Spectralon panel

Five flights at an altitude of 60 m high were carried out during two seasons (three flights in 2014 and two in 2015) with the aim of extending the variability of field and plant condition. Meteorological conditions and phenological stages for each day of data collection were recorded (Table 2). All flights and image acquisition were concurrently done with Ψ_stem field measurements.
Flights in both seasons were carried out between 12:30 and 13:00, to reduce the ‘shadow-effect’ on the images [64 (link)]. Multispectral images were obtained from a MCA-6 camera (Tetracam’s miniature camera array), recording wavelengths at 530, 550, 570, 670, 700 and 800 nm. The image reflectance was normalized using a ‘white reference’ Spectralon panel (Labsphere Inc., Sutton, NH, USA) and compared with a spectroradiometer (SVC HR-1024, Spectra Vista Cooperation, Poughkeepsie, NY, USA) to account for any relative spectral response of each band of the camera as proposed by Laliberte, et al. [65 (link)]. All image processing was carried out using Matlab (MATLAB 2013a, The MathWorks, INC., Natick, MA, USA). The MCA-6 sensor was mounted on an octocopter, Mikrokopter Okto XL, equipped with the FlightNav 2.1 flight and navigation controller, MK3638 motors and 12”× 3.8’’ propellers. The sensor was affixed to a servo-controlled gimbal for stability and to ensure that it pointed directly down during flight.

Remotely sensed spectral information was collected in situ (between 0930–1100 hrs local time) before coral sampling using a GER-1500 portable field spectroradiometer enclosed within a custom made underwater housing (SpectraVista Corp.). The GER-1500 has a spectral range of 278–1094 nm with 512 spectral bands and a Full Width at Half Maximum (FWHM) equal to 2.8 nm). The housing was held by a diver at a 45° angle and a distance of 2.5 cm away from the target to ensure minimal or no signal contamination due to the presence of dissolved or particulate matter in the water column. Five replicate radiance measurements were obtained from each coral colony and used to study possible differences in reflectance within each sample. Additionally, radiance measurements were obtained from a 50% opaque diffuse barium sulfate (BaSO₄) reference Spectralon^® panel (Labsphere Inc.) immediately after sample spectral data collection at the same distance to correct for any atmospheric or wave lensing effects. Coral-leaving radiance was converted to remote sensing reflectance (Rrs) using: Rrs = Lc/Eg, where L_c is the coral-leaving radiance, and Eg = πL_p/Gc with Lp is the diffuse surface panel-leaving radiance and Gc is the calibration factor of the diffuse panel. The spectra were smoothed with a low pass Savitzky-Golay filter [17 –18 ] to eliminate spectral variability at scales shorter than 4 nm.

Leaf reflectance in the visible to shortwave infrared range (350–2500 nm, with a spectral resolution of 3 nm for wavelengths under 1000 nm, and < 8 nm up to 2500) was measured using a portable full range spectroradiometer (SR-3500, Spectral Evolution, Lawrence, USA), following the protocol described in Tóth et al. [60 (link)]. After 20-min dark adaptation, leaves were laid on a flat neoprene plate (reflectance factor < 5%) to minimize background reflection of light transmitted through the leaves. Leaf reflected radiance was measured at contact using a probe equipped with a 5-W internal light source under near-steady state conditions, i.e. 60 s after removing the leaf clip. Each spectrum is the result of 10 averaged scans, and automatic integration time optimization was used, with a maximum allowed of 50 ms per scan. Leaf spectra were eventually calibrated to reflectance using reflected radiance from a Spectralon panel (Labsphere, North Sutton, USA; reflectance factor > 99% for wavelengths under 1500 nm, and > 95% up to 2500 nm) as reference.