Comparison of Vegetation Indices and Spectral Reflectance Observed by Two Types of UAV-mounted Multispectral Camera

概要

Poster
Poster 3

Comparison of Vegetation Indices and Spectral Reflectance
Observed by Two Types of UAV-mounted Multispectral Camera
Ryohei YOKOYAMA1, Chinatsu YONEZAWA2, Mizuhiko NISHIDA2, Ryosuke TAJIMA2, Muxiye2,
Koharu OKADA1, Kouki TAKAMURA3, Kaori AMAYA3 and Ken ICHIKAWA3
1

2

Faculty of Agricultural, Tohoku University, Japan
Graduate School of Agricultural Science, Tohoku University, Japan
3
Fukken Gijyutsu Consultants, Japan

Remote sensing in agriculture using multispectral cameras has increased in availability due to their advantages, such as high
flexibility, ease of operation, and high spatial resolution. Several multispectral cameras with UAVs have been developed and utilized.
The observed value can be converted to spectral reflectance and vegetation indices are calculated. However, the accuracy of the
observed reflectance should be confirmed for vegetation monitoring. In this study, we compared images obtained by the two types of
multispectral cameras mounted on UAVs, RedEdge MX-dual with Inspire-2 and P4 Multispectral (P4M) for paddy rice fields in the
Kawatabi Field Science Center. We obtained spectral reflectance and calculated vegetation indices such as Normalized Difference
Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), and Red Edge Normalized Difference
Vegetation Index (RENDVI). On August 5th, 2022, and September 13th, 2022, we observed target area almost simultaneously by
two cameras. Because of the battery duration, the region was separated into two portions and monitored individually. Radiometric
calibration was performed for the orthomosaic images by RedEdge MX-dual using a reflectance panel. P4M has a sunlight sensor,
and it is possible to obtain vegetation indices without calibration, however, spectral reflectance is uncertain. Therefore, radiometric
calibration was performed for the observation on September 13. For the comparison of RedEdge MX-dual and P4M, the average of
vegetation indices and spectral reflectance were computed on each of the agricultural parcels for paddy rice. The results show that
the spectral reflectance and RENDVI obtained by P4M without calibration were inconsistent with those of RedEdge MX-dual. The
differences were clearly visible, especially in the spectral reflectance. NDVI and GNDVI obtained by P4M without calibration on
September 13 corresponded to those obtained by the RedEdge MX-dual, however, RENDVI was smaller than that of the RedEdge
MX-dual. After the calibration, RENDVI and spectral reflectance, except for blue and red edge, acquired by the first flight of P4M
on September 13 almost corresponded to those obtained by RedEdge MX-dual. Blue was larger than that by RedEdge MX-dual and
red edge was smaller than that by RedEdge MX-dual. RENDVI and spectral reflectance were discordant with those of the RedEdge
MX-dual on the second flight on September 13. Blue, green, and RENDVI were larger than these by RedEdge MX-dual. Red, red
edge, and near infrared were smaller than these by RedEdge MX-dual. These results show that the UAV multispectral remote sensing
technology is instructive for precision agriculture; however, calibration and evaluation are necessary to obtain accurate observation
results. ...