Correlation dependences between crop reflection indices, grain yield and optical characteristics of wheat leaves at different nitrogen level and seeding density

Improvement of crop remote sensing application in precision agriculture systems and development of algorithms for satellite and aerial imagery interpretation necessitate comparing remote sensing and ground-based survey data. This paper is the first to report data on spectral characteristics of the leaf diffuse reflection in spring wheat, their relationship with plant productivity, colorimetric characteristics and reflection indices of the crop vegetation cover, depending on the crop management technologies. The first research objective was to assess the dependence of crop canopy optical characteristics and productivity on seeding density and the rates of pre-treatment with nitrogen fertilizer. The second objective was to reveal correlations between the canopy remote sensing data and the leaf diffuse reflection parameters registered by a contact sensor (on the example of spring wheat Triticum aestivum L. cv. Daria). The plants were grown on the test plots (Menkovo experimental station of Agrophysical Research Institute, Leningrad Province, Gatchina District) in 2020-2021 years. In total, six test plots with an area of 100 m2 were assigned. Nitrogen rates ranged from 0 (no fertilizers applied) to 200 kg/ha with increments of 40 kg/ha, and seeding rates of 500 and 600 seeds per m2. The diffuse light reflection of leaves was registered in situ on stages BBCH 30-31 “booting” and BBCH 53-55 “earing” by a fiber-optical spectroradiometric system (Ocean Insight, USA) in the range from 350 to 1000 nm with a step size of 0.3 nm. After the reflectance spectra recording, the plants were dried to constant weight and each plant was weighed to assess correlations with the optical parameters of leaves. The light diffusion index R800 was determined from the spectra of reflected radiation. The reflectance indices calculated were the following: ChlRI (chlorophyll index), PRI (photochemical index), FRI (flavonoids index), WRI (water content index), ARI (anthocyanins content index) and FRI (flavonoids content index). These indices estimate the intensity of the photosynthetic apparatus function and the efficiency of light use in photosynthesis. The crop canopy remote sensing was performed at BBCH 25-27 (“tillering”), BBCH 30-31 (“booting”), BBCH 53-55 (“earing”), and BBCH 61-65 (“blossoming”) stages using two synchronized digital cameras Canon G7X (Canon Inc., Japan) mounted on a Geoscan 401 quadcopter (Geoscan, Russia). From a height of 75-120 m, the digital images were obtain in the visible and near infrared spectral ranges. The vegetation indices NDVI (Normalized Difference Vegetation Index) and ARVI (Atmospherically Resistant Vegetation Index) were calculated based on the optical characteristics. For quantitative interpretation of the colorimetric characteristics of leaves and the crop canopy, we used a three-dimensional model of the CIELAB color space. Plants were weighted during the growing season, and, after harvesting, the grain productivity was estimated for the plants sampled from 0.25 m2 reference plots. The obtained results indicate that at the early stages of plant development when the vegetation cover remains open, NDVI characterizes the degree of nitrogen supply rather exactly and identifies areas with underdeveloped plants. However, with the development of plants and the formation of a closed vegetation cover this index fails to provide reliable results. ADVI also fails to provide reliable information about the state of spring wheat plants and identifies areas that require additional fertilize application. A close linear correlation between the rate of applied nitrogen fertilizer, the net productivity of plants and spectral characteristics of leaves measured in situ occures until the late stages of development (BBCH 53-55 “earing”, BBCH 61-65 “blossoming”). Crop monitoring based on colorimetric characteristics made it possible to detect changes in the crop canopy associated not only with the plant development and crop density, but also with the spectral characteristics of the diffuse reflection of plant leaves. A comparison of the remote and contact sensing data allows us to conclude that the ChlRI, PRI, FRI and WRI indices can successfully identify areas in which nitrogen deficiency has developed during the closed canopy formation, when the commonly used indices, for example, NDVI, fail to be reliable.