Aerosol Type Recognition Model Based on Naive Bayesian Classifier

Mei Zhou; Jianhua Chang; Sicheng Chen; Yuanyuan Meng; Tengfei Dai

doi:10.3788/AOS202242.1801006

Acta Optica Sinica, Volume. 42, Issue 18, 1801006(2022)

Aerosol Type Recognition Model Based on Naive Bayesian Classifier

Mei Zhou¹, Jianhua Chang^1,2、*, Sicheng Chen¹, Yuanyuan Meng¹, and Tengfei Dai^1,2

¹School of Electronics & Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China

²Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Seasonal mean values of 5 aerosol characteristic parameters at SGP station. (a) Seasonal mean values of $α_{e x t}$ , $α_{a b s}$ and $S_{s s a}$ ; (b) seasonal mean values of $m_{r}$ and $m_{i}$

Download full size

Fig. 2. Flow chart of overall algorithm

Download full size

Fig. 3. Flow chart of NBC model

Download full size

Fig. 4. Proportion of aerosol classification results of SGP station in four seasons from 2020 to 2021 obtained by NBC model. (a) Proportion of spring classification; (b) proportion of summer classification; (c) proportion of autumn classification;(d) proportion of winter classification

Download full size

Fig. 5. Depolarization ratio distribution of HSRL dust aerosol (from14:00 to 19:00 on March 29, 2021)

Download full size

Fig. 6. Time-sharing profile results of HSRL aerosol optical parameters. (a) Depolarization ratio profile; (b) lidar ratio profile; (c) color ratio profile

Download full size

Fig. 7. Observation results of aerosol depolarization ratio of HSRL in typical months of four seasons at SGP station from 2020 to 2021. (a) March 1 to 31, 2021; (b) August 25 to September 24, 2020; (c) November 1 to 31, 2020; (d) January 1 to 31, 2021

Download full size

Fig. 8. Seasonal average of aerosol optical parameters of high resolution lidar SGP station from 2020 to 2021. (a) Seasonal average of depolarization ratio; (b) seasonal average of lidar ratio;(c) seasonal average of color ratio

Download full size

Table 1. Centroids of five aerosol reference clusters

View table

Table 1. Centroids of five aerosol reference clusters

Aerosol type	$α_{e x t}$	$α_{a b s}$	$S_{s s a}$	$m_{r}$	$m_{i}$
Urban industry（UI）	1.76	1.15	0.96	1.40	0.005
Biomass burning（BB）	1.87	1.30	0.89	1.48	0.020
Dust（DU）	0.28	1.75	0.91	1.47	0.004
Marine（MA）	0.59	0.93	0.97	1.40	0.001
Mixed（MT）	1.32	1.20	0.92	1.45	0.011

Table 2. Misjudgment of various aerosols
View table
Table 2. Misjudgment of various aerosols
Aerosol type UI BB DU MA MT Actual number Recognition rate /%
Test number 117 119 156 70 171 633
UI 105 0 6 7 1 119 88
BB 0 102 0 0 4 106 96
DU 9 0 142 0 4 155 92
MA 3 0 0 63 0 66 95
MT 0 17 8 0 162 187 87

Table 3. Average values of various aerosol characteristics determined by NBC model
View table
Table 3. Average values of various aerosol characteristics determined by NBC model
Aerosol characteristic UI BB DU MA MT
N 119 106 155 66 187
$α_{e x t}$ 1.340 1.380 0.840 0.960 1.210
$α_{a b s}$ 1.050 1.190 1.370 0.990 1.160
$S_{s s a}$ 0.960 0.860 0.930 0.980 0.920
$m_{r}$ 1.510 1.560 1.550 1.510 1.530
$m_{i}$ 0.003 0.022 0.006 0.001 0.011

Table 4. Proportion of aerosol types in four seasons obtained by different classification algorithms

View table

Table 4. Proportion of aerosol types in four seasons obtained by different classification algorithms

Classification algorithm	Aerosol type	Spring	Summer	Autumn	Winter
NBC	Urban industry	31	21	15	22
	Biomass burning	4	8	35	15
	Dust	49	16	14	29
	Marine	1	23	2	10
	Mixed	15	32	34	24
Mahalanobis distance	Urban industry	23	18	12	21
	Biomass burning	2	6	23	10
	Dust	29	9	11	20
	Marine	1	17	2	6
	Mixed	45	50	52	43
Threshold criteria	Urban industry	27	14	10	17
	Biomass burning	5	6	31	11
	Dust	43	13	15	33
	Marine	4	28	4	9
	Mixed	21	39	40	30

Tools

Get Citation

Copy Citation Text

Mei Zhou, Jianhua Chang, Sicheng Chen, Yuanyuan Meng, Tengfei Dai. Aerosol Type Recognition Model Based on Naive Bayesian Classifier[J]. Acta Optica Sinica, 2022, 42(18): 1801006

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Atmospheric Optics and Oceanic Optics

Received: Jan. 20, 2022

Accepted: Apr. 22, 2022

Published Online: Sep. 15, 2022

The Author Email: Chang Jianhua (jianhuachang@nuist.edu.cn)

DOI:10.3788/AOS202242.1801006

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Centroids of five aerosol reference clusters

Table 1. Centroids of five aerosol reference clusters

Table 2. Misjudgment of various aerosols

Table 2. Misjudgment of various aerosols

Table 3. Average values of various aerosol characteristics determined by NBC model

Table 3. Average values of various aerosol characteristics determined by NBC model

Table 4. Proportion of aerosol types in four seasons obtained by different classification algorithms

Table 4. Proportion of aerosol types in four seasons obtained by different classification algorithms