UAV-based CWSI estimation for tomato water stress monitoring

Thermal infrared (TIR) imaging has been widely applied as a prominent remote sensing technique to assess plant water conditions. Recent advances in TIR imaging, primarily the use of high-resolution TIR cameras mounted on unmanned aerial vehicles (UAVs) and their integration with water-sensitive indicators, have further enhanced its applicability for crop water stress monitoring. This study compares TIR-based CWSI estimation approaches to identify the most reliable method for precise, real-time water stress monitoring. Assessments were performed for three crop seasons (2021–2023) on processing tomato. Midday UAV surveys and ground truth leaf water potential (LWP) measurements were performed across different growth stages and irrigation levels. Canopy temperature (T c ) retrieved from TIR images was used in CWSI calculations using four different approaches: empirical (CWSI e ), modified empirical (CWSI e3 ) proposed in this study, air-referenced model (CWSI a ), and statistical (CWSI s ). Additionally, the baseline canopy-air temperature differential (T c –T a ) was evaluated as a non-normalized indicator. Results revealed strong LWP-CWSI associations across all physiological stages and irrigation treatments. Phenological variation revealed index-specific sensitivity, with optimal correlations at different stages. When combining the evaluated conditions, CWSI e3 exhibited the highest correlations with LWP (R 2 = 0.63), followed by CWSI a (R 2 = 0.62). In comparison with the normalized CWSI, the referential T c –T a method maintained an important relationship with LWP (R 2 = 0.52), confirming its value as a simple, yet reliable indicator of crop water status. These findings proposed CWSI e3 as a robust estimate of crop water stress and highlighted the growth stage-dependent performance of the CWSI estimations, providing practical guidance for selecting and applying appropriate methods in precision crop water management.