**1. Introduction**

Kelp forests are highly productive three-dimensional coastal marine habitats [1,2] that provide a number of environmental services and contribute substantial economic value to coastal communities globally [3]. In the northeast Pacific, the two dominant surfacecanopy forming kelp species, *Nereocystis luetkeana* and *Macrocystis pyrifera* [4], stabilize shorelines via wave dampening [5,6], support economically important fisheries [7,8], and are commercially harvested for various purposes [9,10]. However, both kelp species are subject to high spatial and temporal variability, correlated with biotic and abiotic drivers of change [11,12]. As such, resource managers are incentivized to monitor the status of these kelp forests, and the corollary effects of the ecosystem services they provide [10,12,13], a task that has been facilitated by remote sensing since the mid-20th century [12,14].

Generally, the remote sensing of surface-canopy forming kelp forests aims to detect the portion of the kelp that forms a canopy, floating at the water's surface; using sensors

**Citation:** Timmer, B.; Reshitnyk, L.Y.; Hessing-Lewis, M.; Juanes, F.; Costa, M. Comparing the Use of Red-Edge and Near-Infrared Wavelength Ranges for Detecting Submerged Kelp Canopy. *Remote Sens.* **2022**, *14*, 2241. https://doi.org/10.3390/ rs14092241

Academic Editors: Simona Niculescu, Junshi Xia and Dar Roberts

Received: 11 March 2022 Accepted: 5 May 2022 Published: 7 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

aboard Earth Observation Satellites (EOS) [15,16], piloted aircraft [10,12], and Uncrewed Aerial Vehicles (UAVs) [17]. In order to use the data provided by remote sensing platforms effectively, it is crucial to understand factors that influence the spectral signature of kelp canopy in water [16]. Floating kelp canopy has high reflectance in the near-infrared wavelength range (NIR) (700–1000 nm), which contrasts with the high NIR absorption by the surrounding water, allowing for binary classification of floating kelp canopy and water within an image [14]. However, there are numerous considerations (e.g., sun glint, bathymetry, turbidity; see [13,16]) that can reduce the separability between the spectral values of kelp canopy and water. One crucial factor that can affect the ability to detect kelp canopy is the submersion of the canopy by tides and associated tidal currents, which can dampen the NIR reflectance of kelp and lead to potential errors when estimating kelp area or biomass [17,18].

In an attempt to minimize classification errors associated with kelp submergence, remote sensing imagery is often acquired at low tides during the peak growing season (mid-late summer) when the majority of the kelp canopy is floating at the water's surface [10,18,19]. However, there are multiple reasons why a remote sensor may also want to detect the submerged portion of the kelp canopy. For example, the northeast Pacific coastline often experiences non-ideal weather conditions for remote sensing data acquisition, leading to imagery being opportunistically collected at higher than ideal tidal heights when more kelp canopy is more likely to be submerged compared to ideal low tide conditions [13,16]. Further, the fixed rate of EOS orbits may result in some regions only having imagery available during high tides even if acquisition conditions are otherwise ideal [13]. Even if remote sensing imagery is captured during ideal tide and weather conditions, portions of kelp canopy may also be continuously submerged depending on the species being targeted. Specifically, if a remote sensor is targeting detection of *Nereocystis luetkeana* (hereafter, *Nereocystis*) surface canopy, one has to consider the two distinct structures with varying buoyancy, the bulb and blades. The bulb is a roughly cylindrical gas-filled structure that floats on the surface of the water and is anchored to the sea floor by a stipe and holdfast [9]. The blades are long thin structures that trail from the end of the bulb, often with many individuals around four meters long per bulb [9]. The blades are not buoyant and are likely to remain submerged below the water's surface regardless of tidal height [16,20]. In addition, floating portions of kelp canopy may be periodically submerged in areas with especially strong currents [20]. Therefore, it is important to understand how submersion of kelp canopy affects the reflectance in the NIR range, as well as whether certain spectral features may allow for higher detectability of kelp when collecting remote sensing imagery from different platforms.

In the past, the red-edge (RE) spectral region (670–750 nm), which includes a range of the shortest NIR wavelengths, has traditionally been used to determine health characteristics of terrestrial plants [21]. However, these wavelength ranges also penetrate deeper into the water column than longer NIR wavelength ranges [22], resulting in the potential for higher above water reflectance in the RE than the longer NIR for submerged vegetation [23–26]. Therefore, given the spectral similarities between kelps and other types of vegetation, it is reasonable to assume that the RE wavelength range may also be beneficial for detecting submerged kelp canopy. Hereafter, the term NIR will refer to only the longer wavelength range above 751 nm, to avoid confusion with the NIR wavelength range that overlaps the RE wavelength range.

To date, there have been no direct comparisons of the ability to detect submerged kelp when using RE or NIR wavelength ranges. Additionally, while the submersion kelp canopy due to tides and currents is well documented using various sensors with different spatial and spectral resolutions [11,17–20,27], there has been no characterization of the changes to the above water spectra of kelp as the canopy is submerged, nor any investigation of the band combinations used in vegetation indices in relationship to accurate detection of the submerged kelp canopy. With this is mind, our goal was to characterize changes to abovewater reflectance of different *Nereocystis* canopy structures as they were submerged and to

relate those changes to depth detection limits. To accomplish this goal, we performed (1) an experiment that documents the effects of kelp submersion on the above-water hyperspectral reflectance of both *Nereocystis* bulb and blade structures. We also compared (2) the detection limits of submerged kelp using RE and NIR vegetation indices, which were calculated from the simulated multispectral bands of high spatial-resolution air- and space-borne sensors.

#### **2. Materials and Methods**
