Retrieval of Planetary Boundary Layer Height by Remote Sensing Fusion Based on Deep Forest

Zhongxing Zhao; Songlin Fu; Junjie Chen; wei Xie

doi:10.3788/AOS241185

Acta Optica Sinica, Volume. 45, Issue 6, 0628004(2025)

Retrieval of Planetary Boundary Layer Height by Remote Sensing Fusion Based on Deep Forest

Zhongxing Zhao... Songlin Fu*, Junjie Chen and wei Xie |Show fewer author(s)

Author Affiliations

College of Physics and Electronic Information Engineering, Zhejiang Normal University, Jinhua 321000, Zhejiang , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(17)

Fig. 1. Multi-grained scanning process

Download full size

Fig. 2. Schematic diagram of cascade forest process

Download full size

Fig. 3. Schematic diagram of improved cascade forest

Download full size

Fig. 4. Atmospheric boundary layer height results obtained by improved fcForest inversion

Download full size

Fig. 5. Inversion results of atmospheric boundary layer height under clear and cloudless weather

Download full size

Fig. 6. Inversion results of boundary layer height under cloudy weather

Download full size

Fig. 7. Comparison of inverted ABLHs of different methods for clear weather. (a) MPL; (b) DPL; (c) fcForest

Download full size

Fig. 8. Comparison of inverted ABLHs of different methods for cloudy weather. (a) MPL; (b) DPL; (c) fcForest

Download full size

Fig. 9. Comparison of inverted ABLHs of different methods during day. (a) MPL; (b) DPL; (c) fcForest

Download full size

Fig. 10. Comparison of inverted ABLHs of different methods at night. (a) MPL; (b) DPL; (c) fcForest

Download full size

Table 1. Comparison of experimental results of random forest models using different datasets
View table
Table 1. Comparison of experimental results of random forest models using different datasets
Dataset R RMSE /km MAE /km
Fusion remote sensing dataset 0.926 0.194 0.135
Micropulse lidar dataset 0.906 0.217 0.148
Doppler lidar dataset 0.910 0.213 0.144

Table 2. Comparison of multi-grained scanning results and original datasets

View table

Table 2. Comparison of multi-grained scanning results and original datasets

Condition	R	RMSE /km	MAE /km
Original dataset	0.926	0.194	0.135
Window scale 2	0.761	0.339	0.232
Window scale 4	0.864	0.259	0.175
Window scale 6	0.895	0.229	0.154
Window scale 8	0.894	0.230	0.155
Window scale 10	0.890	0.233	0.156

Table 3. Feature importance of feature vectors in primitive dataset

View table

Table 3. Feature importance of feature vectors in primitive dataset

Feature	Importance	Choose
MPL_1	0.4007	Yes
MPL_2	0.0112	Yes
MPL_3	0.0085	Yes
MPL_4	0.0078	No
MPL_5	0.0072	No
MPL_6	0.0097	Yes
MPL_7	0.0080	No
MPL_8	0.0085	Yes
MPL_9	0.0089	Yes
MPL_10	0.0095	Yes
DPL	0.1653	Yes
Temperature	0.0472	Yes
Pressure	0.0202	Yes
Speed	0.0145	Yes
Humidity	0.0815	Yes
Hour	0.1913	Yes

Table 4. Comparison of inversion results of feature screening dataset and original dataset
View table
Table 4. Comparison of inversion results of feature screening dataset and original dataset
Dataset R RMSE /km
Original dataset 0.926 0.194
Feature screening dataset 0.929 0.190

Table 5. Experimental results of boundary layer height inversion by different methods

View table

Table 5. Experimental results of boundary layer height inversion by different methods

Method	R	RMSE /km	MAE /km
Gradient method	0.783	0.419	0.302
Threshold value method	0.806	0.384	0.226
Wavelet covariance transform method	0.768	0.445	0.333
Standard deviation method	0.665	0.539	0.413
Random forest	0.926	0.194	0.135
Gradient boosting	0.920	0.201	0.142
Deep forest	0.891	0.234	0.156
Feature screening cascade forest	0.935	0.182	0.126

Table 6. Experimental results of boundary layer height inversion by different methods
View table
Table 6. Experimental results of boundary layer height inversion by different methods
Time 5：41 11：23 17：30 23：30
Sonde 446 648 565 366
MPL 1041 382 742 2300
DPL 105 495 375 165
fcForest 479 723 684 520

Table 7. Atmospheric boundary layer height results at different time
View table
Table 7. Atmospheric boundary layer height results at different time
Time 5：37 11：27 17：29 23：45
Sonde 386 815 956 474
MPL 2151 2330 1881 1551
DPL 105 585 105 105
fcForest 491 920 880 504

Tools

Get Citation

Copy Citation Text

Zhongxing Zhao, Songlin Fu, Junjie Chen, wei Xie. Retrieval of Planetary Boundary Layer Height by Remote Sensing Fusion Based on Deep Forest[J]. Acta Optica Sinica, 2025, 45(6): 0628004

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Remote Sensing and Sensors

Received: Jun. 19, 2024

Accepted: Aug. 16, 2024

Published Online: Mar. 21, 2025

The Author Email: Fu Songlin (fu_songlin@zjnu.edu.cn)

DOI:10.3788/AOS241185

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Comparison of experimental results of random forest models using different datasets

Table 1. Comparison of experimental results of random forest models using different datasets

Table 2. Comparison of multi-grained scanning results and original datasets

Table 2. Comparison of multi-grained scanning results and original datasets

Table 3. Feature importance of feature vectors in primitive dataset

Table 3. Feature importance of feature vectors in primitive dataset

Table 4. Comparison of inversion results of feature screening dataset and original dataset

Table 4. Comparison of inversion results of feature screening dataset and original dataset

Table 5. Experimental results of boundary layer height inversion by different methods

Table 5. Experimental results of boundary layer height inversion by different methods

Table 6. Experimental results of boundary layer height inversion by different methods

Table 6. Experimental results of boundary layer height inversion by different methods

Table 7. Atmospheric boundary layer height results at different time

Table 7. Atmospheric boundary layer height results at different time

微信扫一扫：分享