Determination of rice canopy growth based on high resolution satellite images: a case study using RapidEye imagery in Korea

Abstract

Processing to correct atmospheric effects and classify all constituent pixels in a remote sensing image is required before the image is used to monitor plant growth. The raw image contains artifacts due to atmospheric conditions at the time of acquisition. This study sought to distinguish the canopy growth of paddy rice using RapidEye (BlackBridge, Berlin, Germany) satellite data and investigate practical image correction and classification methods. The RapidEye images were taken over experimental fields of paddy rice at Chonnam National University (CNU), Gwangju, and at TaeAn, Choongcheongnam-do, Korea. The CNU RapidEye images were used to evaluate the atmospheric correction methods. Atmospheric correction of the RapidEye images was performed using three different methods, QUick Atmospheric Correction (QUAC), Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH), and Atmospheric and Topographic Correction (ATCOR). To minimize errors in utilizing observed growth and yield estimation of paddy rice, the paddy fields were classified using a supervised classification method and normalized difference vegetation index (NDVI) thresholds, using the NDVI time-series features of the paddy fields. The results of the atmospheric correction using ATCOR on the satellite images were favorable, which correspond to those from reference UAV images. Meanwhile, the classification method using the NDVI threshold accurately classified the same pixels from each of the time-series images. We have demonstrated that the image correction and classification methods investigated here should be applicable to high resolution satellite images used in monitoring other crop growth conditions.