Spectral range within global aCDOM(440) algorithms for oceanic, coastal, and inland waters with application to airborne measurements

One- and Two-Band Sensors and Algorithms to Derive aCDOM(440) from Global Above- and In-Water Optical Observations

“…Data collection spanned 29 April 2013 to 13 September 2017. The inclusion of near-conservative samples expands the [13] dataset from 613 exclusively conservative samples, as used in Houskeeper et al [9], to 789 samples. The expansion allows a cross-validation analysis, wherein 80% of the total dataset is randomly selected for algorithm fitting and compared to the remaining 20%.…”

“…More importantly, Hooker et al [13] showed that the two-channel (i.e., band-ratio) EMA approach spans a global distribution in a CDOM (440) derived from in-water observations of the diffuse attenuation coefficient, K d (λ). Houskeeper et al [9] extended that work by demonstrating a global relationship between a CDOM (440) and the normalized water-leaving radiance, L W (λ) N , which can be derived from above-and in-water optical observations (discussed below). Some of the above-and in-water optical observations used in Houskeeper et al [9] and Hooker et al [13] were obtained from semi-autonomous platforms manufactured by Biospherical Instruments Inc. (San Diego, CA, USA), respectively: (a) the Compact-Airborne Environmental Radiometers for Oceanography (C -AERO) above-water instrument suite, which operates autonomously on a manned aircraft [14]; and (b) the Compact-Optical Profiling System (C -OPS) in-water instrument suite with Compact-Propulsion Option for Profiling Systems (C -PrOPS) accessory [14], which has two small digitally controlled thrusters to allow profiling independent of the primary platform, e.g., a research vessel or shoreline, or from a USV [15].…”

“…As a log-normal variable, unsampled waters wherein a CDOM (440) exceeds 2.146 m −1 are unlikely to significantly expand the range in a CDOM (440) when considered in log space. Furthermore, algorithms derived using a narrower a CDOM (440) range in Houskeeper [8] produced similar algorithmic coefficients as those derived using an expanded a CDOM (440) range in Houskeeper et al [9], due to the logspace linearity of the relationship between a CDOM (440) and ratios of optical observations at UV and NIR wavebands. A perspective that accounts for observations uniformly spread across oceanic, coastal, and inland waters, while spanning more than three decades of the log-normal variable, is considered herein as global.…”

“…Demonstrated life cycles for instruments built with microradiometers presently exceed 15 years with a failure rate well below 1%. The 10 decades of dynamic range and 15-30 Hz high-speed sampling allow above-and in-water light contamination effects to be properly dealt with, as follows (respectively): surface light perturbations due to oblique wave facets that reflect sunlight can be discretized and properly removed [9], and the central tendency of non-Gaussian brightening from wave focusing can be properly determined with minimal bias [13].…”

“…The other statistics are the root mean square difference (RMSD), the mean absolute difference (MAD) or mean absolute error (MAE), depending on the literature, plus the mean bias (MBIAS). The RMSD follows a standard statistical formulation [9], and both the MAD and MBIAS were derived following Seegers et al [37], as follows:…”