Volume 127, Issue 7 e2021JC018356
Research Article
Open Access

Extent and Magnitude of Subsurface Anomalies During the Northeast Pacific Blob as Measured by Animal-Borne Sensors

Rachel R. Holser

Corresponding Author

Rachel R. Holser

Institute of Marine Sciences, University of California, Santa Cruz, CA, USA

Correspondence to:

R. R. Holser,

[email protected]

Contribution: Conceptualization, Methodology, Software, Formal analysis, ​Investigation, Data curation, Writing - original draft, Visualization

Search for more papers by this author
Theresa R. Keates

Theresa R. Keates

Ocean Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Methodology, Software, ​Investigation, Data curation, Writing - review & editing

Search for more papers by this author
Daniel P. Costa

Daniel P. Costa

Institute of Marine Sciences, University of California, Santa Cruz, CA, USA

Contribution: Conceptualization, Resources, Writing - review & editing, Supervision, Project administration, Funding acquisition

Search for more papers by this author
Christopher A. Edwards

Christopher A. Edwards

Ocean Sciences Department, University of California, Santa Cruz, CA, USA

Contribution: Conceptualization, Methodology, Writing - review & editing, Supervision

Search for more papers by this author
First published: 04 July 2022
Citations: 3

Abstract

Marine heatwaves (MHWs) are prolonged warm water events that are increasing in frequency and magnitude due to rising global temperatures. The Northeast Pacific Blob was an unusually widespread MHW that affected ecosystems across the Northeast Pacific, from producers to top predators. Temperature and salinity data collected by northern elephant seals (Mirounga angustirostris) from 2014 to 2017 show significant (>2 sd) warm anomalies throughout the top 1,000 m of the water column, with peak warming in late 2015. Using temperature and salinity as a tracer of layers of constant density, we looked at how lateral advection may have contributed to the development of the Blob. Temperature and salinity anomalies and the expansion of the water column at the base of the pycnocline both indicate that northward advection of warm, salty water played an important role in the observed accumulation of warm water, in addition to surface warming. These findings contribute to our understanding of the physical dynamics of the Blob, especially the thermal content and structure of the water column, and offer mechanisms for its formation and maintenance, which are crucial to assessing the ecological effects of MHWs now and in the future.

Key Points

  • Animal-borne ocean sensors provide unique insights into the physical mechanisms of a marine heatwave

  • Significant sub-surface temperature anomalies were observed down to 1,000 m of depth between 2014 and 2017

  • Analysis of spice anomalies on deep isopycnals indicates that northward advection of water masses contributed to sustained deep anomalies

Plain Language Summary

Marine heatwaves, extended periods when ocean temperatures are abnormally warm, have occurred with greater frequency and magnitude over the last few decades. As the planet continues to warm, these events will increase, with substantial effects on marine life and on the socioeconomics of communities that depend on marine resources. Here we analyze temperature and salinity data collected by northern elephant seals to better understand the development of the North Pacific Blob, a marine heatwave that began in late 2013 and continued through 2015. This was the largest, longest marine heatwave on record, with surface temperatures over six degrees above normal. Elephant seals cover much of the Northeast Pacific Ocean when they forage, and the high-density data they collected provide a unique look into the evolution of this event. We found that abnormally warm temperatures extended down through the top 1,000 m of the ocean, and that subsurface warming continued until 2017, after surface temperatures had returned to normal. Our analyses show that the observed deep warming was likely caused in part by the northward movement of subtropical water deeper than 200 m. These findings help us understand the complexities of marine heatwaves and better predict ecosystem effects of future events.

1 Introduction

The Northeast Pacific Blob 2015 (the Blob) was the largest marine heatwave (MHW) on record and was categorized as “severe” based on a combination of magnitude, extent, and duration of the anomalies observed (Hobday et al., 2018; Holbrook et al., 2019). Marine heatwaves are discrete, prolonged, anomalously warm water events (Hobday et al., 2016). The surface signal of this MHW lasted for 2 years and, at its peak, the Blob covered 11.7 Mkm2 with temperature anomalies reaching +6.7°C (Holbrook et al., 2019). Sea surface temperature anomalies (SSTa) persisted through winter 2015/2016 and sub-surface anomalies lasted into at least 2017 (Freeland & Ross, 2019; Jackson et al., 2018). The Blob developed because of reduced winds and high sea level pressure anomalies over the Northeast Pacific (NEP) during the boreal winter of 2013/2014, which suppressed heat loss from the surface ocean and weakened cold-water advection (Bond et al., 2015; Whitney, 2015). After the onset of this MHW was the 2015/16 El Niño Southern Oscillation (ENSO) event, which was an extreme, mixed Central and Eastern Pacific El Niño, with forcing that was independent of the Blob (Santoso et al., 2017). This multi-year warm event appears to result in part from coupling between the North Pacific Gyre Oscillation (NPGO) and Pacific Decadal Oscillation (PDO), the two dominant modes of variability in winter SST in the NEP (Joh & Di Lorenzo, 2017).

While the dynamics of the surface features of the Blob have been well studied (e.g., Bond et al., 2015; Di Lorenzo & Mantua, 2016; Freeland & Ross, 2019; Jacox et al., 2019; Joh & Di Lorenzo, 2017; Schmeisser et al., 2019; Tseng et al., 2017; Whitney, 2015), work examining the subsurface structure of the anomaly is sparse and often geographically limited (e.g., Chao et al., 2017; Freeland & Ross, 2019; Jackson et al., 2021; Jackson et al., 2018; Zaba & Rudnick, 2016; Zaba et al., 2020, but see also Hu et al., 2017; Scannell et al., 2020; Zhi et al., 2019). Two studies examining multi-decade Argo and mooring time series across the NEP (Cummins & Masson, 2018; Hristova et al., 2019) provide insight into the relative contributions of lateral and vertical heat transport to maintaining the warm anomaly. Hristova et al. (2019) quantified geostrophic advection anomalies across the southern and northern boundaries of the North Pacific Current bifurcation region (40–50°N, 205–240°E) and found northward transport anomalies in the top 1,500 m across both north and south boundaries continuously from 2014 to 2016. Geostrophic transport in this region exhibits seasonal variability, with greatest northward transport during winter months when the Alaska Gyre is at its largest and strongest. The authors suggest that this geostrophic advection contributed to warming given the covariation of northward transport with temperature and salinity anomalies in the Gulf of Alaska (Hristova et al., 2019). However, Cummins and Masson (2018) suggest that, while anomalous lateral advection likely contributed to anomalies in the last half of 2016, during the first two years of the heatwave heat was transported downward through isopycnals by Ekman pumping. Scannell et al. (2020) also used Argo data to examine connections between surface and subsurface anomalies across four MHWs, including the Blob, with a focus on the 25.4–26.3 kg m−3 isopycnals and isobars down to 220 dbar. They consistently observed warm and salty waters during 2014–2015, particularly in water less dense than 26.5 kg m−3 and found that subsurface temperature anomalies were most strongly correlated with surface conditions lagged by 1–2 years. It is possible for subsurface heatwaves to occur independently of surface conditions, so understanding the relationship between surface and subsurface anomalies is important (Elzahaby & Schaeffer, 2019; Hu et al., 2021). As these studies show, the primary mechanisms for collecting data on subsurface water properties are the Argo profiling float network, moorings, gliders, and oceanographic cruises that target specific regions, all of which have limitations in resolution and coverage.

Advances in tagging technology have enabled the use of animals as ocean-sensing platforms. Animal-borne ocean sensors, such CTD-SRDLs (Conductivity Temperature Depth-Satellite Relay Data Loggers) allow us to collect data at higher temporal and spatial resolutions than those available from the Argo profiling float network, moorings, or ship-based surveys, providing a powerful addition to more traditional sampling methods and enhancing our understanding of the physical ocean (Biuw et al., 2007; Boehlert et al., 2001; Boehme et al., 20092010; Charrassin et al., 2010; Costa et al., 2012; McMahon et al., 2021; Roquet et al., 2014; Siegelman et al., 2019). In this study, we examine temperature and salinity data collected by 71 northern elephant seals (Mirounga angustirostris) across the NEP in 2014–2017. Elephant seals dive continuously throughout their 3,000–12,000 km migrations, with an average dive depth of 516 ± 52 m and a maximum recorded depth of 1,735 m (Robinson et al., 2012). The data collected from instruments carried by these animals allow us to examine the magnitude of subsurface anomalous conditions during this MHW, the temporal evolution of those anomalies, and the potential contribution of lateral advection to observed conditions.

Marine heatwaves are a type of extreme climate event that are poorly understood, largely due to a lack of observation at adequate temporal and spatial scales. An extreme climate event is the occurrence of a weather or climate variable (i.e., – temperature, rainfall) that is outside of the historical range for that variable and location (Ummenhofer & Meehl, 2017). Extreme climate events can result in ecosystem-level perturbations, changing the distribution and abundance of species which can have cascading trophic effects. MHWs have disrupted entire marine ecosystems, from productivity (Kahru et al., 2018; Peña et al., 2018; Whitney, 2015) to top predator survival and reproduction (Gálvez et al., 2020; Hipfner et al., 2020; Jones et al., 2018; Osborne et al., 2020; Wild et al., 2019), and consequently have economic effects on local communities that depend on fisheries. The warming associated with the Blob had extensive ecological consequences in both the Gulf of Alaska and the California Current System (Cavole et al., 2016). Like other extreme climate events, MHWs are expected to increase in frequency, magnitude, and duration in the future and are a growing threat to vulnerable ecosystems (Frolicher et al., 2018; Oliver et al., 2018; Smale et al., 2019). The mechanisms underlying the build-up, persistence, and decay of MHWs are not well understood, nor is the ecological response, which requires knowing the magnitude and duration of a disturbance in addition to the sensitivity and adaptive capacity of the biological system (Frolicher & Laufkotter, 2018; Smale et al., 2019).

The purposes of this paper are (a) to report on a high-resolution, animal-collected CTD data set that complements other observing platforms in the NEP during years spanning the Blob and (b) to investigate deeper water property signatures measured by the animals and its implications for lateral transport. Section 2 details the methods used for data collection, processing, quality control, and calculation of anomalies. Analytical methods for examining spatial and temporal distributions of anomalies are described in Section 3, and results and discussion are presented in Section 4. The project summary and conclusion are in Section 5.

2 Data Collection and Processing

2.1 Animal Handling

Data were collected from the northern elephant seal colonies at Año Nuevo State Park, San Mateo County, California, U.S.A., and San Nicolas Island, Ventura County, California, U.S.A. All animal handling was conducted under NMFS permits #17952 and #19108, and with the approval and oversight of the UCSC Institutional Animal Care and Use Committee.

We deployed CTD-SRDLs (Sea Mammal Research Unit, St. Andrews, UK; a full technical description of these instruments can be found in (Boehme et al., 2009)) on adult female northern elephant seals at Año Nuevo State Park (37.11°N, 237.67°E) during the post-breeding (PB: March–May) and post-molting (PM: June–January) foraging trips from 2014 to 2017, and at San Nicolas Island (33.25°N, 240.50°E) during PM 2015. Animals were sedated and instruments attached following established protocols (Robinson et al., 2012). CTD-SRDLs collect CTD casts (or profiles) every 4–6 hr throughout deployment and data are collected and stored at a 1 Hz sampling rate (Boehme et al., 2009). Profiles are collected on the ascent phase of the dive, after animals have spent ∼15 min at depth. Elephant seals typically ascend from depth at 0.5–1.5 m/s, providing a variable vertical resolution of around 1 m. Subsets of these data are transmitted through the Argos satellite system throughout the animal's trip (Boehme et al., 2009). Whenever possible, instruments were recovered upon the animals' return and the full-resolution stored data were downloaded. When animals did not return or were not accessible, transmitted data were used instead. A summary of the number of deployments and the resulting data used in the following analyses can be found in Table 1.

Table 1. Summary of Conductivity Temperature Depth-Satellite Relay Data Loggers Deployments on Elephant Seals From 2014 to 2017
# CTD deployments #100+ m casts #500+ m casts #800+ m casts
Year Post-Breed Post-Molt Temp Salinity Temp Salinity Temp Salinity
2014 11 12 7,190 7,190 4,712 4,712 553 553
2015 4 12 11,945 10,346 6,666 5,664 452 311
2016 11 7 9,346 5,146 4,699 2763 585 152
2017 0 7 8,133 5,710 4,240 2,841 445 182
Total 26 38 36,614 28,392 20,317 15,980 2,035 1,198
  • Note. Shows the number of casts with temperature and salinity data that reach to at least 100, 500, and 800 m of depth.

2.2 CTD Post-Processing

All CTD data were processed following established methods and are available through the MEOP Project (Treasure et al., 2017). Post-processing of seal-derived temperature and salinity data and comparison to traditional CTD measurements are detailed in Roquet et al. (2011). Temperature and salinity were corrected for thermal mass effects, which can be problematic in regions with strong thermoclines (Mensah et al., 2018; Siegelman et al., 2019). Salinity was also adjusted for spiking and density inversions, with resulting accuracies estimated at ±0.02°C and ±0.03 g kg−1 (Siegelman et al., 2019). All measurements are quality controlled following the quality criteria used for data from Argo profiling floats (Wong et al., 2018). For this study, data qualities 1–3 (1—good data; 2—probably good; 3—bad but potentially correctable) were exported for further quality control and then included in analyses. The number of profiles of temperature and salinity data (shown in Table 1) differs in some years due to the repeated use of some instruments causing the conductivity cell to fail. In these, cases temperature profiles were still collected but salinity data were either not collected or not retained. Six additional instruments which had undergone multiple cycles of deployment and battery replacement were found to have conductivity sensor drift resulting in a steady decrease in salinity values at all depths across their final deployment irrespective of their geographic location (e.g., salinity anomalies of −0.8 g kg−1 at 800 m depth). All measurements for those deployments were removed from the data set. CTD-SRDLs also provide locations multiple times per day via the Argos polar-orbiting satellite system. At the latitudes that northern elephant seals inhabit, there are typically around 30 satellite passes per day, with each pass lasting between 5 and 15 min (CLS, 2022). Elephant seals surface for only 2–3 min between dives lasting from 20 to 40 min (Robinson et al., 2012). Consequently, around 8–10 locations per 24-hr period are received for each individual, therefore additional processing is needed to locate each CTD cast collected. Raw locations were Kalman filtered and then passed through a speed filter to eliminate locations that would represent unrealistic animal movement (Roquet et al., 2014). Retained locations were linearly interpolated through time and assigned to CTD data based on the timestamps of the temperature and salinity profiles (Figure 1 and Figure S1 in Supporting Information S1) (Roquet et al., 2014).

Details are in the caption following the image

Distribution of Conductivity Temperature Depth casts with both temperature and salinity data collected in 2014–2017, each point represents a single cast. The solid black box outlines the core Blob region, and the dashed rectangle outlines the region used to create zonal sections.

2.3 Anomaly Calculations

Potential density (σθ), absolute salinity (SA), and conservative temperature (Θ) were calculated from the quality-controlled temperature and salinity data using the Gibbs Sea Water toolbox (McDougall & Barker, 2011). CTD data were vertically averaged to the depth resolution represented in climatology data: 5 m bins from 0 to 100 m depth, 25 m bins from 100 to 500 m, and 50 m bins below 500 m depth. While climatologies built from measured values are generally more accurate than modeled climatologies, modeled climatologies provide greater spatial coverage, making that data set preferable for our anomaly calculations. To assess the error in modeled climatology values relative to measured climatology values from World Ocean Atlas 2018 (WOA18; 1981–2010 monthly climatology, 1 × 1° grid) (Locarnini et al., 2018; Zweng et al., 2018), we calculated linear regressions of urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0016 ∼ urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0017 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0018 ∼ urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0019 (see Figure S2 and Table S1 in Supporting Information S1). For all the following analyses, modeled (an) climatologies were used. Temperature and salinity anomalies were calculated relative to modeled climatologies (i.e., urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0020 = ΘDataurn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0021) extracted from WOA18 using both time and location to find the nearest match to each CTD cast collected along the animals' tracks. As an additional layer of quality control, means and standard deviations (sd) of both temperature and salinity anomalies were calculated for each depth range (<100 m, 100–500 m, and 500+ m), and outliers, defined as values ±8 sd from the overall mean, were removed (0.022% of 2,078,694 data points).

The uncertainty for each temperature anomaly value relative to climatology was calculated (where possible) using standard error propagation: urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0022 where urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0023 = 0.02 (Siegelman et al., 2019) and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0024 is provided for climatology. The same process was followed for salinity anomalies with urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0001 = 0.03 g kg−1 (Siegelman et al., 2019). The resulting distributions of temperature and salinity anomalies, standardized to urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0025 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0026, respectively, are shown in Figure 2, separated into by potential density into surface (σθ < 25.25), intermediate (25.25 ≥ σθ ≤ 26.5), and deep water (σθ > 26.5). The characteristics of these distributions (i.e., mean, standard deviation, skewness, kurtosis) are summarized in Table 2, along with the standard deviation of the difference in an and mn anomaly fields for both Θ and SA (see also Table S1 in Supporting Information S1).

Details are in the caption following the image

Distribution of all temperature (blue, left) and salinity (green, right) anomaly data in surface (σθ < 25.25 kg m−3), mid (25.25 ≥ σθ ≤ 26.5 kg m−3), and deep water (σθ > 26.5 kg m−3). Histograms are of temperature and salinity anomalies normalized to error with black lines indicating a normal distribution for each. Dashed vertical lines indicate ±2sd.

Table 2. Summary of the Distributions of urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0027 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0028 (Standardized to urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0029 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0030 Respectively) From Surface, Mid, and Deep Water
Surface (<1025.25) Mid (1025.25–1026.5) Deep (>1026.5)
Temp Salinity Temp Salinity Temp Salinity
Mean 1.42°C −0.02 g kg−1 0.19°C 0.09 g kg−1 0.16°C 0.05 g kg−1
sd 1.30°C 0.21 g kg−1 1.04°C 0.24 g kg−1 0.43°C 0.0 g kg−1
Standardized mean 2.09 −0.06 0.70 1.05 0.69 1.20
Standardized sd 4.22 1.98 3.11 3.35 2.32 2.88
Skewness 11.6 0.07 3.99 0.66 6.27 1.85
Kurtosis 209.4 11.3 213.7 10.6 139.5 30.0
Mn-An sd 0.53°C 0.13 g kg−1 0.42°C 0.14 g kg−1 0.21°C 0.032 g kg−1

2.4 Argo Comparison

Animal-collected data were compared to Argo profiling float CTD data to further validate the observations. All Argo profiles from the northeast Pacific during 2014–2017 were downloaded from the World Ocean Database (Boyer et al., 2018). For each animal-collected CTD cast, the Argo data were searched for profiles within ±1 day and ±0.25° latitude and longitude. This yielded 86 unique Argo profiles that were closely co-located in space and time to 304 unique seal profiles (Figure S3 in Supporting Information S1). The seal-collected profile that was closest in space to each unique Argo profile was selected for comparison. Linear models were fit to both temperature and salinity data (SealData ∼ ArgoData; see Figure S4 in Supporting Information S1), and the mean of the residuals was calculated for both (temperature 0.583°C; salinity 0.136 g kg−1). Using the same methods, we also compared 86 instances when different Argo profiles overlapped within the same spatial-temportal window used to match Argo and seal-collected data. The standard deviations of the residuals from linear models for both temperature and salinity were very similar to that found when comparing seal-collected data to Argo profiles (temperature 0.396°C; salinity 0.089 g kg−1; also see Figure S4 in Supporting Information S1).

3 Analytical Methods

3.1 Spatial and Temporal Distribution of Anomalies

T-S diagrams were generated on a 2 × 2° grid across the full range of data collected (Figures 3 and 4). To examine the development of the warm anomaly both spatially and temporally, we generated seasonal zonal sections (40–44°N, 180–240°E) across the sampling period. Within each season (3-month periods: Winter – DJF, Spring – MAM, Summer – JJA, Fall – SON), data were averaged at each depth bin (as described in Section 2) across 1° longitude bins. To account for the varying number of data points contributing to each interval, error propagation for each mean was calculated as follows: urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0002, where N is the number of measurements contributing to the mean. For urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0003 the anomaly was considered not significant and assumed to be 0. The vertical resolution of climatology varies with depth. We generated profiles with a consistent vertical resolution of 5 m across all depths (equivalent to the resolution of near-surface climatology data) using linear interpolation. The same process was used to create sections of urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0004 (Figure 5). Similarly, monthly average temperature anomalies (urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0005) by depth were calculated for all data within the core Blob region (40–50°N, 210–230°E) and we used the same method to determine anomaly significance as described for the zonal sections. These data were used to create contour plots of urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0006, urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0007, and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0008 (Figures 6 and 7).

Details are in the caption following the image

Map of data distribution showing locations (with corresponding letters) of T/S diagrams show in Figure 4.

Details are in the caption following the image

Temperature-salinity plots of data collected by elephant seals from 2014 to 2017 within core Blob region (see Figure 3). Each figure represents a 2° × 2° box in which data were collected from 0 to 1,000 m depth, panel letters correspond to labeled boxes in Figure 3. Contour lines show density anomaly (σθ) calculated at 0 m depth. Black dots show all climatology values corresponding to the locations and months of data collected. The black line is the mean climatology, weighted to match the distribution of observed values. Water mass T/S ranges (as defined in Emery (2001)) are indicated with boxes: Pacific Subarctic Upper Water (PSUW, 3.0–15.0, 32.6–33.6, green), Pacific Subarctic Intermediate Water (PSIW, 5.0–12.0, 33.8–34.3, blue).

Details are in the caption following the image

Depth (0–600 m) by longitude sections of conservative temperature anomaly (a) and salinity anomaly (b) averaged at 1° longitude. Each section includes all data from 40 to 44°N (dashed box in Figure 1) during a 3-month season from Dec 2013—Nov 2017. In both A and B each row depicts a single season. Gray indicates data gaps.

Details are in the caption following the image

Monthly averaged temperature anomaly (a) and standardized anomaly (b) within the core region of the Blob (40–50°N × 150–130°W, solid box in Figure 1) from Jan 2014–Dec 2017. Solid white sections indicate periods when no data were collected within the geographic range. Dashed black lines are 300 m depth reference. Corresponding North Pacific Gyre Oscillation, Pacific Decadal Oscillation, and El Niño Southern Oscillation (ENSO) indices are indicated at the bottom.

Details are in the caption following the image

(a) Climatological monthly averaged spice anomaly (monthly-annual climatology) within the core region of the Blob (40–50°N × 150–130°W, solid box in Figure 1). (b) Observed monthly spice anomaly within the core region of the Blob, from Jan 2014–Dec 2017. Solid white sections indicate periods when no data were collected within the geographic range.

3.2 Isopycnal Surface Variation

Deep warming anomalies can result from surface processes or from lateral advection. To examine the potential contribution of lateral advection to warming, mean Θ, SA, spice, and depth (d) were calculated in a 1 × 1° grid for each year from elephant seal observations and from annual WOA18 climatology on the σθ = 26.5 kg m−3 and σθ = 27.0 kg m−3 surfaces. We subtracted climatological isopycnal depth from observed depth to find depth anomalies of both isopycnals (Figure 8).

Details are in the caption following the image

Depth anomaly of isopycnal surfaces σθ = 26.5 kg m−3 (left) and σθ = 27.0 kg m−3 (right) from climatology (WOA18 annual, 1981–2010). Red (blue) or negative (positive) depth anomaly values indicate the surface is shallower (deeper) than climatology.

North of 45°N, the 26.5 kg m−3 density surface is near the base of the main pycnocline while 27.0 kg m−3 is well below the pycnocline but still has high density elephant seal data coverage. These isopycnals were selected due to their presumed isolation from local surface processes. To test that assumption, we calculated mixed layer depths (MLD) for every cast using the potential-density algorithm developed by Talley and Holte (2009) and compared them to observed isopycnal depths to ensure that the isopycnals under examination were generally deeper than the MLD and therefore isolated from surface mixing. This algorithm calculates a suite of potential MLD values before selecting a final MLD estimate based on patterns within the possible MLDs (Talley & Holte, 2009). In addition, we compared monthly climatological MLD generated from Argo float data (Holte et al., 2017) to monthly climatological isopycnal depths (calculated from WOA18). Both comparisons showed minimal winter ventilation on the 26.5 kg m−3 isopycnal, and no ventilation on 27.0 kg m−3 (see Figure S5 in Supporting Information S1).

Given the isolation of these isopycnals from the mixed layer, it is reasonable to assume that the observed spice anomalies result predominantly from lateral, along-isopycnal transport, and not diapycnal mixing. To identify the most likely geographical origin of observed water properties, we used a non-dimensional objective function to find the nearest potential source water identified within climatology. For each observation, we identified the minimum of the sum of squared deviations, normalized by their climatological variances, between observed values and climatological fields for each climatology grid cell within 5° of our in situ data. Objective functions (also referred to as cost functions) are widely used in data assimilation studies (e.g., Bennett, 2005; Edwards et al., 2015; Wunsch, 1996) and for parameter estimation (e.g., Mattern & Edwards, 2017). Specifically, we searched for the climatological location, Ji,j, that yielded the smallest value for each data cell (Datak,l):
urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0009
where urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0010 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0011, the mean standard deviations for climatological temperature and salinity for the region where seals collected data, were calculated for each density surface (σθ = 26.5: urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0012 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0013 g kg−1; σθ = 27.0,: urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0014 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0015 g kg−1). A Monte Carlo approach accounted for error in both measured and climatology fields: these calculations were repeated 10,000 times while adding a random, zero-centered normally distributed perturbation to both urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0031 and urn:x-wiley:21699275:media:jgrc25116:jgrc25116-math-0032 on each iteration. We found the mean location (latitude and longitude) of the resulting minimum Ji,j value for each grid cell. This procedure found the nearest location within climatology that has values for temperature and salinity that best match seal-collected measurements. We generated density surface maps of temperature anomaly with overlaid arrows to indicate the direction and distance of lateral movement required to account for the temperature and salinity values seen in each year compared to climatology. The anomalies were sustained over multiple years, so we also calculated the difference in lateral advection from 1 year to the next to examine the cumulative movement required to account for year-to-year spice anomalies in the region (Figure 9). Given differences in data coverage from year to year, average values on a 3 × 3° grid were used for these calculations.
Details are in the caption following the image

Mean spice anomaly on isopycnal surfaces σθ = 26.5 kg m−3 (left) and σθ = 27.0 kg m−3 (right) for years 2014, 2015, 2016, and 2017. Green/blue indicates negative spice anomalies (more minty than climatology), and orange/red indicates positive spice anomalies (spicier than climatology). The numeric range of the color scale is narrower for the deeper isopycnal on the right. Arrows originate at geographical locations having closest climatological (WOA18 annual, 1981–2010) T/S properties as observed properties at arrow terminations and indicate displacement from climatology for each year.

4 Results and Discussion

4.1 Distribution of Anomalies

The distributions of standardized anomalies of both temperature and salinity had means >0 across all depth categories, with the strongest anomalies (both absolute anomaly and standardized) in surface water (Table 2). Histograms of standardized anomalies (Figure 2) illustrate the non-Gaussian distribution of temperature anomalies, with high peaks around the mean and significant right skewness (Table 2), resulting in high frequency of anomalies between 2 and 5sd and very few below -2sd at all depths. These distributions illustrate how ubiquitous the warm anomaly was throughout the geographic range and time period sampled here.

Surface temperature anomalies were most pronounced in the core region of the Blob in 2015 and 2016 and were not always accompanied by density-compensating salinity anomalies (e.g., Figures 4a–4f). This warm, low-density water would increase stratification and further enhance the surface warm anomaly due to reduced mixing. In offshore profiles (e.g., Figures 4g–4l) there are consistent warm anomalies and corresponding salt anomalies across isopycnals to the base of the main pycnocline at ∼σθ = 26.5 kg m−3. Nearshore profiles in the region show warm, fresh anomalies that are more isolated to the surface, especially in 2015 and 2016 (e.g., Figures 4e and 4f). Profiles from the periphery of the region with most intense warming, still exhibit some surface warming, but much more notable variability in salinity anomalies, with pronounced freshening in 2016, particularly around the 26.0 kg m−3 isopycnal (see Figure S6 in Supporting Information S1). Zhi et al. (2019) and Liu and Huang (2012) suggest that subduction north of the Kuroshio Extension region induced subsurface fresh anomalies that then propagate eastward toward the dateline and into the North Pacific Current, which may account for the fresh anomalies seen.

4.2 Temporal and Spatial Development

The zonal sections from 40 to 44°N show significant warming throughout, both in surface and subsurface water, although the most pronounced and sustained warming occurred east of 200°E (Figure 5a). While surface anomalies peaked in late 2015, the subsurface anomaly at 200–400 m was sustained through at least the end of 2017. Within the core region of the Blob, the warm temperature anomaly was initially concentrated in the top ∼100 m of the water column, but by late summer 2014 anomalies extended down to 300 m depth (Figure 6). Surface warming became most pronounced in late 2015, with normalized temperature anomalies >5 sd from climatology in the top 100 m. Subsurface anomalies were high throughout 2015–2017. Between 100 and 500 m depth range, ΘAnom of 0.4–1.5°C (2–5 sd) occurred throughout the event. Below 500 m, the maximum anomaly occurred in late 2015 at around 800 m depth (+0.21°C, 4.2 sd). This maximum is both earlier in the event and deeper in the water column than predicted or observed in previous studies (Chao et al., 2017; Freeland & Ross, 2019; Hu et al., 2017). Similarly, positive spice anomalies began in the top 100 m of the water column in 2014 then rapidly began to be observed and were sustained below 300 m into 2017 (Figure 7). The timing, depth, and magnitude of these spice anomalies are substantially different from the seasonal cycle seen in climatology (Figure 7a).

Bands of positive salinity anomalies are seen at ∼150 m in 2014–2015 (Figure 5b) and may be due to reduction in the mixed layer depth because of the profound surface warming occurring at that time. Zhi et al. (2019) shows that fresh anomalies (<−0.2) in the upper 100 m preceded the onset of the marine heatwave (2012–2013) and contributed to shallowing of the mixed layer depth, which then would have inhibited heat penetration at depth and enhanced the development of surface warm anomalies. We cannot directly corroborate this pre-conditioning here, but there are fresh anomalies in 2014 at around 200 m depth, underneath areas with stronger surface warming at that time, which may be a remnant of the eastward-propagating fresh anomaly they reported (Zhi et al., 2019).

Peak warm anomalies co-occurred with the 2015/16 ENSO event (Figure 6c), which was an extreme, mixed Central and Eastern Pacific El Niño (Santoso et al., 2017). While past El Niño events have been linked to sustained warming in the Gulf of Alaska (Whitney & Freeland, 1999), the expression of the 2015/16 event did not match previous observations, particularly compared to prior strong El Niño's (1983/84 and 1997/98) (Jacox et al., 20152016; Santoso et al., 2017; Zhong et al., 2019). Because of the Blob, the Gulf of Alaska and California Current System were both in an unusually warm state prior to the onset of El Niño, which likely influenced the difference in observed effects. For example, the California Current System experienced upwelling favorable winds in summer and fall and shoaling of the 26.0 kg m−3 isopycnal, whereas the opposite was true in prior events (Jacox et al., 20152016). Both positive PDO and ENSO can result in surface cooling in the central north Pacific. The intrusion of cool, fresh water in the second half of 2016 may be related to these dynamics (Figures 5 and 6). Although the cooling coincides with the end of the 2015–2016 ENSO event in the tropics, generally there is a 2–3 months lag in expression in the NEP (Santoso et al., 2017).

4.3 Vertical Displacement of Isopycnals

Isopycnal depth variability in the Gulf of Alaska from 2004 to 2018 was examined by Cummins and Masson (2018). They found that positive PDO values (warmer regional SST) and strengthening gyre circulation are associated with downward pycnocline displacement through the top 1,000 m in the region adjacent to the west coast of North America. The strong positive PDO signal during the Blob should have resulted in downward movement of isopycnals in 2014–2017. They found that a model driven by Ekman pumping generally had good agreement with observed vertical displacement of the base of the pycnocline starting in 2004, however this relationship broke down during the Blob (2015–2017), suggesting different mechanisms driving vertical variability during the MHW (Cummins & Masson, 2018). A longer timescale analysis of data from Station P (1960–2018) completed by Cummins and Ross (2020) found that isopycnals below the permanent pycnocline exhibit a downward trend in time and that the deeper isopycnals (27.4 and 27.8 kg m−3) are separating from shallower surfaces (26.8 and 26.3 kg m−3). At Station P, isopycnals associated with the permanent pycnocline were similar to 1960 or slightly deeper in 2014–2017, while deeper isopycnals exhibited a downward displacement compared to 1960 (Cummins & Ross, 2020).

The data presented here shows regional and temporal variation in the depth anomaly of the 26.5 kg m−3 isopycnal, with shoaling more evident in the western transition zone and a neutral or deepening signal in much of the California Current and Golf of Alaska (Figure 8), generally in agreement with the findings of both Cummins and Masson (2018) and Cummins and Ross (2020). The 27.0 kg m−3 isopycnal, however, is shallower than climatology in all years, and the magnitude of shoaling increases across the period. In 2014 and 2015 we see depths 20–50 m shallower than climatology scattered throughout the NEP. In 2016 the western part of the region has shoaled by 80 m or more, and by 2017 80–100+ m shoaling is evident throughout the sampled region. This differs from expected conditions when there is a positive PDO and strengthened Alaska Gyre (Cummins & Masson, 2018; Hristova et al., 2019), as well as from observations from Station P (Cummins & Ross, 2020), although much of the range sampled here is south of Station P. These findings suggest that other mechanisms may be influencing isopycnal depth during this MHW and warrant further investigation.

4.4 Lateral Advection

The potential contribution of lateral advection along subsurface isopycnals to maintaining the warm anomaly in the NEP has been assessed in two studies using Argo float data. Hristova et al. (2019) quantified geostrophic advection anomalies across the southern and northern boundaries of the North Pacific Current bifurcation region (40–50°N, 205–240°E) and found northward transport anomalies in 0–300 m at both boundary latitudes from 2014 to 2016. Geostrophic transport in this region exhibits seasonal variability, with greatest northward transport during winter months when the Alaska Gyre is at its largest and strongest (Hristova et al., 2019). Hristova et al. (2019) suggest that this advection contributed to warming given the covariation of northward transport with temperature and salinity anomalies in the Gulf of Alaska. However, Cummins and Masson (2018) suggest that anomalous advection likely contributed to anomalies in the last half of 2016, but that heat was transported downward through isopycnals during the first two years of the heatwave.

In the seal-collected data, both deep density surfaces show significant, coherent northward movement into the core region of the Blob in 2014, and northward along the coastal margin (Figure 9). Looking at year-to-year cumulative movement, 2015 showed little additional lateral advection, but subtropical water again shifted northward into the North Pacific Transition Zone in 2016 and 2017. Hristova et al. (2019) saw the strongest northward transport anomalies from late 2014 through mid-2016. To assess the relative contributions of temperature and salinity to the objective function calculations, we completed the calculations as described above using only the temperature or salinity terms. Both salinity and temperature anomalies are independently explained by consistent northward advection in the core warming region and along the coast into the Gulf of Alaska (not shown).

There is also a notable cool, fresh sub-surface anomaly seen in the west on the 26.5 kg m−3 density surface in 2016 and 2017 (and was also evident in the zonal sections—Figure 5). Our calculations here indicate southward transport from the Alaska Stream as potential source water for those anomalies. The Alaska Gyre was both larger and stronger throughout the Blob, with peak gyre strength (maximum transport) at the start of 2016 (Hristova et al., 2019). Increased overall transport combined with the southward shift of the zero meridional transport contour could have resulted in an influx of cool fresh water to that region in 2016.

4.5 Biological Implications

The occurrence of several significant MHWs in the last decade has provoked investigation into the biological response to these extreme events (Frolicher & Laufkotter, 2018; N. J. Holbrook et al., 2019; Neil J. Holbrook et al., 2020; Jacox et al., 2020; Oliver et al., 2018; Smale et al., 2019). Changes in distribution, phenology, and foraging behavior are all expected responses to ongoing warming (Evans & Moustakas, 2018; Hazen et al., 2012), but the sudden and extreme nature of MHWs has the potential to cause more dramatic effects (Frolicher & Laufkotter, 2018; Jacox et al., 2020). Studies to date show that MHWs significantly alter biological systems due to both extrinsic physical forcing (e.g., altered circulation, reduced mixing, reduced dissolved oxygen, etc.) and intrinsic physiological limitations (e.g., increased metabolic rate in ectotherms), and generally the results are ecologically and economically detrimental (Cavole et al., 2016; Frolicher & Laufkotter, 2018; Holbrook et al., 2020; Jacox, 2019; Jacox et al., 2020). The magnitude of the effect on a particular system (or organism) is dependent on the characteristics of the heatwave (duration, intensity, depth) and the ability of the system (or organism) to tolerate or adjust to the extreme conditions (Smale et al., 2019; Wernberg et al., 2013). Forecasting the ecological consequences of MWHs requires a better understanding of the physical dynamics and the potential response of taxa at all trophic levels.

The warming associated with the Blob had extensive ecological consequences in both the Gulf of Alaska and California Current System, spanning all trophic levels (Cavole et al., 2016). Reduced water column mixing resulted in overall suppressed productivity (Kahru et al., 2018; Peña et al., 2018). In the Northern California Current, the expression of upwelling in 2014–2016 was weak despite unusually strong upwelling-favorable winds (Peterson et al., 2017). Altered plankton assemblages were seen throughout the NEP, generally favoring smaller, warmer-water species (Batten et al., 2017; Irigoien et al., 2020; Jiménez-Quiroz et al., 2019; Peterson et al., 2017). Low biomass and smaller prey species can reduce the efficiency of energy transfer to higher trophic levels, causing reductions in body condition, reproductive success, and survival in predator species (Basilio et al., 2017; Gálvez et al., 2020; Hipfner et al., 2020; Jones et al., 2018; Osborne et al., 2020; Piatt et al., 2020; Suryan et al., 2021; Thalmann et al., 2020; Thompson et al., 2019; von Biela et al., 2019). Warm water conditions also supported massive harmful algal blooms resulting in domoic acid buildup in shellfish along the west coast of North America (McCabe et al., 2016; Ryan et al., 2017; Trainer, 2017; Zhu et al., 2017). These algal blooms caused the closure of several fisheries, resulting in substantial economic losses in those communities (Ritzman et al., 2018).

Shifts in species distribution were seen throughout the region and at all trophic levels (Bond et al., 2015; Lonhart et al., 2019; Sanford et al., 2019). Changes in distribution can result both from advection of plankton species (or larval plankton stages of nekton) due to large-scale circulation changes, and from the active movement of nekton to avoid unfavorable conditions. Populations on the warm edge of their range are particularly vulnerable to MHWs, but taxa that are highly mobile are more likely to be able to adjust to new conditions (Cavole et al., 2016; Smale et al., 2019).

Little is known about the potential effect of MHWs on mesopelagic systems. Proud et al. predict that under continued global warming, the primary deep scattering layer will shoal by ∼30 m and fish biomass in the mesopelagic will increase over the next century (Proud et al., 2017). The sustained, deep warm anomalies we observed (Figures 5 and 6) may reduce oxygen content, and the combined temperature and oxygen conditions could alter both the vertical distribution and the vertical migration range of nekton species. The northward advection evident on deep isopycnals (Figure 9) could also contribute to altered species distributions, shifting subtropical mesopelagic species into the Gulf of Alaska. Analyzing the foraging behavior and success of mesopelagic predator species during the Blob could provide insight into changes in the mesopelagic system during MHWs. Predators integrate lower trophic levels; disturbances that impact primary producers or low-level consumers can be seen in dietary shifts, habitat use changes, reductions in reproductive success, or even reduced survival in predator populations (Fretwell, 1987; Lindeman, 1942; Scherber et al., 2010). Northern elephant seals exhibited reduced reproductive output following the first year of the Blob (Holser et al., 2021). Further investigation of their behavior paired with the oceanographic data presented here would benefit our understanding of MHW effects on mesopelagic biomass.

5 Summary and Conclusions

In this study we analyzed a rich data set that contributes to our understanding of the 3-dimensional extent, magnitude, and development of the Northeast Pacific Blob 2015. CTD casts collected by northern elephant seals provide temperature and salinity profiles for the top 1,000 m of the water column at higher temporal and spatial resolution than was achieved with more conventional means during this event. Standardized temperature anomalies >2 sd were found throughout the heatwave at all depths, with peak deep anomalies of 4.2 sd at 800 m in 2015 (Figure 7). These anomalies were deeper and greater than those previously reported for the subsurface (Chao et al., 2017; Hu et al., 2017), and occurred earlier in the event than predicted (Freeland & Ross, 2019). Lateral advection at the base of the pycnocline likely contributed to the observed deep subsurface temperature and salinity anomalies. Spice anomalies in the core region of the Blob suggest northward advection of warm, salty water (Figure 9) drove subsurface warming.

Cummins and Masson (2018) found no evidence of westward advection on similar deep (200–500 m) isopycnals but did suggest that either horizontal advection across temperature gradients or vertical diffusion may have contributed to the formation of the Blob. Scannell et al. (2020) discussed several potential mechanisms for subsurface expression of MHW anomalies, including lateral advection, although their analyses focused more on the contribution of vertical processes. By assuming no diapycnal mixing across deeper isopycnals, we find that northward lateral transport along isopycnals could be responsible for the subsurface warm, salty water seen throughout the region from 2014 to 2017. This finding also aligns with the northward transport anomalies observed across both northern and southern boundaries of the bifurcation region (Hristova et al., 2019). Lateral advection is supported in other works analyzing subsurface features of the Blob (Chao et al., 2017; Zaba et al., 2020), although those analyses were more geographically constricted and shallower (0–300 m) than what is presented here. Our analyses focused on deep isopycnals, isolated from surface ventilation, and indicate that lateral advection was an important process in generating the anomalies seen below 200 m. At present, it is not clear if this advection is directly or indirectly related to the local surface warming of the Blob. More likely, motion on the 26.5 kg m−3 and 27 kg m−3 surfaces results from large scale circulation patters related to broader wind stress curl. Nonetheless, anomalies deeper than 500 m in the geographical region of the Blob are substantial and experienced by marine fauna during this period.

MHWs are expected to increase in frequency, magnitude, and duration as global temperatures continue to rise (Frolicher et al., 2018; Jacox, 2019; Oliver, 2019; Oliver et al., 2018). These events have significant ecological consequences for the affected systems, as well as economic consequences for the local communities that rely on those systems (Bograd et al., 2019; Frolicher & Laufkotter, 2018; Richerson et al., 2018; Ritzman et al., 2018; Smale et al., 2019; Smith et al., 2021). Understanding the physical mechanisms and associated climate drivers that cause and maintain MHWs is essential to predicting the onset and development of these events in the future and will help us anticipate and mitigate their ecological and economic consequences (Bograd et al., 2019; N. J. Holbrook et al., 2019; Smale et al., 2019; Smith et al., 2021).

Oceanographic data collected from animal platforms will continue to increase in abundance and type. As technological advances continue to produce new sensors in small, robust packages, a greater variety of parameters will be measured. Animal-borne ocean sensors have collected temperature, salinity, and, more recently, chlorophyll fluorescence data throughout the world's oceans (Boehme et al., 20092010; Guinet et al., 20132014; Keates et al., 2020; Roquet et al., 20112014), and data collected by Autonomous Pinniped Bathythermographs have already been incorporated into the World Ocean Database (Boyer et al., 2018). The high spatial and temporal resolution and low relative cost of these data make them a valuable addition to the Global Ocean Observing System (Boehme et al., 2010; Harcourt et al., 2019; McMahon et al., 2021). The analyses presented here highlight the complementary role these data play to more traditional ocean-sampling techniques, shedding light on physical processes that are otherwise difficult to measure and are critical to understanding and predicting a changing ocean.

Acknowledgments

D.P.C., R.R.H. and T.R.K. were supported by the Office of Naval Research (ONR) (Grant Nos. N00014-13-1-0134, N00014-18-1-2822). D.P.C. was further supported by the E&P Sound and Marine Life Joint Industry Programme (JIP) of the International Association of Oil and Gas Producers (IOGP) (Grant No. 00-07-23). Additional support for oceanographic data collection was provided by the Central & Northern California Ocean Observing System and the US Animal Telemetry Network. The authors would also like to thank Año Nuevo State Park for their ongoing support of elephant seal research. In additional, all the students, volunteers, and researchers who have contributed to the data collection completed at the Año Nuevo colony, in particular P.W. Robinson, L.A. Hückstädt, S.H. Peterson, and E.A. McHuron. Finally, we appreciate the thoughtful and constructive comments provided by three reviewers–your questions and recommendations helped us refine and clarify this manuscript.

    Data Availability Statement

    All data used in this research are publicly available through the MEOP Project (Treasure et al., 2017), the World Ocean Atlas 2018 (WOA18; 1981–2010 monthly climatology, 1 × 1° grid) (Locarnini et al., 2018; Zweng et al., 2018), and World Ocean Database (Argo profiling float data) (Boyer et al., 2018). All data analyses were completed using MATLAB© R2021a (2021) and required use of the Gibbs Sea Water toolbox (McDougall & Barker, 2011). All code used for these analyses are publicly available at GitHub and Zenodo (http://doi.org/10.5281/zenodo.6590749).