Models Of Estimating Nutrient Elements In Leaves And Canopy Of Juglans Regia Using Hyperspectral Data

Juglans regia is one of the main economic tree species in the characteristic forest and fruit industry of Tarim Basin around Xinjiang.For a long time,due to the lack of rapid,timely and non-destructive methods for the diagnosis of nutrient elements in foliar,the development of a large area of tree nutrition diagnosis and scientific fertilization in southern Xinjiang has been restricted.Therefore,on the scale of foliar and canopy,the Juglans regia in Hongqipo area of Aksu,Xinjiang was studied.The spectral characteristics of Juglans regia were analyzed from different observation periods by using the method of spectrum analysis and indoor chemical analysis.The spectral sensitive band of nutrient content and the hyperspectral estimation model of nutrient content were studied.The main results are as follows:The spectral characteristics showed that the spectral reflectance of foliar was mainly affected by pigment and tissue structure,and there was no significant difference during the observation period.The spectral reflectance of canopy was related to the reflectance of ground objects,water content,canopy structure and vegetation growth and development level,and the spectral reflectance in June was the highest,followed by September,August and July.The contents of nutrient elements in the foliar of Juglans regia were different,and the dynamic changes of the foliar were different in different periods.In June and July,the moving speed of nutrient elements in the foliar of Juglans regia was faster due to the Juglans regia in the fruit-setting stage and the fast-growing period.There was a great difference in content,but in August and September,the content fluctuated little and was in a relatively stable period.The correlation analysis between foliar nutrient element content and foliar and canopy spectral analysis showed that the correlation between spectrum and nutrient element content could be improved by using normalized first-order differential pretreatment method.The contents of nutrient elements in foliar and canopy of Juglans regia had different spectral sensitive bands.Because of the influence of foliar pigment and cell structure,the spectral sensitive bands of foliar are mostly distributed in visible and near infrared bands.Under the influence of pigment,illumination,canopy structure,ground cover reflection and water vapor,the sensitive band of canopy spectrum is distributed in the whole spectral range of 350～2500 nm.The study on the hyperspectral estimation model of the nutrient elements content in foliar and canopy shows that the principal component regression method is used to establish the nutrient element content model at the foliar and canopy level.The fitting degree of each nutrient element spectral model of Juglans regia foliar in different periods was above 0.60,and the prediction accuracy was more accurate when the nutrient content was higher.The optimum forecasting period of element content of Mg and Ca is June,and the best prediction period of Cu is July,Mn,P,K is August and Fe,Zn is September.The fitting degree of each nutrient element spectral model of Juglans regia canopy in different periods was above 0.49,and the fitting effect was slightly lower than that of foliar level.The best modeling effect of N element is in June and P,K,Ca,Mg,Mn elements are in August.Fe Cu and Zn have the best prediction effect in September.The influence of canopy structure and ground cover reflection may be the main factor that the spectral model of the canopy is lower than the foliar level.