Biochar-amended soils enhance drought resilience in lettuce: Integrating Hyperspectral Imaging (HSI) and CNN-based moisture prediction

The increasing frequency and severity of drought events threaten global food security by degrading soil fertility and reducing crop yields. In this study, we employed biochar as a soil amendment to improve soil water-holding capacity and support plant growth under limited irrigation. While biochar offers a practical means to enhance crop performance, its evaluation has traditionally relied on destructive, labor-intensive measurements. Although we confirmed the positive effects of biochar through experimental assessments, we recognized the need for a rapid, non-destructive method to evaluate drought resilience under reduced irrigation more efficiently. To address this demand, we integrated biochar amendments with hyperspectral imaging (HSI) and convolutional neural network (CNN) modeling into a unified agronomic-computational framework. Corn-stalk-derived biochars, produced at 350 o C, 550 o C, and 700 o C and applied at 1%, 3%, 5 wt%, by weight, were tested in lettuce under well-watered and water-stressed conditions. Under drought stress, highly pyrolyzed biochars (CSB550 and CSB700) at 5 wt% increased fresh biomass by 33.2%, plant height by 29.8%, and leaf moisture content by 5.63% relative to the drought-stressed control without biochar amendment, while improving soil moisture retention by up to 24%. Leveraging HSI’s rich spectral signatures and CNN’s ability to model complex, non-linear relationships, plant moisture content was predicted with an R 2 of 0.93 and RMSE of 0.28%, closely matching destructive measurements. This integrated approach delivers both a practical soil management strategy to mitigate drought impacts and a scalable, high-throughput tool for real-time crop water monitoring, supporting precision irrigation and climate-resilient agriculture.