Optics and Precision Engineering, Volume. 32, Issue 11, 1773(2024)
Establishment and optimization of aerial multispectral field straw mulch quantity inversion model
Figures & Tables(14)
Fig. 2. Experimental process for determination of staw mulch quantity
Fig. 3. UAV route planning and multi-spectral camera whiteboard calibration
Fig. 5. Correlation coefficients between multi-spectral variables and straw mulch quantity in different periods
Fig. 6. Scatter plot of measured versus and predicted values for different data
Fig. 8. Scatter plot of measured versus and predicted values for optimized model
Fig. 9. Comparison of corn straw mulch quantity estimates at sampling sites
Table 1. Spectral index and its formula
View tableView in ArticleTable 1. Spectral index and its formula
Spectral index Formula Reference NDVI NDVI=（NIR-R）/（NIR+R） ［ 28 ］DVI DVI=NIR-R ［ 29 ］SR SR=NIR/R ［ 30 ］GNDVI GNDVI=（NIR-G）/（NIR+G） ［ 31 ］SMI SMI=NIR×（NIR-G）×（R-G） ［ 32 ］SRRE SRRE=RE/NIR ［ 30 ］NDRE NDRE=（NIR-RE）/（NIR+RE） ［ 33 ］RNDI RNDI=（RE-R）/（RE+R） ［ 33 ］
Table 2. Description of modeling set and validation set
View tableView in ArticleTable 2. Description of modeling set and validation set
Data set Sample Study area 1 Study area 2 Mean/kg Min/kg Max/kg Standard deviation/kg Autumn Spring Spring Modeling set 126 42 42 42 0.248 0.120 0.373 0.066 Validation set 63 21 21 21 0.225 0.073 0.395 0.067
Table 3. Correlation coefficients between multi-spectral variables and straw mulch quantity in different study areas
View tableView in ArticleTable 3. Correlation coefficients between multi-spectral variables and straw mulch quantity in different study areas
Multi-spectral variable Study area 1 |r| Study area 2 |r| B 0.347 7* 0.36* G 0.330 2* 0.345 6* R 0.445 8** 0.437 8** RE 0.438 6** 0.439 2** NIR 0.401 6** 0.398** NDVI 0.555 4*** 0.558 9*** DVI 0.530 5*** 0.505 9*** SR 0.577 6*** 0.560 6*** GNDVI 0.506 6*** 0.533 8*** SMI 0.344 5* 0.328 3* SRER 0.607 5*** 0.610 7*** NDRE 0.558 1*** 0.553 2*** RNDI 0.481 1** 0.479**
Table 4. Machine learning models based on different sets of variables
View tableView in ArticleTable 4. Machine learning models based on different sets of variables
Model algorithm Variable group Modeling set Validation set R2c RMSEc R2v RMSEv SVM Band group 0.45 0.078 0.36 0.086 Spectral index group 0.51 0.072 0.50 0.081 All variable group 0.58 0.070 0.47 0.071 RF Band group 0.64 0.066 0.67 0.061 Spectral index group 0.66 0.059 0.69 0.056 All variable group 0.65 0.065 0.64 0.060 BPNN Band group 0.42 0.076 0.35 0.087 Spectral index group 0.55 0.068 0.47 0.073 All variable group 0.59 0.072 0.45 0.075 ELM Band group 0.58 0.070 0.33 0.088 Spectral index group 0.65 0.060 0.47 0.082 All variable group 0.63 0.063 0.43 0.076
Table 5. Estimation of straw mulch quantity in study area
View tableView in ArticleTable 5. Estimation of straw mulch quantity in study area
Research area Area/hm2 Ms/t Es/t As/% Es/% Study area 1， autumn 2.55 17.952 16.524 91.36 7.95 Study area 1， spring 2.55 15.572 14.076 89.37 9.61 Study area 2， spring 2 11.360 12.587 90.25 10.80
Tools
Get Citation
Copy Citation Text
Yuanyuan LIU, Yu SUN, Xuebing GAO, Libin WANG, Yueyong WANG, Mengqi LIU, Shuran CUI. Establishment and optimization of aerial multispectral field straw mulch quantity inversion model[J]. Optics and Precision Engineering, 2024, 32(11): 1773
Paper Information
Category:
Received: Feb. 19, 2024
Accepted: --
Published Online: Aug. 8, 2024
The Author Email: Yueyong WANG (yueyongw@jlau.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20