018-08-22 | 2019-01-11 | 河南科技大学 | A kind of method of high light spectrum image-forming identification microorganism |
CN111460681A (en) * | 2020-04-17 | 2020-07-28 | 南京信息工程大学 | Satellite remote sensing method for monitoring depth of true light layer of offshore water body |
Non-Patent Citations (1)
Title |
---|
张凯;陈军屹;祝佰航;李振春;丁鹏程;: "基于程函方程的VTI介质初至波走时层析反演", 石油地球物理勘探, no. 06, 15 December 2018 (2018-12-15) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Schneising et al. | Three years of greenhouse gas column-averaged dry air mole fractions retrieved from satellite–Part 1: Carbon dioxide | |
CN111795936B (en) | Multispectral remote sensing image atmospheric correction system and method based on lookup table and storage medium | |
Hoogenboom et al. | Retrieval of chlorophyll and suspended matter from imaging spectrometry data by matrix inversion | |
Pan et al. | An improved spectral optimization algorithm for atmospheric correction over turbid coastal waters: A case study from the Changjiang (Yangtze) estuary and the adjacent coast | |
Li et al. | Atmospheric correction of geostationary satellite ocean color data under high solar zenith angles in open oceans | |
Brajard et al. | Atmospheric correction of MERIS data for case-2 waters using a neuro-variational inversion | |
CN110927120B (en) | Early warning method for vegetation coverage | |
CN110909491A (en) | Sea surface salinity inversion algorithm based on wind and cloud meteorological satellite | |
CN111191380B (en) | Atmospheric aerosol optical thickness estimation method and device based on measurement data of foundation spectrometer | |
Babin et al. | Ocean Colour Remote Sensing in Polar Seas. | |
CN111198162B (en) | Remote sensing inversion method for urban surface reflectivity | |
Andrade et al. | An assessment of semi-analytical models based on the absorption coefficient in retrieving the chlorophyll-a concentration from a reservoir | |
Fan et al. | Machine learning algorithms for retrievals of aerosol and ocean color products from FY-3D MERSI-II instrument | |
Zhang et al. | Turbidity dynamics of large lakes and reservoirs in northeastern China in response to natural factors and human activities | |
Mélin | From validation statistics to uncertainty estimates: Application to VIIRS ocean color radiometric products at European coastal locations | |
Zhao et al. | Retrieval of red solar-induced chlorophyll fluorescence with TROPOMI on the Sentinel-5 precursor mission | |
CN114218786A (en) | On-orbit polarization radiation characteristic inversion method for non-polarization satellite sensor | |
CN113763272A (en) | Remote sensing inversion method for photosynthetic effective radiation attenuation coefficient of eutrophic lake | |
CN111650128B (en) | High-resolution atmospheric aerosol inversion method based on surface reflectivity library | |
CN113849490A (en) | Residual error correction method for satellite remote sensing reflectivity product data | |
Matishov et al. | Satellite estimation of chlorophyll-a concentration and phytoplankton primary production in the Sea of Azov | |
KR20210018737A (en) | Apparatus and method for calculating optical properties of aerosol | |
Razali et al. | Quantifying haze from satellite using haze optimized transformation (HOT) | |
Wu et al. | A novel semi-empirical soil multi-factor radiative transfer model for soil organic matter estimation based on hyperspectral imagery | |
Kudo et al. | Global 3-D distribution of aerosol composition by synergistic use of CALIOP and MODIS observations |
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 |