Prediction of water depth in paddy fields using water index calculated from images taken by UAV

<p>If the spatial characteristics of rice paddies and waterways in relation to water management—specifically, water-depth—can be grasped in an areal and timely manner, then an important environmental consideration for aquatic organism habitat analysis can hence be used. This study verifies whether the water-depths of paddy fields and waterways can be successfully predicted using water indices calculated using a low-altitude-remote- sensing (LARS) approach. A DJI Matrice 300 system equipped with a MicaSense Altum was attached to an unmanned aerial vehicle (UAV) and used to capture images of paddy fields in Sogabe, Kameoka City, Kyoto Prefecture. The resulting images were processed using photogrammetry to produce orthomosaic representations. A normalized difference water index (NDWI) was calculated from near-infrared and green wavelengths, with water-depths of subject paddy fields and waterways then accordingly measured within 1h immediately after the capturing the image using the UAV. The NDWIʼs of the water-depth measurement points were extracted using geographic-information-system technology, with a linear regression equation and coefficient of determination between water-depth and NDWI ultimately calculated. NDWI values were relatively high for flooded paddy fields and there were NDWI differences even for paddy fields with only a single section. Moreover, NDWI values were relatively high for waterways, indicating that NDWIʼs can be used to help understand the heterogeneity of water-depths within paddy fields and evacuated water areas, such as waterways. Although predicting the water depth by generalizing one regression equation and collectively predicting water-depths during an entire irrigation season through the methods used in this study is difficult, accurately assessing individual-case water-depths depending on paddy field conditions within a discrete study area, especially during the frog breeding season and the start of mid-drying season is possible. Such determinations will ultimately provide critical data in support of new aquatic habitat evaluations for a variety of Kyoto Prefecture wetland settings.</p>