Deep-Water Renewal Events; Insights into Deep Water Sediment Transport Mechanisms

Deep-water renewal (DWR) events are characterized in the Strait of Georgia, Canada using 11 years of real-time physical and chemical oceanographic data and seafloor videos. At least 6 DWRs occur per year at 300 m water depth and each event continues for over 3 days. They initiate during neap tides and are associated with increased turbidity. In the spring, DWRs introduce cold, oxygenated and nutrient-poor waters, and in the fall they introduce warm, oxygen-depleted, nutrient-rich and saline waters. Although the timing and magnitude of DWRs differ from year to year, we demonstrate that they are not restricted to two seasons, but continue throughout the year. High-resolution videos of DWRs show that these events comprise a plume of high suspended sediment concentration that flows parallel to the basin axis and deposits approximately 1.5 cm per event.

www.nature.com/scientificreports www.nature.com/scientificreports/ oceanic properties of DWRs in 2009 which comprises one of the most complete datasets of the eleven years of data (Fig. 2). Other years are included as supplementary data ( Supplementary Fig. S1).
Variations in oceanographic parameters throughout the year render it challenging to compare individual DWRs both intra-and inter-annually. In particular, discharge from the Fraser River, variations in tidal currents, and climate conditions over the Pacific Ocean, and even living organisms (e.g., consuming excessive oxygen during certain times of the year) all appear to impact the timing and characteristics of DWRs 7,9,12,19 . Although there are some annual variations in the timing of these events, when eleven years of data are compared, generalities emerge in the character of these events ( Supplementary Fig. S1). We divide DWRs into three cycles based on their chemical characteristics. These cycles occur at different times in the year and the start and/or end of each cycle In general, temperature during C1 DWRs decreased, although some DWRs towards the end of C1 exhibited a gradual shift from a decreasing to a slight increasing pattern (e.g., 2009 and 2012; C1 in Fig. 2 and Supplementary  Fig. S1). Dissolved Oxygen (DO) shows a consistently increasing pattern for all DWRs in C1 while salinity does not show a well-defined trend with the exception of a generally increasing pattern during DWRs in the second half of C1 (Fig. 2). In 2009, the average temperature change during DWRs was −0.53 °C (standard deviation (stdev): 0.8), average salinity variation was +0.02 psu (stdev: 0.2), and average DO variation was +1.30 ml l −1 (stdev: 0.6). Turbidity also increased by an average of 6.56 Nephelometric Turbidity Units (NTU; Supplementary Table S1) (stdev: 11.4).
The second cycle (C2) extends from approximately June to October, in 2009. Fifty DWRs were recorded in C2 between 2009 and 2019, and 7 of these occurred in 2009. The chemical and physical characteristics of DWRs in C2 are consistent and of greater amplitude compared to those during other cycles. Salinity values show the most striking increasing trends with an average increase of 0.44 psu (stdev: 0.2) during events in 2009 (Fig. 2). Temperature typically shows a sharp increase at the beginning of C2 DWRs and slowly returns to ambient temperatures throughout the cycle. In 2009, the average temperature increase during C2 DWRs was 0.91 °C (stdev: 0.3). Dissolved oxygen generally decreases during C2 DWRs, and in 2009, the average DO decrease was 0.43 ml l −1 (stdev: 0.1). Turbidity during 2009 C2 DWRs increased on average by 6.56 NTU (Supplementary Table S1) (stdev: 5.7).
The third cycle (C3) extends from October to February, in 2009. Forty-seven DWRs were observed in C3 between 2009 and 2019 and three of these were in 2009. DWRs in C3 are characterized by weak signatures and they show noticeable variations between years ( Fig. 2 and Supplementary Fig. S1). Salinity and temperature variations are represented by increasing patterns, but in some years are represented by weak decreasing patterns ( Supplementary Fig. S1). DO shows an increasing pattern through events. The  Supplementary Fig. S2, and Supplementary Video S1). We call this DWR "DWR0312".
Chemical and physical characteristics of the DWR0312 event. The DWR0312 event started with a sharp decrease in both temperature and salinity, and a sharp increase in DO (Fig. 3b). Temperature dropped from 8.87 °C to 7.78 °C (−1.09 °C) and, salinity decreased from 30.91 to 30.80 psu (−0.10 psu). Conversely, DO sharply increased from 2.44 to 4.10 ml l −1 (+1.66 ml l −1 ). Both temperature and salinity remained depleted throughout the DWR reaching minima of 7.48 °C and 30.75 psu, respectively. DO remained high through most of the event reaching a maximum of 4.53 ml l −1 and this coincided with minimum salinity (Fig. 3b). The onset of the DWR0312 event corresponds to the flood tide, although it reaches the instrument array during the ebb tide (Figs. 3b and 4b). All the changes in chemical oceanographic properties during the DWR0312 event temporarily return to their ambient values during flood tidal currents for 5-7 hours each time (Figs. 3b and 4b). In the first three days, these trends are consistent while in the following seven days the chemical properties gradually reset back to ambient values (Fig. 3b).
The DWR0312 event started with an increase in turbidity from 1.58 to 4.82 NTU (+3.24 NTU) followed by two more significant increases coinciding with the end of strong flood tides (Fig. 4a). Turbidity reached its peak of 29.69 NTU following the second flood tide, and after which it returned to ambient values and remained low for the rest of the event.
Backscatter intensity and current velocity data, measured by ADCP provide insight into the hydrodynamics within the 300 m water column during the DWR0312 event (Fig. 4b,c). When the DWR0312 event initiated, a slight increase in particle intensity occurred (Fig. 4a,c) and northward directed flow velocity reached approximately 0.5 m s −1 (Fig. 4b). Two major and strong pulses appear in the following two days matching with increases in particle intensity at the end of strong flood tides (Fig. 4a-c). Following these two pulses bottom water particle intensity reduces and returns to ambient conditions. Video footage of the DWR0312 event. During the DWR0312 event, the camera recorded a clean image of the seafloor, including exposed pig bones and epifaunal shrimp (Figs. 4 and 5, Supplementary Fig. S2, and Supplementary Video S1). In order to describe the DWR0312 event, we analyzed all available video data between March 1 and 16, 2012. Five minutes before the event initiated, bottom waters showed no signs of increased suspended sediments (i.e., turbidity) and a large number of shrimp were present (Fig. 4d). Video recording was paused three minutes before the DWR0312 event so the onset of the DWR was not recorded. The next video started 8 minutes after initiation of the event at which time visibility was significantly reduced due to increased turbidity (Fig. 4e).
Two hours from the initiation of the DWR0312 event, the bones were still partially visible, and sediment began to accumulate on the bones (Supplementary Fig. S2a). Approximately 2.5 hours after initiation of the event, visibility in the water column improved in response to reduced turbidity ( Supplementary Fig. S2b), and epifaunal shrimp feeding on the pig carcasses remained during this stage (Supplementary Fig. S2b). Nine hours after www.nature.com/scientificreports www.nature.com/scientificreports/ initiation, the skull showed more accumulation of sediments (Fig. 4f). The water column remained relatively free of turbidity until approximately 12:40 pm on March 6, 2012. At this time the first of two major turbidity peaks were recorded and following which, visibility was reduced to zero ( Supplementary Fig. S2c). Visibility returned at approximately 8:00 pm, on March 6 ( Supplementary Fig. S2d) as turbidity in the water column subsided (Fig. 4a). Following the first major turbidity peak, more sediment had accumulated on the pig skull compared to the first slight turbidity increase ( Fig. 5f and Supplementary Fig. S2e).
The second and last turbidity peak occurred on March 7, 2012 at 12:55 pm (Fig. 4a), and in under 30 seconds visibility was reduced to zero ( Fig. 4g and Supplementary Video S1). Two hours after the second peak (at 3:57 pm) visibility returned and most bones were partially buried, the meshed bottom was covered, and all shrimp had left the area (Supplementary Fig. S2h). Shrimp returned to the area 3 hours later, but in low numbers. Eleven hours following initiation of the second turbidity peak, approximately 1.5 cm of sediment had accumulated on top of the www.nature.com/scientificreports www.nature.com/scientificreports/ skull and around the meshed-cage bottom (Fig. 4h,i). After 2 days the population of shrimp reached high levels, but did not reach the same density as before the event.
In order to compare sedimentation rates during DWRs, all available video footage from 2012 was analyzed and similar increased turbidity was recorded during other DWRs (Figs. [3][4][5]. Eight DWRs (7 in C1, 1 in C2) occurred after the DWR0312 event (March 5, 2012) and before the removal of the camera (August 1, 2012). Each DWR deposited sediments and contributed to the burial of the bones (Fig. 5a-h). Although it is not possible to precisely and directly measure the sedimentation rate in each event due to both the lack of vertical scale in the video footage and to organism disturbance (e.g., shrimp or octopus; Supplementary Fig. S2g) of bones, we approximated the thickness of the humerus bone (black arrow, Fig. 4d) as 6 cm using the bottom mesh size (1 cm) as a scale. The humerus bone was completely buried following three DWRs after the DWR0312 event, indicating that each DWR deposited approximately 1.5 cm of sediment.

Discussion
Regardless of the controlling factors, dense Pacific Ocean waters penetrate into the Juan de Fuca Strait, especially during upwelling periods, and get trapped by sills at the Boundary Pass before entering into the central SoG ( Fig. 1; 9 ). During spring tides, strong mixing occurs at the Boundary Pass preventing the dense waters from www.nature.com/scientificreports www.nature.com/scientificreports/ passing the sills, thus concentrating them at the bottom of the Haro Strait (Supplementary Fig. S3). The mixing probably causes resuspension of settled sediments at the sills increasing the turbidity of dense waters as well. During neap tides, weaker tidal currents cause less mixing and results in strong stratification of fresh surface waters and dense bottom waters (Supplementary Fig. S3). Following strong flood tidal currents, these dense waters are pushed into the central SoG changing physical and chemical characteristics of bottom waters for 3 days or longer (Supplementary Fig. S3; 7,9,13 ). chemical characteristics of DWRs. Of the three cycles of DWRs described in this study, C1 and C2 show the most pronounced and consistent chemical variations ( Fig. 2 and Supplementary Fig. S1), and DWRs in these two cycles (spring and fall) are identical to those described previously 7,9,13 . Conversely, DWRs in C3 are difficult to recognize 20 mainly because chemical signatures are weak or absent.
We present three hypotheses for the weak or absent signatures during C3. (1) Intrusion of Pacific Ocean water into the SoG is either very limited or not present during these cycles. (2) During C3, DWRs do not reach the central SoG largely because the flows are weak 21 . (3) The chemical characteristics of DWRs and SoG bottom waters are similar during C3, and hence, DWRs cannot be easily discerned. 1) The first hypothesis is plausible and may be explained by climate conditions. For example, seasonal changes in major wind directions may affect dense water intrusions causing a lack of DWRs at certain times of the year as is the case in some fjords (e.g., the Gullmar Fjord and Norwegian fjords 22,23 . 2) The second hypothesis can play a role, particularly when there is strong mixing causing weak water column stratification. 3) The third hypothesis seems the most probable as turbidity increases noticeably during C3 DWRs that coincide with neap tide cycles. The similarity in the timing and physical character of turbidity increases during C3 and those in C1 and C2 suggests that turbidity pulses mark the onset of weak DWRs during October and March, but where the chemical characteristics of DWRs are close to those of the central SoG. Regardless of the driving mechanism, the weak, intermittent DWRs during C3 suggests that the number of DWRs annually might be even higher than those documented in this study and this would increase the annual sedimentation rate estimated in this study.
physical DWR characteristics. DWRs initiate during neap tides and at or close to the end of strong flood tide currents. Chemical characteristics typically show rapid changes compared to ambient values at the start of these events while turbidity progressively increases over the course of them (Fig. 3). Although we do not have enough evidence to define the source of increased suspended sediments during DWRs two hypotheses may explain it. 1) Mixing of waters contributes to re-suspension of sediments at Boundary Pass, such that DWRs are already rich in suspended sediments before flowing into SoG. 2) Elevated suspended sediment concentration originates from re-suspension of previously deposited sediments after the DWRs pass Boundary Pass where downslope moving DWRs re-suspend seafloor sediments 21 . It is also possible that these two mechanisms both contribute to turbidity increases in DWRs. For example, mixing can form a high-fluid content sediment layer at the bottom of dense waters and these sediments can re-suspend as the dense waters start flowing in to SoG triggering gravity flows 24 .
The calculated sedimentation rate during each DWR in 2012 was 1.5 cm. Given that 6-15 DWRs occur every year ( Supplementary Fig. S1), the annual sedimentation rate can, theoretically, exceed 9 cm yr −1 and may reach up to 22.5 cm yr −1 . Over 11 years, these values would suggest that DWRs deposited approximately 2 m of sediment in the SoG. This value is significantly higher than the reported maximum sedimentation rate in the Fraser prodelta (~3 cm yr-1 25 ) suggesting that either sedimentation rates are significantly higher along the pathway of the DWRs (base of the SoG) or that some DWRs and bottom currents contribute to seafloor erosion and reduce the long-term sedimentation rate. Further sampling and monitoring is needed to accurately assess long-term sedimentation rates at the base of the SoG.
In summary, we demonstrate episodic long-lived fluctuations in the physical and chemical ocean characteristics of the Strait of Georgia, Canada showing strong evidence of deep water renewal events (DWRs). Several factors control the timing and character of the DWRs including Pacific Ocean upwelling, Fraser River discharge, El-Nino cycles, and tidal currents 9,13,26 . We document at least six DWRs occur every year refreshing bottom waters in the Strait of Georgia. DWRs are divided into three cycles (C1-C3). C1 and C2 are characterized by pronounced fluctuations in seawater characteristics and are consistent with previously defined DWRs. C3 events display weak fluctuations in chemical parameters and are documented here for the first time. Regardless of the cycle, DWRs initiate during neap tides and are associated with significant increases in turbidity.
For the first time in the literature, DWRs are documented using video footage. Video footage of DWRs in 2012 provides insight into how these events transport and deposit sediments. Available data suggest 1.5 cm of sediment is deposited during each DWR in C1 and C2. With a sedimentation rate of potentially 9 cm yr −1 , DWRs should be considered an important sediment transport mechanism in strait and enclosed seaways (e.g., fjords).

Methods
Data collection. We utilized an extensive dataset of real-time physical and chemical ocean measurements and seafloor video footage. Oceanographic data is derived from instruments situated at 300 m water depth (Fig. 1). All instruments are situated between 1 and 2 m above seafloor. The instrument station is operated and maintained by Ocean Networks Canada (ONC) who supports a network of recording stations on the seafloor on the west coast of Canada. Oceanographic data used in this study include salinity, dissolved oxygen (DO), temperature, turbidity, and acoustic doppler current profiler (ADCP) measurements. Supplementary Table S2 displays the instrument models, sensors and sampling intervals. More detailed information can be found in ONC webpage (https://www.oceannetworks.ca/). Average, minimum and maximum values for each oceanographic property are calculated from time-averaged data.