Geometric and radiometric analysis of dual wavelength NIR-LiDAR scanning for estimating physiological and free water of leaves in apple tree canopies

Current challenges in the production of fruit and vegetables are increasing with climate change, impacting water requirements in different production operations. Although water status of plants and surface wetness detection have been widely investigated using diverse sensors and well established methodologies, spatial resolution and the request to monitor potentially all plants are continuous trade-off in large orchard systems. Thereof, light detection and ranging systems (LiDAR) are gaining rapid interest in horticultural research due to its ability to provide geometric and radiometric data of plants in field conditions by means of three-dimensional point clouds. Previous experiments have been approaching close-range sensing of leaf water content ( LWC ) based on the absorption of electromagnetic wave by water in the near infrared wavelength range. This paper is aimed at assessing the water status of apple canopies using a dual-wavelength LiDAR scanner at 1320 and 1450 nm. Physiological water was measured by LWC , leaf water potential, and crop water stress index. Additionally, the sensor responses to spray application of water on the canopies was investigated. A semi-automated processing algorithm was developed to isolate points of leaves from wood structures. Normalized returned signal strength intensity ( I N o r m ) was obtained at 1320 and 1450 nm after radiometric calibration. A normalized difference NIR index ( N D N I R ) is proposed, based on the two wavelengths. Only marginal differences were found between I N o r m values, comparing fully irrigated and drought stressed canopies for all LiDAR derived variables ( I 1320 , I 1450 and N D N I R ). Only LWC was slightly correlated to I 1450 , with adjusted R 2 of 0.3. On the other hand, significant differences were found between dry and sprayed canopies, with free water on the leaves, comparing patterns at I 1450 and N D N I R at 95 % confidence interval using Kolmogorov-Smirnov statistical test. Benefits and challenges of the methodology point to feasible applications in detecting free water on the leaf surface under field conditions.