Extensive analysis of the periods when irrigation starts in individual paddy fields using Sentinel-2 satellite data

Understanding the periods when irrigation starts in paddy fields is important for grasping the current situation of usage of paddy irrigation water. I developed a method to extensively analyze periods of irrigation using multiple Sentinel-2 satellite data, multispectral images available free of charge. In this method, the periods when irrigation starts for individual paddy fields was determined by the following procedures; (1) improvement of spatial resolution of short-wave infrared band data by the pan-sharpening method, (2) creation of Modified Normalized Difference Water Index (MNDWI) images, (3) creation of binary images by the Otsu' threshold method, (4) determination of whether individual paddy fields are flooded on each satellite observation day, (5) determination of whether irrigation has started in individual paddy fields on each satellite observation day, (6) determination of the period when irrigation starts in each paddy field. The periods when irrigation started in individual paddy fields located in the target area were determined by this method using the eight Sentinel-2 satellite data observed from April to June, and paddy parcel boundary data. As a result, the paddy fields where irrigation started from April 14th to April 20th were the most numerous. In addition, accuracy verification and a breakdown analysis of the results for judging whether irrigation had started was conducted for each paddy field in part of the target area. As a result, the correct determination of whether irrigation had started in each paddy field was 98% on May 15th. On June 4th, many paddy fields where rice plants were relatively large were judged as "no flooding". However, since they were judged as "flooded" at least once in April to May, they were judged to be already irrigated. Therefore, the periods when irrigation starts for individual paddy fields can be determined with high accuracy by applying this method.