CHEN Bing, WANG Fang-Yong, HAN Huan-Yong, LIU Zheng, DENG Fu-Jun, LIN Hai, YU Yu, LI Shao-Kun, WANG Ke-Ru, XIAO Chun-Hua
Cotton Science. 2013, 25(3): 254-261.
Using data from red-edge parameters of haperspectra, we endeavored to provide an expedient way to extract leaf nitrogen contents(LNC) of cotton infected with Verticillium wilt, and to lay the groundwork for estimating cotton yield infected by Verticillium wilt using remote sensing technology. The relationship between LNC and red edge parameters were analyzed, and diagnose models of spectra red edge parameters were established for cotton leaves infected by Verticillium wilt. The main results are as follows: (1) With the increase of leaf severity level, LNC little by little decreased, difference is significant. (2) In all red edge parameters, REP, Dr, Lo, Depth672 and Area672 all decreased, and the degree of decrease was maximum to Area672 value, the degree of decrease was minimum to Dr value; but Lwidth largely increased. (3) LNC had best significant positive correlations with REP, Lo, Depth672 and Area672, had best significant negative correlations with Lwidth, and no best significant correlations with Dr. (4) Diagnose models of LNC to disease cotton leaves on the basis of spectra red edge parameters all attached best significant correlations (P<0.01). The diagnose models based on Area672 had best estimated precision for LNC, and R2 were over 0.7, RMSE less than 0.6, RE less than 0.007. So using red edge parameters of haperspectra monitor LNC of cotton infected by Verticillium wilt is accurate and convenient, and those models can better diagnose LNC.