Monitoring Of Wheat Powdery Mildew By Using Hyperspectral Remote Sensing And Dynamics Of Airborne Conidia Of Blumeria Graminis F. Sp. Tritici

Wheat powdery mildew is one of the world-wide foliar disease of wheat caused by Blumeria graminis f. sp. tritici. Conidia dispersed predominantly by wind are the main inoculum source for the disease epidemics, growing resistant cultivars and fungicide are the main measures for disease control it's useful for control disease pertinently by strengthening disease monitoring and predicting. In this paper, wheat powdery mildew in fields was monitored by using hyperspectral and volumetric spore trap in 2007 and 2008, respectively. The main results were as follows:There were differences between canopy reflectance of health wheat at different growth stages with its maximum at the heading stage, and fungicide had no effect on canopy reflectance. As disease index increasing, reflectance in near infrared band decreased significantly, and there were highly significant negative correlations between them in both cultivars and years at the anthesis stage and the milk-filling stage. derivative transform indicated that derivative reflectances in near infrared band were negative correlations with disease index. And significant correlations were existed between red edge slope, area of Red Edge and disease index. Several vegetation indices such as DVI, RVI, NDVI, SAVI were chosen, and DVI, SAVI had high the absolute values of correlation coefficients with disease index. As a result, we can estimate disease in fields by using reflectance in near infrared band (NIR), area of Red Edge (SDr), DVI and SAVI. Based on above results, the models of relationships between disease index and hyperspectral parameters were constructed for both varieties, and no significant differences were found between varieties at different growth stages. But for the same cultivar, there were significant differences between the two years at the anthesis stage, while at the initial filling stage, the differences between years were not significant.Based on fungicide had no significant impact on chlorophyll content, when wheat was damaged by powdery mildew, chlorophyll content decreased significantly, however, rates of decreasing were different in varieties with different resistance levels. There was highly significant correlation between chlorophyll content and DI, and significant correlation between chlorophyll content and green band. The models based on multiple bands were better than that based on single band. We can estimate chlorophyll content by using reflectance in green and red band. Red edge parameters were also well correlated with chlorophyll content, and the area of red edge was higher than the others.Yield loss was mainly caused by the reduction of 1000-kernel weight, and protein content was reduced significantly when wheat was infected by B. graminis f. sp. tritici. There were the highest relationships between yield or protein content and disease at heading stage, however, the highest relationship between1000-kernel weight and disease was at the anthesis stage. Correlations between spectral parameters at filling stage and yield, 1000-kernel weight or protein content were significant, the correlation coefficients of DVI, R760-850 and these three components were all above 0.7, so yield prediction models were constructed based on DVI and R760-850, respectively. As wind speed increased with height above the ground became higher in or above the canopy, there was difference between conidium numbers at different heights. During the seasonal dynamics of aerial conidia of B. graminis f. sp. tritici, numbers of conidia in the air increased with disease expanding, and reduced after reached peak in both years, whereas in daily dynamics, conidia dispersed in daytime were more than that in evening. The number of conidia in the air was related with meteorological factors, and among the weather variables considered, airborne conidial numbers was significantly and positively correlated with mean temperature and negatively with vapour pressure deficit, relative humility and wind speed had effects on conidial numbers. Conidial numbers in the air was closely related to the weather conditions on the same day and also on previous days. According to the highly correlationships between disease index and cumulative number of conidia in the air, disease prediction models based on conidium numbers at the same period or seven days before and cumulative number of spores at the same period or seven days before were constructed, respectively. And there were no significant difference among models based on cumulative number of conidia at the same period at different regions in 2 years, prediction model was constructed based on cumulative number of spores at the same period at different regions in 2 years: y=19.187Ln(x)-44.871.