Non-destructive method using UAVs for high-throughput water productivity assessment for winter wheat cultivars

Abstract

Grain yield or biomass production per unit water consumption is defined as crop water productivity (WP). Using cultivars with high WP is important for reducing the negative influences of water shortages on agricultural production. Common methods for obtaining the WP of different cultivars are time-consuming and required considerable labor input. Developing nondestructive and high-throughput methods is essential for phenotyping cultivars with high WP. Unmanned aerial vehicles (UAVs) capture high spatiotemporal resolution remote sensing data, offering an opportunity to accurately estimate evapotranspiration (ET) and biomass during crop growing seasons to assess WP. In this study, the WP at the main growing stages of 10 winter wheat cultivars was assessed under three irrigation levels based on UAV-derived ET and biomass. Continuous daily ET was estimated by a new method combining the SEBAL (Surface Energy Balance Algorithm for Land) model, crop coefficient (K_c) and soil water balance equation. Biomass was estimated from multispectral data, and five machine learning algorithms were compared, with random forest selected as the best performer. Using the ET and biomass estimates from the UAV flights, the WP for different growing periods of various winter wheat cultivars was obtained. The WP at the biomass level around the flowering stage was significantly correlated with the WP at the grain yield level for all the cultivars under the three irrigation conditions. Therefore, the WP monitored using UAVs during this period was used to assess the final WP of different cultivars, as biomass accumulation during this stage was critical for final grain production, and the daily ET also peaked at this time. The results from this study showed that UAVs based on non-destructive and high-throughput methods was feasible for assessing the WP of multiple cultivars to save labor and time.