Aerosol particle type identification method based on optimal estimation
Through the optimal estimation method and decision tree model, the lidar inversion data is converted into a state vector, which solves the uncertainty and model universality of aerosol type recognition in the prior art, and realizes the accurate identification of aerosol type and Concentration-independent parameters are obtained.
Patent Information
- Application Number
- CN202410957678.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-11-26
AI Technical Summary
The existing aerosol particle type identification method based on lidar has a high dependence on characteristic parameters, resulting in high uncertainty in classification results. The aerosol characteristic parameters caused by the variety of lidar models are different, making it difficult to establish universality and ability. Transplant classification model.
Using the optimal estimation method, the parameter information obtained by lidar inversion is converted into a state vector through the decision tree and the forward model, and linear approximation is performed using the Jacobian matrix to establish an objective function to minimize errors and realize vertical retrieval of aerosol type. .
Aerosol optical and microphysical parameters independent of concentration were obtained, which improved the accuracy of aerosol type identification and the universality of the model, and reduced the uncertainty of classification results.
Smart Images

Figure CN119026026A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of particle type identification, and in particular to an aerosol particle type identification method based on optimal estimation. Background Art
[0002] Aerosol type information is of great significance for studying the radiative forcing of the ground-atmosphere system, improving radiation transfer and forecasting models, and providing pollution warnings and governance decisions. The use of lidar can obtain the optical microphysical property characteristics of aerosols with high temporal and spatial resolution, providing favorable data support for particle type identification.
[0003] At present, the identification of aerosol particle types based on lidar is mainly based on optical characteristic parameters (such as extinction scattering ratio), depolarization ratio, Angstrom index, backscattering color ratio, etc. Using one or more of the above parameters, mathematical methods such as K-means clustering, decision tree classification and artificial neural network are used to realize the aerosol type identification of single-wavelength or multi-wavelength lidar. However, the above identification methods are highly dependent on characteristic parameters. Only relying on sparse observation samples to create an aerosol type index library will bring great uncertainty to the classification results. In addition, due to the diversity of lidar models, the aerosol characteristic parameters that can be generated are also different. It is necessary to establish a more universal and portable classification model.
[0004] The present invention establishes a new aerosol classification method based on lidar data. After a systematic calibration, quality control and inversion process, the optical and microphysical parameters of aerosols that are independent of concentration are obtained, and the aerosol type in the vertical direction is retrieved using the optimal estimation method. Summary of the invention
[0005] In view of this, the present invention provides an aerosol particle type identification method based on optimal estimation.
[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:
[0007] A method for identifying aerosol particle types based on optimal estimation, characterized by comprising the following steps:
[0008] Step 1: Using the most recognizable optical parameters, the aerosol component information that the target may contain is estimated through the decision tree method, and the first initial value S of the state vector is returned;
[0009] Step 2: Write the parameter information obtained by lidar inversion into a state vector matrix Xa = [LR387, PDR355, PDR532, BAE]. The relationship between S and Xa can be established through the forward model F: S = F (Xa) + ε, where ε is the error term;
[0010] Step 3: The target to be estimated by the model is S, and the observable is Xa. It is necessary to establish the objective function J to minimize the error between S and K*Xa, that is, the objective function J = (S-KXa) T *(S-KXa)+λK T K;
[0011] Step 4: Exploitation To minimize the objective function, we can obtain the global optimal estimate S, which is the proportion of the five basic types of aerosols in the real atmosphere.
[0012] Preferably, in step 2, the laser radar inversion process is:
[0013] Taking Raman-polarization lidar as an example, the inversion method of Raman-polarization lidar is mainly based on the atmospheric Mie scattering signal and vibration Raman scattering signal it receives. The lidar equations of these two signals can be expressed as follows:
[0014]
[0015] in and are the laser radar Mie scattering echo power signal and nitrogen Raman echo power signal respectively; z is the detection distance, and is the radar system constant, O(z) is the radar geometric overlap factor, and are the backscattering coefficients of aerosol and atmospheric molecules at the height z when the emission wavelength is λ0, and α is the same; N R (z) is the molecular density of nitrogen at height z, is the Raman scattering cross section of nitrogen.
[0016] Preferably, the optical parameters of step 1 include: extinction coefficient α aer , backscatter coefficient β aer , depolarization ratio, and backscattering related Angstrom exponent BAE.
[0017] Preferably, the extinction coefficient
[0018] Preferably, the backscatter coefficient
[0019] Preferably, the backscatter-related Angstrom index is
[0020] Preferably, it also includes an aerosol lidar ratio: S aer (z,λ0)=α aer (z,λ0) / β aer(z,λ0)
[0021] Preferably, in step 2, the forward model is based on aerosol mixing theory, such as the aerosol depolarization ratio δ can be written as: Where j is the aerosol type index, S j is the volume percentage of each component, and the other parameters are the same.
[0022] Preferably, in step 2, since there is a nonlinear relationship between Xa and S, linear approximation can be performed through the Jacobian matrix, that is: S=K*Xa+ε, where K is the Jacobian matrix.
[0023] Compared with the prior art, the present invention has achieved the following technical effects:
[0024] The present invention establishes a new aerosol classification method based on lidar data. After a systematic calibration, quality control and inversion process, the optical and microphysical parameters of aerosols that are independent of concentration are obtained, and the aerosol type in the vertical direction is retrieved using the optimal estimation method. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 This is an algorithm technology roadmap for an aerosol particle type identification method based on optimal estimation in the present invention. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0027] Embodiment 1:
[0028] A method for identifying aerosol particle types based on optimal estimation, characterized by comprising the following steps:
[0029] Step 1: Use the most recognizable optical parameters, where the optical parameters include: extinction coefficient α aer , backscatter coefficient β aer , depolarization ratio, and backscattering related Angstrom exponent BAE.
[0030] Extinction coefficient
[0031] Backscatter coefficient
[0032] Backscatter-related Angstrom exponent
[0033] Also includes aerosol lidar ratio: S aer (z,λ0)=α aer (z,λ0) / β aer (z,λ0),
[0034] The aerosol component information that the target object may contain is estimated through the decision tree method, and the first initial value S of the state vector is returned;
[0035] Step 2: Write the parameter information obtained by lidar inversion into a state vector matrix Xa = [LR387, PDR355, PDR532, BAE]. The relationship between S and Xa can be established through the forward model F: S = F (Xa) + ε, where ε is the error term;
[0036] Among them, the laser radar inversion process is:
[0037] Taking Raman-polarization lidar as an example, the inversion method of Raman-polarization lidar is mainly based on the atmospheric Mie scattering signal and vibration Raman scattering signal it receives. The lidar equations of these two signals can be expressed as follows:
[0038]
[0039] in and are the laser radar Mie scattering echo power signal and nitrogen Raman echo power signal respectively; z is the detection distance, and is the radar system constant, O(z) is the radar geometric overlap factor, and are the backscattering coefficients of aerosol and atmospheric molecules at the height z when the emission wavelength is λ0, and α is the same; N R (z) is the molecular density of nitrogen at height z, is the Raman scattering cross section of nitrogen;
[0040] Among them, the forward model is based on the aerosol mixing theory. For example, the aerosol depolarization ratio δ can be written as: Where j is the aerosol type index, S j is the volume percentage of each component, and the other parameters are the same.
[0041] Step 3: The target to be estimated by the model is S, and the observable is Xa. It is necessary to establish the objective function J to minimize the error between S and K*Xa, that is, the objective function J = (S-KXa) T *(S-KXa)+λK TK; Since there is a nonlinear relationship between Xa and S, linear approximation can be performed through the Jacobian matrix, that is: S = K*Xa+ε, where K is the Jacobian matrix.
[0042] Step 4: Exploitation To minimize the objective function, we can obtain the global optimal estimate S, which is the proportion of the five basic types of aerosols in the real atmosphere.
[0043] According to previous studies, outliers were removed from the above parameters, and the specific screening range is as follows:
[0044]
[0045]
[0046] In addition, in order to ensure that the model can provide comprehensive and typical aerosol characteristic information, the most suitable basic aerosol category must be selected. According to existing research, basic aerosol types are generally divided into five types: dust type, ocean type, urban type, polluted urban type, and biomass burning type. The following table is the statistical mean of different parameters of the five main aerosols based on existing observations:
[0047]
[0048] It can be seen from the above embodiments that the present invention establishes a new aerosol classification method based on lidar data. After a systematic calibration, quality control, and inversion process, aerosol optical and microphysical parameters that are independent of concentration are obtained, and the aerosol type in the vertical direction is retrieved using the optimal estimation method.
[0049] The above description is only a preferred embodiment of the present invention and does not limit the technical scope of the present invention. Therefore, any slight modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention are still within the scope of the technical solution of the present invention.
Claims
1. A method for identifying aerosol particle types based on optimal estimation, characterized in that: The following steps are involved: Step 1: Using the most recognizable optical parameters, the aerosol component information that the target may contain is estimated through the decision tree method, and the first initial value S of the state vector is returned; Step 2: Write the parameter information obtained by lidar inversion into a state vector matrix Xa = [LR387, PDR355, PDR532, BAE]. The relationship between S and Xa can be established through the forward model F: S = F (Xa) + ε, where ε is the error term; Step 3: The target to be estimated by the model is S, and the observable is Xa. It is necessary to establish the objective function J to minimize the error between S and K*Xa, that is, the objective function J = (S-KXa) T *(S-KXa)+λK T K; Step 4: Exploitation To minimize the objective function, we can obtain the global optimal estimate S, which is the proportion of the five basic types of aerosols in the real atmosphere.
2. The method for identifying aerosol particle types based on optimal estimation according to claim 1, characterized in that: In step 2, the laser radar inversion process is: Taking Raman-polarization lidar as an example, the inversion method of Raman-polarization lidar is mainly based on the atmospheric Mie scattering signal and vibration Raman scattering signal it receives. The lidar equations of these two signals can be expressed as follows: in and are the laser radar Mie scattering echo power signal and nitrogen Raman echo power signal respectively; z is the detection distance, and is the radar system constant, O(z) is the radar geometric overlap factor, and are the backscattering coefficients of aerosol and atmospheric molecules at the height z when the emission wavelength is λ0, and α is the same; N R (z) is the molecular density of nitrogen at height z, is the Raman scattering cross section of nitrogen.
3. The method for identifying aerosol particle types based on optimal estimation according to claim 1, characterized in that: The optical parameters of step 1 include: extinction coefficient α aer , backscatter coefficient β aer , depolarization ratio, and backscattering related Angstrom exponent BAE.
4. The method for identifying aerosol particle types based on optimal estimation according to claim 3, characterized in that: The extinction coefficient 5. The method for identifying aerosol particle types based on optimal estimation according to claim 3, characterized in that: The backscatter coefficient 6. The method for identifying aerosol particle types based on optimal estimation according to claim 3, characterized in that: The backscatter-related Angstrom exponent 7. The method for identifying aerosol particle types based on optimal estimation according to claim 2, characterized in that: Also includes aerosol lidar ratio: S aer (z,λ0)=α aer (z,λ0) / β aer (z,λ0).
8. The method for identifying aerosol particle types based on optimal estimation according to claim 1, characterized in that: In step 2, the forward model is based on the aerosol mixing theory. For example, the aerosol depolarization ratio δ can be written as: Where j is the aerosol type index, S j is the volume percentage of each component, and the other parameters are the same.
9. The method for identifying aerosol particle types based on optimal estimation according to claim 1, characterized in that: In step 2, since there is a nonlinear relationship between Xa and S, linear approximation can be performed through the Jacobian matrix, that is: S=K*Xa+ε, where K is the Jacobian matrix.
