Detection of water stress in an olive orchard with thermal remote sensing imagery

Abstract

An investigation of the detection of water stress in non-homogeneous crop canopies such as orchards using high-spatial resolution remote sensing thermal imagery is presented. An airborne campaign was conducted with the Airborne Hyperspectral Scanner (AHS) acquiring imagery in 38 spectral bands in the 0.43–12.5 μm spectral range at 2.5 m spatial resolution. The AHS sensor was flown at 7:30, 9:30 and 12:30 GMT in 25 July 2004 over an olive orchard with three different water-deficit irrigation treatments to study the spatial and diurnal variability of temperature as a function of water stress. A total of 10 AHS bands located within the thermal-infrared region were assessed for the retrieval of the land surface temperature using the split-window algorithm, separating pure crowns from shadows and sunlit soil pixels using the reflectance bands. Ground truth validation was conducted with infrared thermal sensors placed on top of the trees for continuous thermal data acquisition. Crown temperature ( T c), crown minus air temperature ( T c − T a), and relative temperature difference to well-irrigated trees ( T c − T R, where T R is the mean temperature of the well-irrigated trees) were calculated from the ground sensors and from the AHS imagery at the crown spatial resolution. Correlation coefficients for T c − T R between ground IRT sensors and airborne image-based AHS estimations were R 2 = 0.50 (7:30 GMT), R 2 = 0.45 (9:30 GMT) and R 2 = 0.57 (12:30 GMT). Relationships between leaf water potential and crown T c − T a measured with the airborne sensor obtained determination coefficients of R 2 = 0.62 (7:30 GMT), R 2 = 0.35 (9:30 GMT) and R 2 = 0.25 (12:30 GMT). Images of T c − T a and T c − T R for the entire field were obtained at the three times during the day of the overflight, showing the spatial and temporal distribution of the thermal variability as a function of the water deficit irrigation schemes.