Study Of Monitoring Sclerotinia Sclerotiorum Based On Satellite Remote Sensing

Object: During the whole growing period, we used domestic environmental satellite to research spectralindexes’ character of rapeseed’s health and disease fields, analyzed these spectral indexes relationship withagronomy parameters, aiming to exploring the theoretical method of monitoring rapeseed field’s growingand disease condition through the way of satellite remote sensing.Methods: First we ascertained the best agronomy parameters and12satellite spectral indexes to representthe rapeseed’s growth conditions in its each growth stage. Then during the rapeseed growing time in2011to2012, we determined these parameters and indexes’ value for sample points when the satellite flewthrough, and created mathematical model monitoring the growth condition by synthesizing them. Thehealthy and disease sample points’ date were compared, significance test were also done to inspect thosedate’s difference before and after disease occurred between healthy and disease fields. On the basis ofabove work, in the rape sclerotium disease high-incidence season, we used4kinds of supervisedclassification method of spectral parameters (The parallelepiped method, Mahalanobis distance, Minimumdistance method, Maximum likelihood method) detecting the disease. Finally classification accuracy testwas done.Results: Seedling stage to bud (22-62days after sowing), the LAI had extremely significantly related(r=0.906**) with rapeseed biomass, to be the best agronomic parameter for this period of rapeseedgrowing, it also had extremely significant correlations with all spectral indexes (except TSAVI),|r|>0.8**(n=76), RVI and DVI were two kinds of indexes had the minimum relative error to estimate LAI,errors were22.9%and22.4%respectively, and the fitting equations’ R~2>0.8.Bud to full flowering stage (62~78days after sowing), FAI had significant correlation with TSAVI,r=-0.261*(n=71), had extremely significant correlations with other spectral indexes,|r|>0.78**(n=71),NDVI were the index that had the minimum relative error to estimate FAI, errors were42.4%, and thefitting equations’ R~2>0.7.Full flowering stage (78~97days after sowing), PAI had extremely significant correlations with allspectral indexes (except TSAVI),|r|>0.52**(n=52), TVI were the index that had the minimum relativeerror to estimate PAI, errors were18.7%, and the fitting equations’ R~2>0.4.After disease occurred, disease fields’ agronomy parameters and satellite spectral indexes bothshowed large different, significant t-test results showed that the NDVI, RVI, and DVI had no significantdifference before disease, but had extremely significant difference after disease occurred. Maximumlikelihood method’s result were closest to actual situation, in4kinds of supervised classification method ofspectral parameters. Its Kappa coefficient is0.72, reached substantial level.Conclusion: According to rapeseed growth feature, using satellite spectral data and vegetation parameters,which can represent the rapeseed’s growth conditions best, to do quantitative analysis and created spectralvegetation index fitting LAI, FAI and PAI estimating model, in phases, as all the results had good accuracy,it’s indicated that using domestic environmental satellite to realize the real-time monitoring the conditions of rapeseed growing is feasible. After rape sclerotium disease occurred in the fields, the key spectralvegetation indexes, which were on behalf of the growing conditions, NDVI, RVI and DVI all hadextremely significant difference with healthy fields. Among supervised classification methods of spectralparameters, Maximum likelihood method’s result were closest to actual situation. It was indicated thatusing spectral vegetation indexes of domestic environmental satellite to realize the monitoring rapesclerotium disease is feasible.