Paddy rice is one of the most important crops in the world.Accurate estimation and monitoring of paddy rice phenology is necessary for management and yield prediction.Remotely sensed time-series data are essential for estimation of crop phenology stages across large areas.Here,the paddy rice phenological stages(i.e.,transplanting,tillering,heading,and harvesting)were detected in Jiangxi Province,China.A comparison study was conducted using ground observation data from 10 agricultural meteorological stations,collected between 2006 and 2008.The phenological stages were detected using Moderate Resolution Imaging Spectroradiometer(MODIS)time-series enhanced vegetation index(EVI)data.Savitzky-Golay filter and wavelet transform were used to reduce the noise in the time-series EVI data and reconstruct the smoothed EVI time-series profile.Key phenological stages of double-cropping rice were detected using the characteristics of the smoothed EVI profile.The root mean square errors(RMSEs)for each stage were ±10 days around the ground observation data.The results suggest that Savitzky-Golay filter and wavelet transform are promising approaches for reconstructing high-quality EVI time-series data.Moreover,the phenological stages of double-cropping rice could be detected using time-series MODIS EVI data smoothed by Savitzky-Golay filter and wavelet transform.
Li ShihuaXiao JiangtaoNi PingZhang JingWang HongshuWang Jingxian
