Bioenergy crops monitoring using Sentinel-2 and UAV-derived imagery

Abstract

Purpose. Establishing the regularities for the monitoring of the bioenergy crop conditions using Sentinel-2 and UAV-derived imagery. Methods. A field experiment was carried out in an experimental field of the Institute of Bioenergy Crops and Sugar Beet National Academy of Agrarian Sciences of Ukraine (50.023194, 30.173895), located in a zone of unstable soil moisture in the Right Bank Forest Steppe, in 2022−2023. Results. Among the studied traditional crops, sugar sorghum, sugar beet, and fodder beet are interesting crops from the standpoint of biomass for energy, as they can produce 110 t/ha, 120 t/ha, and 135 t/ha of biomass, respectively, or 20 t/ha, 26 t/ha, and 24 t/ha of dry matter, respectively. The yield of energy crops can be predicted with an acceptable level of accuracy using many known vegetation indices. However, the relationships between vegetation indices and crop yield are not consistent at every stage of plant growth and development, leading to low accuracy in yield estimation. The combination of vegetation indices related to the structural characteristics of the canopy and chlorophyll content in the aboveground biomass can improve the accuracy of yield estimation. Moreover, combining vegetation indices related to canopy structure, chlorophyll content, and stress indices as input to yield prediction models may provide even higher correlations for yield prediction. Conclusions. Usually, plantations of miscanthus and switchgrass are fertilised annually, while willow and poplar are fertilised every 3–4 years, after each harvesting. Therefore, the nutrient provision of tree species may not be sufficient. It was found that the content of total nitrogen in the soil of bioenergy plantations highly correlates with NDVI. To use NDVI for bioenergy crop prediction, an algorithm for determining the level of plant nutrients should be developed. It was found that at the end of the growing season (September or early October), the use of vegetation indices allows for the accurate estimation of the size and condition of bioenergy plantations since most crops in this period are already harvested. Sentinel-2-derived imagery is useful for monitoring bioenergy crop plantations as it provides images with a resolution of 10 m at 3–5-day revisiting time. We investigated NDVI on a total area of Miscanthus × gigantheus plantations of 2.9 ha. The satellite-derived data accumulated and aggregated by the OneSoil application as of August 20 and September 19 resulted in NDVI values of 0.80 and 0.70, respectively, while the NDVI obtained with UAV imagery was 0.82 and 0.77, respectively. Consequently, the satellite can provide quite acceptable NDVI data for use in the monitoring of bioenergy plantation yield at the national level.