Evaluation of Paddy Field Drainage Using Sentinel-2 Satellite Data

Abstract

To support subsurface drainage planning using pipes in paddy fields, we have developed a method to determine the drainage quality for each paddy field using Sentinel-2 satellite data and GIS polygon data showing paddy field parcels. The areas of interest are paddy field parcel polygons where rice has been harvested and wheat has not been planted after harvesting rice. First, parcel polygons, whose parcel mean of image pixel values of the near-infrared band data is above a certain threshold value, are determined as not cultivated and are excluded. Next, the remaining parcel polygons, whose parcel mean of image pixel values of the short-wave infrared band data is less than a certain threshold, are determined to have bad drainage. The remaining parcels are determined to have good drainage. The results obtained using the Sentinel-2 satellite data taken on February 21, 2021 (6 days after heavy rain) were collated with visually interpreted results using the aerial photograph taken on February 20, 2021. The overall correct answer rate (area basis) obtained by assuming the latter judgement result as the true value was 84%. Since the developed method uses the dry and wet conditions of the paddy field soil surfaces as a judgment index, it has the disadvantage that the result is dependent on the number of days between heavy rain events and satellite observations.