Premium
Simple in‐field evaluation of moisture content in curing forage using normalized differece vegetation index (NDVI)
Author(s) -
Lim Jihyun,
Watanabe Nariyasu,
Yoshitoshi Rena,
Kawamura Kensuke
Publication year - 2020
Publication title -
grassland science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.388
H-Index - 19
eISSN - 1744-697X
pISSN - 1744-6961
DOI - 10.1111/grs.12275
Subject(s) - spectroradiometer , multispectral image , normalized difference vegetation index , remote sensing , mean squared error , environmental science , multispectral pattern recognition , water content , reflectivity , mathematics , statistics , leaf area index , geography , geology , optics , ecology , physics , geotechnical engineering , biology
Production of hay at a proper moisture level is critical to reduce the spontaneous heating and nutrient loss during the preservation phase. Herein, we explored a new method for estimating the moisture content (MC) of field curing forages that uses a portable spectroradiometer (400–900 nm) and a hand‐held normalized difference vegetation index (NDVI) meter. The spatial distributions were assessed by a multispectral camera with unmanned aerial vehicle (UAV) platform. Field spectral measurements were conducted daily at 10 random plots over six trials at four forage fields across the entire curing period. Multispectral imagery acquisition by UAV flights was conducted at one trial. The MC (wet basis) was determined using a forced‐air oven. For 400–900 nm spectral reflectance measured by the spectroradiometer, we conducted linear regression analyses between the MC and all available combinations of a normalized difference spectral index (NDSI (Band1, Band2) = ( ρ Band1 − ρ Band2 )/( ρ Band1 + ρ Band2 ), where ρ λ is the apparent reflectance and λ is the wavelength in nm). The combinations that are generally used as greenness indices showed high coefficients of determination, for example, red and near infrared ( R 2 = .76, RMSE = 10.66%) and green and red ( R 2 = .81, RMSE = 9.57%). The NDVI meter and multispectral imagery showed the feasibility of NDVI ( R 2 = .79, RMSE = 8.58% and R 2 = .89, RMSE = 6.74%, respectively) as a parameter to estimate the MC. We were able to verify the spatial variability of the MC in the field based on the NDVI imagery, which indicates that our method provides information for site‐specific management (e.g., partial swath manipulation) and for decision‐making regarding the harvest time and location.