Abstract: Abstract: Lodging of cereal crops due to strong wind and rain is a common agricultural disaster in many regions of the world, leading to crop growth retardation and yield reduction. Area of crop lodging is very critical for agricultural hazard assessment and agricultural insurance claim. A survey method of lodging using high resolution remote sensing data acquired from unmanned aerial vehicle or satellite is fast and efficient. Worldview-2 multi-spectral image has a 1.8 m spatial resolution in 8 bands (from 400 to1040 nm), which is a type of useful remote sensing data of agricultural information in many aspects, but there are few researches about lodging survey. Therefore, a maize lodging survey using Worldview-2 image was discussed in this paper, which was aimed to find an optimum method for estimating the lodging area of maize. In 2012, a Worldview-2 image with the swath of 5 km × 5 km was acquired on September 14th after a lodging event occurred on September 12th in Wanzhuang Agricultural Park of Chinese Academy of Agricultural Sciences, and the precise area of lodging at 3 fields was measured using the photogrammetry of unmanned aerial vehicle. In this paper, the features of spectrum and texture were analyzed for finding out the optimum features and bands in classification. Firstly, typical pixels of normal maize (285 pixels) and lodging maize (284 pixels) were sampled. Spectrum was acquired from multi-spectral image after ortho-rectification and atmospheric correction. Mean texture feature was acquired from reflectance data by applying co-occurrence based filter. The spectral curves showed that there was a significant difference between normal maize and lodging maize. The reflectance of maize increased after lodging, and the values of the 3 bands including red edge, infrared 1 and infrared 2 were greater than 0.1. The change of canopy structure might be the main reason of this phenomenon. Comparing texture feature of lodging maize to normal maize, the values of the 8 bands between the 2 kinds of maize were different, and especially the band of green, red edge, infrared 1 and infrared 2. According to the feature analysis, red edge, infrared 1 and infrared 2 were chosen as the optimum bands for classification in this study. Then, 8 classification methods were carried out by using different features, bands and algorithms of classification. The type of features included spectrum and texture, the bands used for classifying were respectively 3 optimum bands and all 8 bands, and the algorithms of classification were respectively based on Mahalanobis distance and the maximum likelihood. Finally, an optimum method was chosen according to the error. The estimation results of 8 methods showed that, the errors of those methods that used 3 optimum bands were lower than those using 8 bands, and when using the 3 optimal bands, the errors of those methods based on the algorithm of the maximum likelihood were lower than those based on the algorithm of Mahalanobis distance. According to the performance of 8 methods, the method that used reflectance of red edge, infrared 1 and infrared 2 based on the maximum likelihood was the best one, for which the minimum error was 2.2%, the maximum error was 8.9%, and the average of error was 4.7%. According to this study, using Worldview-2 multi-spectral image can estimate the area of lodging maize accurately based on remote sensing data of unmanned aerial vehicle.