![<?echo $_SERVER['SERVER_NAME'];?>](/template/twentyseventeen/skin/images/header.jpg)
The original name "UAV frame painting hyperspectral remote sensing application based on the fitting of leaf area index based on spectral features and PLSR"
Beijing Agricultural Information Technology Research Center/National Agricultural Information Technology Research Center/Key Laboratory of Agricultural Information Technology, Ministry of Agriculture
School of Geography and Ocean Sciences, Nanjing University
School of Surveying and Land Information Engineering, Henan Polytechnic University
Using winter wheat LAI as the research object, the UAV S185 hyperspectral image obtained from the booting stage, flowering stage and filling stage and the ground data (the winter wheat canopy ASD reflectivity and winter wheat LAI) obtained from the booting stage, the flowering stage and the filling stage were used to demonstrate the spectral characteristics (red edge parameters). Or the application value of the modified LAI fitting method combined with the partial least squares regression algorithm in the ULI detection of UAV format hyperspectral remote sensing.
Firstly, S185 and ASD are compared from the correlation of spectral reflectance correlation and vegetation index, and the accuracy of S185 data is verified. The results show that the hyperspectral data of S185 in the 3rd to 96th bands (458~830nm) is better. The spectral quality is suitable for detecting winter wheat LAI.
Secondly, the correlation between spectral features (6 planting index and 4 red edge parameters) and LAI is analyzed, and the traditional LAI fitting method and improved based on red edge parameters or vegetation index are sequentially determined by independent verification and cross-validation methods. The LAI prediction method of winter wheat LAI prediction accuracy is evaluated. Compared with the traditional LAI fitting method, the improved LAI fitting method can greatly improve the prediction accuracy of winter wheat LAI, especially PLSR+REPs.
From October 2014 to June 2015, the winter wheat experiment was carried out at the National Precision Agriculture Research and Demonstration Base in Changping District, Beijing. The nitrogen application in the experimental field is shown in Figure 1.
Figure 1 Overview of the geographical location of the study area and the variable fertilization experiment of winter wheat
N1~N4 indicates the nitrogen application level, which is 0, 195, 390, 585 kg/hm−2 in order.
Two model validation methods were used to objectively evaluate the LAI fitting method. Figure 2 shows that the overall sample, modeled samples, and validation samples of the first method are normally distributed, which lays the theoretical foundation for this study.
Figure 2 Normal distribution of the sample, modeled, and validated samples
For S185, the predecessors have initially verified the spectral reflectance accuracy, but the influence of the target features and the verification data source (ie the instrument used to collect the verification spectrum), Tian Minghao et al based on S185 and SVC HR-1024i The morphological difference analysis of cotton canopy spectrum of non-imaging full-spectrum ground spectrum spectrometer indicates that the spectral information of S185 1st to 100th bands (450~850 nm) is accurate and reliable. Although this conclusion has certain indications, for different crop types and verification data sources, it is still necessary to further explore the optimal range of LAI for UAV S185 data estimation. Considering the difference in spectral resolution between S185 and ASD, the ASD band is resampled to the S185 band and the correlation between the two is calculated;
The results show that the spectral reflectance of S185 and resampled ASD in the range of 3rd to 96th bands (458~830nm) is highly correlated: R2=0.996 at booting stage, R2=0.998 at flowering stage, and R2=0.996 at filling stage.
Based on this, the 6-plant index was calculated using ASD data and S185 data in the range of 458-830 nm, and the differences between VIASD and VIS185 were analyzed in more depth from the correlation.
The results showed that more than 93% of the VIASD and VIS185 samples were in the best estimation interval, and the correlation between VIASD and VIS185 was better, R2>0.8 (Fig. 3).
Figure 3 Correlation analysis of VIS185 and VIASD
Based on the above analysis, the canopy spectral information of winter wheat obtained by S185 has good radiation resolution and spectral quality in the range of 3rd to 96th bands (458~830nm), which can be used to estimate the winter wheat LAI.
Figure 4 shows that PLSR+REPs is more suitable for estimating LAI than PLSR+VIs. Independently, the R2 and RMSE of PLSR+REPs were increased by 0.048 and decreased by 0.068, respectively. Compared with PLSR+VIs, the R2 and RMSE of PLSR+REPs were increased by 0.061 and decreased by 0.091, respectively.
Figure 4 Comparison of LAI estimation accuracy based on PLSR+REPs and PLSR+VI
a and b are independent verifications; c and d are cross-validations
in conclusion
Compared with the traditional LAI fitting method, the improved LAI fitting method can make full use of the UAV S185 hyperspectral information to obtain higher precision LAI prediction values, and the LSR precision predicted by PLSR+REPs is higher than PLSR+VIs. It is expected to provide some ideas for the physical parameter detection of UAV hyperspectral remote sensing.
Attached to the original
Red Light Therapy is a quick, simple way to brighten your mood and energize your cells, leading to a wide range of clinically proven health benefits.
Near Infrared Light Therapy Device,Red Light Therapy,Led Red Light
Shenzhen Bonliter Optoelectronic Co., Ltd. , https://www.szbonliter.com