Estimating Tea Yield Using Multirotor-type UAV Remote Sensing

Abstract

In this study tea yields were estimated using vegetation indices such as the normalized difference vegetation index (NDVI). These calculations were made from the aerial image data of tea fields taken by a multirotor unmanned aerial vehicle (UAV，drone) equipped with a multispectral camera. There was a high correlation between NDVI and the yield of tea during the growing period of first flush (R²=0.61-0.97). Another vegetation index, green NDVI, showed a varying correlation depending on observation conditions such as the test day and location of the field (R²=0.002-0.91), while normalized difference red edge index (NDRE) showed a high correlation (R²=0.65-0.99). The root mean squared error of the tea yield obtained using both NDVI and NDRE was as small as 40kg/10a or less. These results suggest that the vegetation indices NDVI and NDRE obtained by the aerial image data can be used to estimate the tea yield. Since regression lines between yield and vegetation index were different for each survey, the correction under different observation conditions with additional explanatory variables such as the accumulated temperature is a subject for future research.