How Sensitive Is Thermal Image-Based Orchard Water Status Estimation to Canopy Extraction Quality?

Accurate canopy extraction and temperature calculations are crucial to minimizing inaccuracies in thermal image-based estimation of orchard water status. Currently, no quantitative comparison methods exists the context precision irrigation. The accuracies four were compared, effect on status was explored for these methods: 2-pixel erosion (2PE) where non-canopy pixels removed by thresholding morphological erosion; edge detection (ED) edges identified morphologically dilated; vegetation segmentation (VS) using histogram analysis spatial watershed segmentation; RGB binary masking (RGB-BM) a layer statistically extracted from an image masking. field experiments occurred four-hectare commercial peach during primary fruit growth stage (III). relationship between stem potential (SWP) crop stress index (CWSI) established 2018. During 2019, large dataset ten infrared two images acquired. had different accuracies: 12 August, overall accuracy 83% 2PE method, 77% ED 84% VS 90% RGB-BM method. Despite high between-row weeds misidentified as canopy. Canopy CWSI calculated average 100% (CWSI_T100%) coolest 33% (CWSI_T33%). CWSI_T33% produced similar SWP–CWSI models irrespective method used, while CWSI_T100% yielded inferior models. results highlighted following: (1) contribution is not significant extraction. can be with reliability solely images. (2) T33% approach calculation more robust superior simple mean all pixels. These noteworthy findings step forward implementing imagery irrigation management.