Proxy detection of wheat water stress from photochemical reflectance index and land surface temperature data

Abstract

In semi-arid and arid regions, crops face elevated atmospheric demands and endure prolonged periods of moderate to severe water scarcity. In this context, this study investigated the effectiveness of the photochemical reflectance index (PRI) and a normalized surface temperature index (Tₙₒᵣₘ) for proxy detection of the water stress of winter wheat crops. Furthermore, the potential of PRI for characterizing water, atmospheric or photo-inhibition stress, and wheat transpiration was assessed over experimental drip-irrigated crop fields in the Haouz plain, central Morocco. In practice, PRI observations were compared to agro-environmental variables such as Leaf Area Index (LAI), Available Water Content (AWC) at a root zone depth, net Radiation (Rₙ), Vapor Pressure Deficit (VPD) and the wheat transpiration derived from sap flows, lysimeters and a crop water balance model. Due to the strong relationship between PRI and LAI (R² = 0.91), another index named PRIⱼ was derived to correct for this effect. The PRIⱼ was found to be independent of structural effects related to LAI and significantly correlated with AWC (R² = 0.85). Using the PRIⱼ index, we can reflect the level of water stress experienced by the wheat field throughout the experiment with an R² of 0.69 for a FAO-56 water stress coefficient (Kₛ) of less than 1. Under dry conditions, for an AWC below 30%, the correlation between AWC and Tₙₒᵣₘ gives an R² of 0.29. However, comparison of PRIⱼ with the Tₙₒᵣₘ index showed that PRIⱼ is an early water stress index and provides information on the state of the vegetation cover at all stages of wheat development. The study's findings can have a significant impact on the use of the PRI as a water stress indicator, helping in the optimal irrigation of crops.

Code and data availability

The supplied blocks contain only figure captions and the reference list of the article. No public phenotype/trait datasets, plant images or sensor inputs, author analysis code, trained models, or supplements with paper-specific data or code are mentioned, and no availability statements or repository URLs appear.

No evidence-backed public reproduction asset is currently recorded.