Application Of Remote Sensing And Pathogen Conidia Trap For Monitoring Of Wheat Powdery Mildew

Wheat powdery mildew is one of the important diseases on wheat production. In this study, hyperspectral and unmanned aerial vehicle remote sensing technology was used to detect wheat powdery mildew at different planting densities. The losses on yield,1000-kernel weight and crude protein content caused by the disease and the estimations on them by using remote sensing technology were also studied. Two7-day volumetric spore samplers were used to monitor the dynamics in concentrations of Blumeria graminis f. sp. triciti (Bgt) conidia and their relationship to local weather conditions was analyzed. Disease prediction models based on the concentration of Bgt spores and meteorological factors were constructed. The concentrations of Bgt spores in the air estimated by microscopy on tapes and those determined with the Real-time PCR assay were compared. The main results are as follows.1. Diease indexes of powdery mildew in flowering, early and late filling stages have significant negative correlations with red edge slope(dλred) and the area of the red edge peak (SDr) at two planting densities. Only SAVI and DVI were significantly correlated with disease indexes in all three growth stages for both planting densities. Spectrum parameters got from the plots at nomal planting density have more significant correlations than those at1/2nomal planting density. According to the results of correlation analyses, several spectrum parameters (NIR, dλred, SDr, DVI&SAVI) were choosed to construct disease detection models for both density levels at different gowth stages. Moreover, there were no significant differences in slope and intercept of the constructed models between two density levels, data at both planting densities were gathered up to construct new disease detection models. RED, RVI in early filling stage, RED, NDVI in flowering stage and RED, RVI in flowering stage were used to construct the loss estimation models of wheat yield,1000-kernel weight and crude protein content.2. All parameters extracted from digital images of the plots at1/2nomal planting density were significantly correlated with disease indexes. However, in those plots at nomal planting density, the correlations between S,(R-B)/(R+B) and diease indexes are more significant than other parameters. Digital image parameters of the plots at1/2nomal planting density have more significant correlations than those at nomal planting density. Disease indexes estimation models based on S,(R-B)/(R+B) at200m, based on S,(G-B)/(G+B),(R-B)/(R+B) at300m and based on S,(R-B)/(R+B) at400m were constructed. Also, the yield estimation models based on S,(R-B)/(R+B) and the1000-kernel weight estimation models based on S,(G-B)/(G+B) were constructed after correlation analyses.3. The occurrence of wheat powdery mildew caused losses to wheat yield,1000-kernel weight and crude protein content. The models of the percentage of yield,1000-kernel weight and crude protein content losses were constructed using disease index (DI) at the critical point (CP) of growth stages and area under disease progress curve (AUDPC).4. The concentrations of Bgt conidia in the air increased with the disease index and reached the top at filling stage, then they began to decline and finally disappeared. The concentrations of Bgt conidia in the air have significant negative correlations with temperature, while other meteorological factors, such as relative humidity, rainfall and wind speed also have influence on them. Prediction models of Bgt conidial concentrations in the air based on meteorological factors were constructed using multiple regression analysis and time series analysis. Regression coefficients of linear model and ARIMA (1,1,0) model both reached significant levels. Conidial concentrations within the canopy were higher than those above the canopy, they were significant positive correlated. Three sets of models were derived by weekly-accumulated number of Bgt spores per cubic meter of air only, by weather variables only, and by both number of Bgt spores and weather variables to the disease index. Also models using inoculum7days before only, using weather variables7days before only, and using both inoculum and weather variables7days before were constructed for predicting disease indexes.5. A significant linear relationship between conidial concentrations counted with microscope and those determined with the real-time PCR assay was obtained, using the same samples of spore traps. Conidial concentrations determined with the real-time PCR assay was inherently lower than by counting with microscope.