Measuring canopy nitrogen nutrition in tobacco plants using hyper spectrum parameters Yong Zou, Xiaoqing YE, et al. Shenzhen Tobacco Ind. Co., Ltd. of CNTC
Layout Background Experimental Program Experimental Results Conclusions
Background as real-time, real-field and no-harm diagnostic technology of nitrogen is deficient for flue-cured tobacco production in China. Therefore, inefficient use of nitrogen fertilizer has become a common phenomenon, resulting in poor utilization of nitrogen fertilizer and decline of tobacco leaf quality.
a simple, rapid and non-destructive monitoring method of crop nitrogen is needed for nitrogen precision management. As we know, ground-based remote sensing has been widely applied in crop agronomic parameters monitoring and extraction. Therefore, monitoring tobacco plant nitrogen accumulation by using of canopy reflectance has theoretical significance and practical value.
Experimental Program Crop monitered was tobacco, cultivar K326. Spectrometry adopted the multispectral field portable radiometer, MSR-16R. Flue-cured tobacco after topping was to be measured under cloudless or less cloud weather. Randomized block design was used with nitrogen as variable at four levels, namely (N), 15(N1), 15(N2), 195 kg hm-2 (N3).
The total biomass, aboveground weight, root weight, total fresh leaves weight, total dry leaves weight, total leaves area were measured. the Leaf Nitrogen Accumulation, Plant Nitrogen Accumulation, Root Nitrogen Accumulation were calculated with the following formula.
Experimental Program Calculation L LNA = L LNC L LDW....(1) P PNA = P PNC P PDW (2) and R RNA =R RNC R RDW.. (3) L LNA Leaf Nitrogen Accumulation(g N/m 2 ); L LNC Leaf Nitrogen Content(%); L LDW Leaf Dry Weight(g/m 2 ); P PNA Plant Nitrogen Accumulation(g/m 2 ); P PNC Plant Nitrogen Content(%); P PDW Plant Dry Weight(g/m 2 ); R RNA Root Nitrogen Accumulation(g N/m 2 ); R RNC Root Nitrogen Content(%); R RDW Root Dry Weight(g/m 2 ).
Algorithm of different multispectral indices Spectral Parameters Abbreviation Calculation Formula Ratio Vegetation Index RVI(λ 1,λ 2 ) R λ1 /R λ2 Difference Vegetation Index DVI(λ 1,λ 2 ) R λ1 -R λ2 Normalized Difference Vegetation Index NDVI(λ 1,λ 2 ) R R λ1 λ1 + R R λ 2 λ 2 Enhanced Vegetation Index EVI (1 + R 2.5( R NIR NIR + 6R R 68 68 ) 7.5R 46 ) Red edge position wavelength Soil-adjusted Vegetation Index Optimization of Soil Adjusted Vegetation Index λ rep SAVI OSAVI 71 + (R + R 66) / 5 R 76 R 1.5 R 81 2 71 ( R87 R 68) 87 + R 68 +.5 1.6( R81 R 68) ( R81 R 68 +.16) R 71
Results Original spectrum changes of flue-cured tobacco canopy 7 6 5 4 3 2 1 46 51 Reflectance/% 56 61 66 N N1 N2 N3 68 71 76 81 Wavelength/nm 87 95 11 122 13 15 Flue-cured tobacco canopy reflectance under different N fertilization levels 165 8 7 6 5 4 3 2 1 46 51 Reflectance/% 56 61 April 28th May 17th May 29th June 6th June 2th 66 68 71 76 81 87 Wavelength/nm 95 11 122 13 15 Flue-cured tobacco canopy reflectance in different periods 165
Results Relating ANA to canopy reflectance 1.8.6.4.2 -.2 -.4 -.6 -.8-1 Correlation coeffici ANA LNA SNA t.5 t.1 wavelength/nm 46 51 56 61 66 68 71 76 81 87 95 11 122 13 15 165 Correlation between ANA and spectra reflectance of flue-cured tobacco
Results Relating ANA to canopy reflectance 7 6 5 4 3 ANA/g.m-2-2 y = 1.2674x +.2992 R 2 =.9386 2 1 LNA/g.m -2 1 2 3 4 5 Correlation between LNA and ANA
Results Relating ANA to canopy spectra index 6 5 4 3 2 1 Nitrogen concentrat RVI(81,68) RVI y =.18x 1.9791 R2 =.922 5 1 15 2 Correlation between RVI and LNA 7 6 5 4 3 2 1 Nitr ogen concentr at DVI(81,68) y =.4e.1334x R 2 =.8746 DVI 1 2 3 4 5 6 Correlation between DVI and LNA
Results Relating ANA to canopy spectra index 6 NDVI(81,68) 6 EVI81 5 4 3 2 1 Nitrogen concentrat y = 9.3899x 7.5838 R 2 =.8425 NDVI.2.4.6.8 1 Correlation between NDVI and LNA 5 4 3 2 1 Nitrogen concentrat y = 3.9623x 2-1.949x + 7.6741 R 2 =.8125 EVI.5 1 1.5 2 2.5 3 Correlation between EVI and LNA
Results Relating ANA to canopy spectra index 6 SAVI 6 OSAVI 5 4 3 2 1 Nitrogen concentrat y = 35.749x 2-69.746x + 34.27 R 2 =.8639 SAVI.5 1 1.5 Correlation between SAVI and LNA 5 4 y = 7E-15x 1659.8 3 R 2 =.9119 2 1 OSAVI 1.1545 1.155 1.1555 1.156 1.1565 1.157 Nitrogen concentrat Correlation between OSAVI and LNA
Results Relating ANA to Dλ rep 6 5 4 3 2 1 Nitrogen concentrat y = 8.376x 2.91 R 2 =.8997 Dλ rep.2.4.6.8 Correlation between Dλ rep and LNA
Regression models based upon vegetation indices against LNA and its validation Vegetation Indices Regression Model Model Building Coefficient of Determination Model Verification Coefficient of Determination RMSE RVI y =.18x 1.9791.92 **.91 **.264 DVI y =.4e.1334x.875 **.792 **.245 NDVI y = 9.3899x 7.5838.843 **.839 **.193 EVI 81 y=7.6741+3.9623x 2-1.949x.813 **.929 **.188 SAVI y=34.27+35.749x 2-69.746x.864 **.921 **.22 OSAVI y = 7E-15x 1659.8.912 **.93 **.26 Dλ rep y = 8.376x 2.91.9**.898**.197
Conclusions There existed an absorption valley in the vicinity of 68 nm and 46 nm, a reflection peak at about 56 nm and a high reflectance platform in the near infrared region of 81-11 nm. Canopy spectral reflectance of flue-cured tobacco decreased in the visible region (46-71 nm) and increased significantly in the near-infrared band ( 76-122 nm) as nitrogen level increased.
RVI (68,81), DVI (68,81), NDVI (68,81), EVI81, SAVI (68,81) OSAVI (68,81) and Dλ Red all could be used to establish good tobacco nitrogen accumulation regression models, and the EVI81, OSAVI(68,81) and Dλ Red had better fitting effect, which significantly improved the inversion accuracy. It could be concluded that ANA in flue-cured tobacco could be monitored effectively by key vegetation indices, especially EVI 81, OSAVI and DλRed. The regression model were y=7.6741+ 3.9623x 2-1.949x, y = 7E-15x 1659.8, y = 8.376x 2.91 respectively.
Thank You!