Detection ranges of blue whale vocalizations from a glider-based hydrophone Open Access

Ocean gliders have been used to study cetacean distribution and habitat, as they can be equipped with hydrophones to monitor cetacean vocalizations^1–3 as well as a wide variety of oceanographic sensors.^4–6 Gliders are increasingly used to relate the acoustic presence of cetaceans to oceanographic conditions.^1,3,4,6–10 Using gliders for passive acoustic monitoring (PAM) offers the advantage of utilizing in situ temperature and salinity data to determine contemporaneous sound speed profiles (SSP), thereby estimating the detection range (DR) of the hydrophone. The DR, the maximum distance a hydrophone can detect acoustic signals from vocalizing cetaceans,⁹ is crucial for interpreting results from paired acoustic and oceanographic glider surveys that examine cetacean acoustic presence. Variability in local bathymetry and oceanographic conditions significantly impact these detection distances. This highlights the need for DR estimation to understand the spatial scale of acoustic monitoring to study the drivers of whale distribution. Additionally, DR is necessary for cetacean density estimates¹¹ since it enables the conversion of calls per unit time to calls per unit area.^9,12–14

DR depends on site-specific variables such as water column temperature,¹⁵ bathymetry,¹⁶ and ambient noise levels (dB re:1μPa).¹⁷ Thus, each study requires site-specific DR estimations. For instance, a bottom-mounted hydrophone in southern New Zealand estimated the DR for blue whale D-calls between 5.5 and 34.8 km,¹⁸ whereas in the Corcovado Gulf (Northern Chilean Patagonia), it ranged from 1.4 to 6 km.¹⁹ When using gliders, added complexity occurs because the glider moves horizontally through a study area and vertically through the water column.

In this study, we asked two research questions: (1) How do the different oceanographic/bathymetric conditions affect the DR of southeast Pacific blue whale vocalizations (SEP2²⁰ song and D-call^21,22) as the glider moves horizontally through the study area? (2) How does the vertical position of the glider in the water column (which we will call receiver depth) affect the DR of blue whale vocalizations? In a study that investigated blue and sei whale distributions using acoustic/oceanographic glider surveys in Northern Chilean Patagonia without examining DR,⁴ gliders surveyed a wide range of temperature and salinity conditions between the estuarine and oceanic poles of the Patagonian Inner Sea; this study utilizes the same dataset.

The study area was Northern Chilean Patagonia, specifically the area between the Inner Sea of Chiloé, the Corcovado Gulf and Guafo Island, between 44.2°S, 75.6°W and 42.6°S, 72.8°W [Fig. 1(a)]. Based on a visual analysis of the average salinity values below mixed layer depth from three glider deployments, the study area was divided into four distinct zones: ES, Estuarine zone; ET, Estuarine Transition zone; OT, Oceanic Transition zone; OC, Oceanic zone [Fig. 1(b)].

Between 4 April and 6 June, 2019, three glider deployments were conducted, covering a total of 2241.3 km and recording oceanographic data and 1729 h of acoustic data [Fig. 1(a)]. Two Slocum gliders (model G1 Teledyne Webb Research, Falmouth, MA) were utilized simultaneously, each equipped with a conductivity, temperature, and depth sensor, along with a digital acoustic monitoring (DMON) instrument that recorded continuous acoustic data at a 2 kHz sampling rate (see Buchan et al.⁴ for detailed methods).

In the present study, no signal processing was applied, as no detections of songs or vocalizations were presented. This type of analysis was not part of the main objective of this work.

Here, FOM was the figure of merit, which was the maximum allowable TL of an acoustic signal before it was no longer detectable by the hydrophone; SL was the source level of the signal (dB at 1 m re:1μPa); DT was the detection threshold (dB); and N was the noise level (dB re:1μPa). SL values were obtained from the literature, and N was calculated from the recorded acoustic data (Table 1). To simplify the data analysis, we established a DT value of 10 dB as a theoretical threshold; any vocalization with a DT of 10 dB or higher was considered detected, while those below this threshold were not. This DT value was assumed and not calculated from a specific frequency band, and it was used to calculate the FOM, which allowed us to derive the DR.

TL was modeled across the water column along a horizontal transect [Fig. 1(b)]. The TL, the dB loss per distance from the whale to the glider, was computed using the Range Dependent Acoustic Model²⁴ in matlab 2021a.²⁵ Input parameters for the model are found in Table 1.

To compute TL, bathymetric data were sourced from the National Centers for Environmental Information (NCEI).³¹ Within each zone, a single north-south bathymetric transect of approximately 30 km was chosen [see Fig. 1(b) for locations]. SSPs were derived using the Seawater toolbox³² in matlab, utilizing salinity, temperature, and pressure data from the glider. Along each bathymetric transect, TL was computed at intervals of 10 m horizontally and 1 m vertically.

The DR was defined as the maximum distance at which the smoothed TL was less than FOM. However, as part of the DR calculation, it was necessary to first determine the FOM and the distance at which it occurred. The FOM was calculated using the passive sonar equation, and the distance where TL equaled or exceeded the FOM was identified. To calculate DR, we then smoothed each TL and determined the maximum distance where the smoothed TL remained below the FOM. This approach allowed us to identify the distance at which the signal would no longer be detectable due to increased TL.

Mean N for the 54 Hz frequency band of D-call vocalization and the 23.6 Hz frequency band of SEP2 were determined from the glider data using PAMGuide³³ (power spectral density analysis, 1 s window, 50% overlap, Hanning window, end-to-end calibration, −169.8 dB system sensitivity). Noise levels were not determined for every depth and every zone, given the focus of this analysis on the effects of SSP, bathymetry, and receiver depth on DR.

To explore how different oceanographic conditions impacted the DR of blue whale vocalizations for the DMON (first question), we calculated the TL and the distance at which TL ≥ FOM occurs for each zone and vocalization (Fig. 3). TL in each zone was modeled using the bathymetric transect and the average SSP (Fig. 2). Subsequently, FOM was calculated for each vocalization type and zone, assuming whale calls originated from a fixed depth of 20 m and the glider was at a fixed depth of 30 m.

To explore how the DR changed for each call type with varying receiver depths (second research question), DR was calculated at every 1-meter depth for each call type and zone. Additionally, to assess the relative impact of bathymetry vs sound speed conditions on DR variations with glider depth, DR was computed using both variable bathymetry and SSP for each zone. Then, only SSP was varied while keeping bathymetry constant across all zones (using only data for the OC zone).

The FOMs calculated from the passive sonar equation were 74.4 dB for SEP2 song calls and 63.7 dB for D-calls. The N for SEP2 was 89.6 dB re:1μPa and for D-calls was 82.3 dB re:1μPa.

The distance where TL ≥ FOM of SEP2 (up to 20.1 km) was greater than D-calls (up to 9.2 km) in all zones (Fig. 3). This distance for D-calls was similar in all zones (∼6–9 km). For SEP2, the distances where TL ≥ FOM were similar in the ES, ET, and OC zones (∼11–15 km), but higher in the OT zone (∼20 km).

We observed significant DR variation with receiver depth (second research question) for both call types (Fig. 4). In OT, SEP2 DR nearly tripled between the surface and 60 m depth [Fig. 4(b)]. However, at greater depths, DR for both calls sharply decreased due to high TL, likely caused by proximity to the bottom. SEP2 exhibited higher DR than D-calls due to greater SL (Fig. 4).

Bathymetry primarily influenced DR differences among oceanic zones: when using SSP for each zone and keeping bathymetry constant, minimal differences in DR between zones are evident. This is highlighted by comparing Figs. 4(a) and 4(b) (variable bathymetry and SSP) with Figs. 4(c) and 4(d) (variable SSP only).

For studies aiming to use PAM to understand cetacean spatial distribution, calculating DRs for emitted acoustic signals is important. This computation enables ecological analyses to be better tailored to relevant spatial scales.³⁴ A core challenge in PAM is distinguishing whether calls originate from animals within a given area.³⁵ This challenge intensifies with glider-based PAM, given gliders' horizontal and vertical water column movements. Here, we estimate blue whale call DRs from a glider-based recorder. This is crucial for robustly interpreting the glider-based acoustic whale detections to link whale acoustic presence with oceanographic conditions sensu Buchan et al.⁴ 

It was found that the distances where TL ≥ FOM and DRs of D-calls were consistently lower than those for SEP2 in every zone [Figs. 3, 4(a), and 4(b)]. This is probably due to the higher estimated SL of SEP2. Since the SL of SEP2 is 18 dB higher than that of D-calls (174 vs 156 dB, respectively), this call type propagates over greater distances despite higher TL and N values (Table 1). For D-calls, DRs in ES, ET, and OC were no greater than 9 km, and in OT were less than 14 km, regardless of glider depth [Fig. 4(a)]. These values are within the same order of magnitude as the 6 km range reported by Buchan et al.¹⁹ For SEP2, the 15.3 km reported by Buchan et al.¹⁹ is exceeded in every zone when the glider is between 40 and 85 m depth, i.e., midwater [Fig. 4(b)]. The higher DR values at medium depths in this study compared to the results of Buchan et al.¹⁹ are probably due to lower TL in the midwater and because that study was from a hydrophone deployed in a relatively enclosed basin (Fig. 3).

For both D-calls and SEP2, DRs indicate that calling whales were no further than 20 km from the glider. This supports the conclusion of Buchan et al.⁴ that the spatial variations in call presence observed were due to differences in whale acoustic presence and not differences in DR. However, the nearly two times greater DR of SEP2 compared with D-calls, at least in part, might explain the higher presence of SEP2 (43 566 calls) vs D-calls (5656 calls).⁴ 

The different oceanographic and bathymetry conditions that the glider moved through in the study area affected the DR of blue whale vocalizations. We found that DRs in OT were two (D-calls) to three (SEP2) times higher than DRs in the ES, ET, and OC [Figs. 4(a) and 4(b)]. We did not find a consistent (decreasing or increasing) trend between the estuarine pole and the oceanic pole of the study area. Thus, we cannot assume higher DRs occur in oceanic areas and smaller DRs in estuarine areas (or vice versa). This emphasizes the need to estimate site-specific DR.

In this relatively shallow study area, differences in DRs among oceanographic zones were primarily influenced by bathymetry rather than SSPs (Fig. 2). Given that the study area is influenced by several water masses of varying salinities (Buchan et al.⁴ and references therein), we anticipated that DR variations would be influenced by SSPs. However, as depicted in Figs. 4(c) and 4(d), DR remained consistent when bathymetry was constant and only SSP varied. Oliveira et al.³⁶ observed significant TL effects from bathymetric changes off Long Island.³⁶ Glider depth notably impacted DR, with the highest DR when the glider was at mid-depths vs at the surface and near the bottom, similar to Johnson et al.³⁷ This study reveals a two to fourfold DR increase at mid-water depths compared to surface and bottom depths particularly for SEP2 calls, which could be linked to higher TL values in those areas of the water column.

Understanding ocean sound propagation is complex. Precisely determining TL and calculating hydrophone DR is challenging due to environmental conditions, call characteristics, modeling methods, and instruments used.³⁸ DRs are vital for estimating whale population density from acoustic data. In this study, DRs varied up to threefold due to changes in bathymetry. Although SSPs appeared to have minimal influence on DR in this study, temperature, salinity, and stratification may have a greater impact on glider-based DRs in other study areas and seasons (e.g., deeper waters, higher stratification). Because the greatest DRs occurred when the glider was at midwater depths, future glider-based acoustic surveys of cetaceans may benefit from examining call detections only at certain receiver depths (e.g., midwater) for more robust spatial comparisons among regions and with oceanographic drivers. Finally, the estimation of DR helps elucidate the relationship between whale presence, the ocean environment, and potential anthropogenic impacts on cetaceans in an area thereby aiding conservation efforts.

Financial support was provided by the Office of Naval Research Grant No. N00014-17-1-2606. Partial funding was provided by Centro COPAS Sur-Austral PFB31 and ANID APOYO CCTE AFB170006, and Centro COPAS Coastal FB210021, funded by the Chilean Agencia Nacional de Investigación y Desarrollo (ANID) (https://www.anid.cl/). All deployments were carried out under SHOA resolutions 13270_24_136 and 13270_24_155. A matlab, 2021a Academic License to the University of Concepción was used in this research. L.G. was supported by a scholarship from COPAS Coastal. L.B.-R. was also supported by IDEAL ANID 15150003. This work contributes to the Southern Ocean Research Partnership (SORP) Acoustic Trends Working Group.

The authors have no conflicts of interest to disclose.

The raw data supporting the conclusions of this article will be made available by the authors upon request, without undue reservation.