Cold Injury Detection And Assessment For Camellia Sinensis By Terahertz Spectroscopy And Hyperspectral Imaging

China is a leader in the world's tea?Camellia Sinensis?production.However,tea plants are grown in humid and warm hilly regions,prone to cold damage at freezing temperatures.In order to increase the production,it is essential to protect the plant against cold damage through rapid freezing injury detection techniques which will lead to assessment of the semi lethal temperature and the critical cold temperature as well as the identification of cold hardy cultivars.At the moment,it is challenging to come up with the most effective cold injury detection technique which is rapid and can provide the needed information for effective frost protection strategies to help reduce production loss to tea growers in the country.In recent times,terahertz?THz?spectroscopy and hyperspectral imaging have been used to evaluate the change in the physiological properties of several foods and crops due to freezing.In some cases,they are combined with other techniques for comprehensive analysis of such properties.Injury detection is a unique case of excellent THz wave technology application.Therefore,THz technology can be used to discriminate cold injured leaf from normal leaf because of their unique spectral fingerprints in THz band.Besides,there is a direct correlation between the degree of cell damage and the water that escapes out of the cold injured cell which could make THz spectral response proportional to the degree of the leaf injury.Similarly,water comes out of the leaf when frozen,reducing the optical path length through which the THz wave is transmitted and can lead to higher absorptions.Though,tea leaf samples are from the same cultivar,the absorption spectra of the samples are likely to differ due to differences in the chemical compositions of the samples.Therefore,the main research content and results of this study are as follows:?1?Freezing tolerance of different tea cultivars?Maolv,Zhong Cha 102 and Zhong Cha 108?was assessed by relative electrical conductivity?REC?,the index of injury and lethal temperature(LT₅₀)when the tissues were subject to 0 to-10°C.Subsequently,terahertz spectra responses to physiological changes in tea leaves under cold stress were assessed.The results showed that REC increased with decreasing temperature and LT₅₀ of Maolu,Zhong Cha 102 and Zhong Cha 108were-3.38°C,-6.11°C and-8.92°C,respectively.Zhong Cha 108,Zhong Cha 102 and Maolu had11.4?10^-8,9.8?10 ^8-and 8.8?10 ^8-as principal power peaks respectively.The average power had obvious drop and appeared to recover subsequently within typical range of temperatures.The possible critical cold temperature could be found within such temperature ranges of-3 to-5°C for Maolu,-5 to-7°C for Zhong Cha 102 and-6 to-8°C for Zhong Cha 108.The trend of the terahertz response made it possible to rank the cold tolerance of the tea plants as?Zhong Cha108>Zhong Cha 102>Maolu?.Terahertz appears as a promising technique that could be utilised to quantify the physiological changes in leaves.?2?Protection to tea plants against cold injury requires knowledge of nutrients effect on freezing tolerance.The influence of 0,25,50,75 and 100%NPK dosage on cold tolerance were evaluated by terahertz time domain spectroscopy based on the absorption spectral as potential analytical tool.Subsequently,quantitative determination of the cold injury was improved by combining the original partial least-squares regression?PLSR?model with two dimensional correlation spectroscopy?2DCoS?model.Absorbance of cold injury with different NPK dosage was obtained by THz-TDS in the band between 0.1THz and 1.6THz.The results showed that average absorption increases with increase in NPK dosage.Subsequently,the 25,50,75 and 100%NPK dosage could reduce the cold tolerance of the control plant by 2.92,4.58,6.31 and 6.72%respectively.The trend indicates no significant reduction after 75%dosage which suggests that NPK could improve tea cold hardness but only from 75%dosage upwards.The quantitative assessment result obtained for the 2DCOS-PLSR?R_p=0.9963,RMSE=0.05548?,is considerably better than the original PLSR?R=0.9912,RMSE=0.0632?.The technique highlighted the spectra differences of NPK dosage,improved the accuracy of detection result with excellent quantitative analysis.?3?Cold injury?CI?threatens the quality and successful cultivation of tea and so requires quick non-destructive technique to assess the CI for effective protection.As a result,the absorbance of tea leaf subjected to 4,0,-2.5,-5.0,-7.5 and-10.0°C freezing temperatures was measured with terahertz?THz?spectroscopy in the band of 0.1-1.6THz.To enhance quantitative analysis and improve separation of overlap signals resulting from scattering of injured leaf cells,two-dimensional correlation spectroscopy?2DCOS?was introduced into partial least square regression?PLSR?and the performance evaluated.The mean absorption plotted against the temperatures showed a slight decrease from 4.0 to 0°C with an obvious decrease at-2.5°C and a sharp increase from-5.0 to-10°C.The-2.5°C was critical to the crop so the lowest possible critical temperature could be found around-2.5°C.The quantitative assessment result obtained for the 2DCOS-PLSR(R_2D=0.8873,0.9113 and 0.9963;RMSE-_2D=0.6032,0.5763 and0.5221),is significantly better than AI-PLSR(R_AI=0.8647,0.8707 and 0.9473;RMSE-_AI=0.6038,0.5962 and 0.5797).The combination of THz spectroscopy and 2DCOS-PLSR provided better benchmark for the input interval selection and improved the accuracy of detection result.?4?Protection to tea plants against cold injury requires knowledge of nutrients effect on freezing tolerance.The influence of 0,30,60 and 100%N concentrations and 0 to-10°C,on freezing injury and cold tolerance by hyperspectral imaging based on the reflectance as potential analytical tool was evaluated.Subsequently,quantitative determination of the cold injury was improved by combining the original partial least-squares regression?PLSR?model,principal component analysis?PCR?and Linear model.Reflectance of cold injury with different N concentration was obtained by hyperspectral imaging system in the band between 900-1800 nm.The results showed that average reflectance increases with increase in N dosage.Amongst the five specific spectral domains tested,the best predictive accuracy for tea leaf cold injury was achieved by PCR? R²=0.9971,RMSE=0.0609?in 1410-1766 nm wavelength followed by LM? R²=0.9999,RMSE=0.0805?in 900-100 nm.The selected wavelength range where all the models had optimal predictions accuracies is 1296-1380 nm followed by 1001-1295 nm.The whole interval had the worse prediction accuracies which could be due to large variation in the data for a specific treatment and high absorption peak occurring around 1450 nm wavelength in the reflectance curve.Taking the average R² and RMSE calculated for all the three statistical models showed that the worse prediction accuracy occurred in 1410-1766 nm followed by the whole interval 900-1766 nm.The low prediction accuracy could be due to the strong water absorption peaks that appeared in both range of wavelengths.