Relationship between NDVI and Canopy Cover Sensed by Small UAV under Different Ground Resolution

Abstract

Canopy Cover (CC) is a significant indicator of crop development and estimation of the evapotranspiration volume of crop leaves within crop simulation models. During the last three decades, monitoring CC for crops using Normalized Difference Vegetation Index (NDVI) obtained from satellite sensors has been studied worldwide. However, a few studies have estimated the CC of crops using NDVI by UAVs. One of UAV imagery’s crucial advantages is a high resolution of less than 0.10 m, while the resolution of satellite imagery is usually larger than 10 m. Now that the UAV has become a popular method in agriculture science, it is necessary to prove the interchangeability of UAV and satellite imagery of monitoring CC. In this study, small UAVs took RGB and multispectral images of the experimental peanuts field in Hokkaido. Orthomosaic and reflectance map of the field were constructed using the UAV imagery and then were obtained CC and NDVI values with GIS software. CC was calculated as the green canopy area ratio, extracted from the orthomosaic using a GIS supervised classification tool. CC was compared with NDVI values under various resolutions of 0.50 m, 1.0 m, 2.5 m, 5.0 m, and 10 m. The NDVI showed a highly correlated linear relationship with CC under each ground resolution from 0.10 m to 10 m (R² led a range of 0.88** to 0.94**). The shapes of NDVI and CC’s regression equations closely resembled each other, with the slopes of 1.06 to 1.11 and the intercepts of 0.247 to 0.250, respectively. From the result of ANCOVA, the UAV imagery resolution has no significant impact on NDVI and CC’s relationship. Although more irrelevant factors, such as soil and mulching seat, got included within one pixel of the images, the regression equations stayed the same with the increased ground resolution.