Observations of coastal ocean currents in the Barents Sea (Porsangerfjord) during the summers of 2014 and 2015

Accepted Manuscript Observations of coastal ocean currents in the Barents Sea (Porsangerfjord) during the summers of 2014 and 2015 Malgorzata Stramska, Knut Yngve Børsheim, Andrzej Jankowski, Henrik Søiland, Agata Cieszyńska PII:

S0272-7714(16)30793-4

DOI:

10.1016/j.ecss.2018.02.035

Reference:

YECSS 5777

To appear in:

Estuarine, Coastal and Shelf Science

Received Date: 28 December 2016 Revised Date:

18 January 2018

Accepted Date: 27 February 2018

Please cite this article as: Stramska, M., Børsheim, K.Y., Jankowski, A., Søiland, H., Cieszyńska, A., Observations of coastal ocean currents in the Barents Sea (Porsangerfjord) during the summers of 2014 and 2015, Estuarine, Coastal and Shelf Science (2018), doi: 10.1016/j.ecss.2018.02.035. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Observations of coastal ocean currents in the Barents Sea (Porsangerfjord) during the summers of 2014 and 2015

RI PT

Malgorzata Stramska a, b, *, Knut Yngve Børsheimc, Andrzej Jankowskia, Henrik Søilandc, Agata Cieszyńskab, a a

Institute of Oceanology, Polish Academy of Sciences, Powstańców Warszawy 55,

b

SC

Sopot 81-712, Poland

Department of Earth Sciences, Szczecin University, Mickiewicza 16, Szczecin 70-383,

Poland

Institute of Marine Research Nordnesgaten 33, NO-5017 Bergen, Norway

M AN U

c

* corresponding author: Phone: (+48 58) 73 11 600, Fax: (+48 58) 55 12 130, email: [email protected]

Abstract

TE D

Keywords: Arctic; fjords; currents; high frequency radars.

Surface current data collected with a high frequency radar system in the summers of 2014 and 2015 in the coastal waters of the Barents Sea (Porsangerfjord) have been analyzed.

EP

Application of a least-squares fit coupled with nodal modulation enabled separation of tidal and residual currents. The most important tidal component was M2 (semidiurnal

AC C

lunar, period of 12.42 hours) and the second largest one was S2 (principal solar semidiurnal, period of 12 hours). Residual currents were significantly influenced by winds. The relationship between winds and currents is complicated, since wind speed, direction, and fetch are highly variable. The data reveal a significant contribution from inertial currents to residuals. Forcing of surface currents by winds resulted in more frequent measurements of stronger residual currents in situations when wind speeds were higher. Stronger winds and residual currents were most often associated with winds blowing from west to east. The most frequent residual direction was to the right by 10-30 degrees in comparison to the wind azimuth. Freshwater runoff to the fjord influences the 1

ACCEPTED MANUSCRIPT

vertical distribution of water salinity. Due to runoff, surface water transport is on average directed out of the fjord, but its variability on synoptic time scales is governed by winds.

AC C

EP

TE D

M AN U

SC

RI PT

Research Highlights: Surface currents in the Porsangerfjord were analyzed. Surface currents are forced by tides (10-30% of variance) with the M2 and S2 components being the most important. Surface currents are forced by winds. Stronger winds and residual currents were most often associated with winds blowing from west to east. Surface water transport is on average directed out of the fjord as a result of freshwater runoff, but its variability on synoptic time scales is governed by winds.

2

ACCEPTED MANUSCRIPT

1. Introduction Studies of the Arctic often focus on highly climate-sensitive regions. One such region is the Barents Sea (BS), where the sea surface temperature (SST) has increased by an average of 0.03°C per year in the last 32 years, this increase being simultaneous with a

RI PT

significant decrease in sea ice coverage (Jakowczyk and Stramska, 2014). The warming trend estimated in the BS is exceeding global trends (IPCC, 2007). It is natural to expect that climate-related changes are also affecting coastal areas of the Barents Sea. This study focuses on the Porsangerfjord (Porsangerfjorden), near the North Cape (Nordkapp), one

SC

of the largest fjords in northern Norway. Significant ecosystem changes have been observed in recent years in this fjord. In the 1980s the ecosystem there experienced

M AN U

strong pressure as a result of fish stocks being devastated by harp seals migrating from the Barents Sea (Broderstad and Eythórsson, 2014). In addition, coastal cod gradually disappeared from its usual spawning sites, possibly because of changes in water circulation (Broderstad and Eythórsson, 2014). Other ecological transformations include the depletion of kelp forests (Sivertsen and Bjørge, 2015). Finally, the migration to the Barents Sea and Porsangerfjord of the red king crab (a marine species alien to this region)

TE D

has given rise to another potential threat to the ecosystem (Hvingel et al., 2012; Windsland et al., 2014). The mechanisms behind all these changes have yet to be fully explained, but it is likely that they were forced by adjustments in the physical conditions within the fjord (Sivertsen and Bjørge, 2015). Coastal settlements in the area, in

EP

particular the Sámi villages, are strongly dependent on fishing for their economic and cultural wellbeing, and the ecological changes in recent years have led to substantial social changes in this region as a consequence (Broderstad and Eythórsson, 2014).

AC C

The Porsangerfjord extends from about 25.0 to 26.5oE and from 70.0 to 71.0oN

(Figure 1). It is approximately 100 km long, 15-20 km wide, and has a maximum depth of more than 260 m. Based on its bottom topography (Figure 2a), this fjord can be divided into three different zones: inner (0-30 km), middle (30-70 km) and outer (70-100 km). The inner zone is separated from the rest of the fjord by an underwater sill at 60 m depth situated approximately 30 km from the inland end of the fjord. The middle part of the fjord starts from this 60 m sill and extends as far as the island of Tamsøya, some 70 km from the head of the fjord (Figure 2b). The outer zone stretches from Tamsøya

3

ACCEPTED MANUSCRIPT

towards the Barents Sea and ends at a sill 180 m below the surface. The middle and outer parts of the fjord are thus well-connected with the Barents Sea. This stands in contrast with the inner part of the fjord, which has limited exchange with the open sea: a unique arctic ecosystem has formed in this inner zone (Eilertsen and Frantzen, 2007). Cushman-

RI PT

Roisin et al. (1994) estimated that the width of the Porsangerfjord is approximately three times the deformation radius and suggested that it could be classified as a fjord of intermediate size. Based on the ratio of land runoff to the fjord’s surface area, Svendsen (1995) classified Porsanger as a fjord with a relatively low runoff, but pointed out that in

SC

the spring/summer season freshwater runoff could significantly influence the hydrography. Porsangerfjord waters are stratified from May to October as a consequence

M AN U

of seasonal river runoff and surface heating (Svendsen, 1995). The fjord is surrounded by mountains that reach altitudes of about one thousand meters on the south- western side of the fjord (the inner part of the fjord, Figure 2b). The land around the outer part of the fjord is generally below 300 meters. The mountains disrupt local wind patterns, especially in the inner part of the fjord. Hence, wind directions and speeds above the fjord water surface may differ significantly from the large-scale patterns.

TE D

The coastal area off Finnmark in northern Norway, where the Porsangerfjord is situated, is a part of the Norwegian Shelf (Figure 1b). Near the fjord’s entrance, the bathymetry of the coastal area is characterized by a steep shelf break falling off to 250 m, followed by a relatively flat bottom at depths of 250-300 m. There are two seamounts in

EP

this region: one (at 45 m depth) is situated between the Porsangerfjord and Laksefjord, near the coastline, while the other (at 250 m depth) is farther out to sea. Two major

AC C

current systems dominate the waters near the Norwegian coast. These are the Norwegian Atlantic Current (NAC) with Atlantic Water (AW) and the Norwegian Coastal Current (NCC) with Norwegian Coastal Water (NCW). Off Tromsø the NAC splits into two branches: one flows northwards as the West Spitsbergen Current, and the other – the North Cape Current – flows along the Norwegian coast, parallel to the NCC (flowing closer inshore). Current directions and speeds (15-40 cm s-1) vary in time and with depth (Skagseth, 2011). Surface currents are sensitive to winds, which are mostly NE –NW during the summer and SW during the winter (Svendsen, 1995). Two different water masses (AW and NCW) interact near the entrance to the Porsangerfjord. Pedersen et al.

4

ACCEPTED MANUSCRIPT

(2005) have shown a number of mesoscale eddies with spatial scales of 15-40 km propagating along the coast of northern Norway. Despite the diversity of studies performed in the Porsangerfjord region (listed above), many questions remain unanswered regarding the variability of water properties

RI PT

(temperature, salinity), currents and interactions with the area outside the fjord. It is important to extend knowledge of these environmental conditions if a better understanding of the ongoing ecosystem changes in the Porsangerfjord is to be achieved. The main objective of this study was to obtain detailed information about surface currents

SC

in the fjord’s outer zone during the summers of 2014 and 2015. Surface currents affect the transfer of momentum, heat and mass across the air-water interface, consequently

M AN U

influencing physical, chemical and biological processes in the upper ocean. The prevailing surface circulation patterns will be described in this paper and the influence of winds evaluated. This information will be of use to scientists studying the dispersal of fish eggs, parasites or other drifting particles (e.g. Asplin et al., 2014, Johnsen et al., 2016; Myksvoll et al., 2011; 2013); it should also be of interest to local fishermen. In addition, information regarding the advection of surface water masses is necessary for the

TE D

correct interpretation of ocean color data in coastal regions (Arnone et al., 2016); the authors are currently working on such problems. Another objective was to obtain quantitative estimates of surface water transport outside the fjord. This paper is a contribution to the interdisciplinary NORDFLUX project

EP

(Application of in situ observations, high frequency radars and ocean color to the study of suspended matter, particulate carbon and dissolved organic carbon fluxes in coastal

AC C

waters of the Barents Sea). A preliminary discussion of surface currents, based on data from 2014 and focusing on details of tidal analysis, was presented in Stramska et al. (2016). In contrast, the present paper uses longer data series (from 2014 and 2015) and concentrates more on the detailed description of the variability of residual currents and their relation to winds. Other project relevant papers published so far include a detailed description of meteorological variability and climate trends in the Porsangerfjord region (Cieszynska and Stramska, 2017) and an analysis of the spatial and temporal variability of total suspended particulate matter based on in situ bio-optical experiments (Bialogrodzka et al., 2017).

5

ACCEPTED MANUSCRIPT

2. Methods 2.1 Surface currents from High Frequency radars Observations of surface currents were carried out with a high frequency (HF)

RI PT

WERA radar system (e.g., Helzel et al., 2010, 2011) from June 10 to October 11, 2014 (year days 161- 284) and from May 28 to August 17, 2015 (days 148-229). The region monitored by the radars is situated in the outer zone of the Porsangerfjord (Figure 2a). WERA systems are very useful for deriving spatial information about surface currents in

SC

coastal regions and have demonstrated good accuracy and reliability at numerous installation sites around the world (e.g., Helzel et al., 2010, 2011; Kokkini et al., 2014).

M AN U

The system and the methods used to analyze WERA data were described in greater detail in our previous paper, where we presented the results of a tidal analysis of sea level and surface currents recorded in 2014 (Stramska et al., 2016). Only the most important information is provided here for completeness.

The HF system in the Porsangerfjord included two sites positioned at about 27 km from each other. The first site was at Repvaag (70.76 N, 25.69 E), the second at

TE D

Nordvaagen (70.97 N, 26.02 E), as indicated in Figure 2a by the letters A and B, respectively. The radars were applied in phased-array mode, and each station was equipped with 4 transmitting and 12 receiving antennas. In our experiment, the frequencies were set to about 26 MHz and the nominal bandwidth was 250 kHz. The

EP

radar-measured ocean currents thus represent a surface water depth of approximately 0.5 m. Additional information on subsurface water currents was obtained from an ADCP

AC C

current meter, as described below. The two-dimensional surface current velocity vectors were estimated by

combining the radial velocity components obtained from the two radar stations in the area of signal overlap. This was done using software provided by the manufacturer (Helzel Messtechnik GmbH). The data processing steps are explained in detail in the literature (e.g., Barth et al., 2010; Stanev et al., 2014). The spatial resolution of the radar data after vector field construction by the WERA software is about 0.75 km (see Figure 3). In this paper, the terms “cell” or “pixel” are used interchangeably to indicate the position of the time series discussed in the Results section. Surface current maps were generated every

6

ACCEPTED MANUSCRIPT

half hour. In addition, the final HF radar data set included estimates of the accuracy for each value of the current vector components (e.g., Stanev et al., 2014), accounting for the geometry of the radar sites (geometric dilution) and random errors. The accuracy of 99% of the recorded u and v components was better than 0.04 ms-1 and 0.035 ms-1,

RI PT

respectively, at the cells discussed in this paper. The HF radar data return is sensitive to weather conditions, ionospheric reflection, sea-surface conductivity as well as noise from interference, which leads to the problem of missing data. Around 90% or more data were recorded in the central region, where the cells discussed in this paper were located (see

SC

Stramska et al., 2016). The positions of the selected pixels referred to in the Results sections are indicated in Figure 2a by the symbols A1, A2, A3 and M.

M AN U

In the present work the same methods were applied as in Stramska et al. (2016) in order to separate the tidal from the nontidal (or residual) currents. The method is based on a least-squares fit coupled with nodal modulation (Foreman 1977, 1978). Our previous paper was based solely on data from 2014, whereas the present one uses a more extensive data set (from 2014 and 2015) and includes a more detailed discussion of the variability of residual currents.

TE D

Specific time intervals were defined in order to examine the temporal changes in surface current statistics. Time interval #1 in 2014 and 2015 corresponds to the dates when the in situ bio-optical experiments were carried out. Thus, interval #1 (also referred to as late spring) in 2014 is from June 10 to June 29 (days 161-180), while interval #1 in

EP

2015 is from May 28 to June 18 (days 148-170). Note that the exact dates are somewhat different in each case, because the dates of the bio-optical experiments were different.

AC C

(The bio-optical data are discussed in a separate paper – Bialogrodzka at al., 2017). Time interval #2 (referred to as summer) includes the same dates in 2014 and 2015, from July 1 to August 17 (days 182 - 229). Time interval #3 (referred to as early fall), from August 18 to October 11 (days 230 - 284), is defined only for 2014. In 2015 the HF radar system was dismantled on August 17, after which no further data were collected. Using this approach, one can check whether there are significant differences in the current statistics depending on the time of the year (spring, summer, early fall) and between the two years in question. Unfortunately, these data sets do not allow variability in surface currents to be investigated over the entire annual cycle or longer time intervals.

7

ACCEPTED MANUSCRIPT

Spectral analysis was carried out in addition to tidal harmonic analysis. Autospectra of physical variables were determined using standard signal processing techniques (Bendat and Piersol 2010; Bloomfield 2000). A Fourier transform was applied to the autocovariance and cross-covariance functions in order to obtain power spectra and

RI PT

squared coherence of scalar variables. The time series record consisted of a total of N = 2048 data points, representing around 85 days. The autocovariance functions were calculated with a maximum time lag of M = 220. The standard error in these power spectral estimates was 34%. A squared coherence greater than 0.28 was statistically

SC

significant at the 95% confidence level. In addition, rotary spectra (Gonella, 1972) were estimated for vector time series (surface currents and residuals). In this case, the

M AN U

individual time series were divided into subsamples consisting of 512 data points, and power spectra were averaged. Prior to these calculations, all data gaps were linearly interpolated, and series with gaps longer than four hours were discarded. The HF data were also used to estimate the net transport of surface waters out of the Porsangerfjord in the direction of its exit. This was based on data recorded at seven cells located on a transect defined by stations A1, A2, and A3 (Figure 1): cell #2

TE D

corresponds to station A1 and cell #6 to station A3. Current vectors were decomposed into components parallel and perpendicular to the transect line. Time series of components perpendicular to the transect segments were used to estimate the daily average water transport at each of the sectors delimited by pixels 1 through 7. The total

EP

net daily transport of water was estimated as the sum of these estimates. These calculations are based on the assumption that the velocity of the 1 m surface layer is

AC C

approximately the same as the surface velocity recorded by radars.

2.2 Subsurface currents from ADCP Subsurface current data series were collected using a Nortek Continental 190 kHz

ADCP (www.nortek.no) equipped with compass, tilt, pressure, temperature sensors and internal recorder. The instrument was deployed from a mooring located at 70° 52.00’ N, 26° 10.91’ E for 17 days (June 8 – 24, 2014). The bottom depth at this site was 239 m. The ADCP was mounted in an upward-looking configuration, and was applied to profile currents in 5 m vertical bins from the instrument position (~ 140 m depth) to the surface.

8

ACCEPTED MANUSCRIPT

The center points of all the good bins were located between 135.5 and 15.5 m. The accuracy of a 5-minute sample with the ADCP setup used was 2.1 cm s -1. The normal error was further reduced by filtering (averaging): e.g., the error in an hour average based on 12 samples was reduced to 0.6 cm s -1. Time series of ADCP data were collected over

RI PT

a time interval too short to apply the method of Foreman (1977, 1978). Therefore, in order to carry out comparisons between the surface and subsurface currents at pixel M, both data sets (ADCP and WERA data for pixel M) were additionally filtered using a 13hour moving average filter. After being filtered, the data were subsampled using a 1-hour

SC

time resolution and compared.

M AN U

2.3 Other data sets

During in situ experiments from June 6-29, 2014 and May 29-June 18, 2015, vertical profiles of water temperature (T) and salinity (S) were also gathered from a small boat. This was done with the aid of an SBE 49 FastCAT (Seabird Electronics) CTD Sensor. The depth range for the profiles was usually 0-50 m since it was difficult to deploy the instrument packages from a small boat, but profiles occasionally extended

TE D

down to about 100 m. Data were collected at several stations located along the fjord. More information about data from these boat surveys is given in another paper (Bialogrodzka et al., 2017). In the present one, these data are used to show the influence of freshwater runoff on the spatial distribution of salinity.

Norwegian

EP

Other data sets are also used in this paper. Sea level records were provided by the Hydrographic

Service

(http://kartverket.no

and

http://vannstand.no).

AC C

Estimates of terrigenous water discharge into the fjord were based on data from E-HYPE model, version 3.11 (Donnelly et al., 2015). Data were downloaded from the Swedish Meteorological

and

Hydrographical

Institute

server

(http://hypeweb.smhi.se/europehype/time-series/). Wind data from the Norwegian Meteorological Institute (http://www.yr.no) included wind speed and direction for Honningsvaag and Lakselv. Honningsvaag is situated near the Nordvaagen radar station, while Lakselv lies in the inner part of the fjord (Figure 2). A Vaisala WAV252 Heated Wind Vane and a Vaisala WAA252 Heated Anemometer were used at Lakselv, whereas a Vaisala WAV151 Wind Vane and a Vaisala WAA151 Anemometer were used for wind

9

ACCEPTED MANUSCRIPT

measurements at Honningsvaag. In order to obtain wind direction information consistent with surface current data, wind direction was converted to wind azimuth according to oceanographic convention: here, a wind azimuth of 0 degrees indicates a wind blowing from south to north, 90 degrees represents a wind blowing from west to east, -90 degrees

to south. The same convention is used for the current directions.

3. Results and Discussion

SC

3.1 Tidal and residual currents

RI PT

a wind blowing from east to west, and 180 (or -180) degrees a wind blowing from north

Examples of surface current maps obtained from the HF radars are shown in

M AN U

Figure 3 a and b. These maps were generated on the same day (June 27, 2015), but 4 hours apart. It is evident that the spatial pattern of surface currents changed significantly during those 4 hours. More examples of recorded surface currents, presented as maps and time series, can be found in Stramska et al. (2016).

Wind speed and azimuth time series recorded in time interval #1 in 2015 are exemplified in Figure 4 a and b. These plots show that wind speed and direction in the

TE D

Porsangerfjord are highly variable: there are large fluctuations (of ~ 2 m/s and more) in wind speed at a time scale from hours to days. During the time interval shown in Figure 4, there were 4 events when wind speeds reached 10 m/s. The prevailing wind direction was W-NW (azimuth between 50 and 120 degrees). Figures 4 c-f show plots of surface

EP

currents and residuals (speed and direction) for pixel A3 for the same time interval as the wind data shown in Figure 4a-b. Pixel A3 is located on the eastern side of the fjord

AC C

(Figure 2a). Note that the current speed is almost always less than 0.6 ms-1. Current speed varies at a time scale of hours: this is expected, because of the influence of tides (Stramska et al., 2016). Residual current speed is also quite variable and is generally less than 0.4 ms-1. Patterns of variability in current speed and direction are different at different pixels (data not shown). Tidal analysis of surface current data from 2014 and 2015 separated the most important tidal constituents in the Porsangerfjord. The present results confirmed earlier findings based on a shorter data set (Stramska et al., 2016) that M2 (semidiurnal lunar constituent with a period of 12.42 hours) had the largest amplitude. The second largest

10

ACCEPTED MANUSCRIPT

tidal constituent was S2 (the principal solar semidiurnal component with a period of 12 hours). Significant contributions from the N2, L2 and K1 components were also noted. The 36 most important tidal constituents were used for reconstructing the residual

RI PT

currents shown in Figure 4.

3.2 Winds and residual currents

From comparisons of total, reconstructed/tidal and residual currents, it was estimated that tidal components contribute about 10-30% to the total variance of the

SC

surface currents recorded. The remaining variance is attributed to the residual currents that are a superposition of all flows not associated with tides, such as geostrophic currents

M AN U

or currents forced by changing meteorological conditions. For instance, inertial ocean currents can play a significant role, since winds are variable in this region: events with relatively high winds are often followed by periods of calmer weather. In their simplest form, inertial oscillations are circular motions rotating clockwise in the northern hemisphere. They occur intermittently in response to individual wind events, decaying rapidly within a few cycles. The period of inertial oscillations depends on the latitude and

TE D

is equal to 12.67 hours at pixels A1, A2 and A3, which is similar to the semidiurnal N2 tidal constituent with a period of 12.66 hours. In order to investigate in more detail the influence of winds on surface currents, power spectra were estimated, and the simple statistical analysis presented in the following paragraphs was undertaken.

EP

The rotary spectra of total and residual currents for pixel A1 are shown in Figures 5 a-d. The main features of such spectra estimated at other locations were similar (not

AC C

shown). The most prominent peak for surface currents is for the semidiurnal frequency. It is not possible by means of spectral analysis to resolve the semidiurnal tidal components that contribute to this peak, since the temporal resolution of the present data is too low. Figure 5 also shows the diurnal peak, which is smaller than the semidiurnal one. Comparison of Figures 5 a-b with Figures 5 c-d shows that that the diurnal peak has been removed by harmonic analysis, but that the semidiurnal peak is still present in the residual time series, although it is significantly smaller than in the original data and mostly restricted to CW rotation. Since the removal of tidal components did not eliminate

11

ACCEPTED MANUSCRIPT

the semi-diurnal peak from the clockwise rotating current velocity component, one can conclude that inertial currents made a significant contribution to the residuals. The dependence of surface currents and sea level on winds is also illustrated in Figures 5e-h. Figures 5e and g compare the power spectral densities for sea level and

RI PT

wind speed recorded at Honningsvaag, with power spectral densities for current speed measured at station A1. In each case, the power spectrum has been normalized to the total variance of the respective time series. Figure 5e shows that a relatively larger portion of the total variance is attributed to the diel and semidiurnal variability of sea level than in

SC

the surface current speed time series. The corresponding maxima in the power spectrum for surface current speed are much less pronounced than in the power spectrum for sea

M AN U

level. The power spectra for wind and current speed indicate that proportionally more variance is attributed to time scales longer than a day in comparison with the sea level data. This is in agreement with previous findings that as much as 99% of variance in sea level but only 10-30% of variance in surface currents are attributable to tides (Stramska et al., 2016). Figures 5f and h summarize the results of cross-spectral analysis between selected pairs of variables. Figure 5f indicates the squared coherence between the sea

TE D

level and the current speed, and Figure 5h indicates the squared coherence between the wind and the current speed. A coherence greater than 0.28 is statistically significant at the 95% level (Bloomfield 2000; Bendat and Piersol 2011). In an ideal linear system, the coherence would be unity (one – 1). It is clear that the coherence between the wind speed

EP

and the current speed is above the critical level for periods longer than about 2.5 days. This indicates that both variables are significantly correlated at synoptic time scales. The

AC C

time delay (not shown) between the wind and currents is about 2 hours, which shows that surface currents respond rather quickly to winds. Similar findings have been published for other coastal regions (for example, see Cosoli et al., 2008, 2012). In contrast, the sea level and current speed are not significantly correlated at time scales longer than ~3.5 days, although there is a significant correlation at a time scale of ~2–3 days, probably due to atmospheric pressure variability (see also Stramska et al., 2016). These results support the notion that the temporal and spatial variabilities of the wind’s influence add to the complexity of the surface currents in the Porsangerfjord. Strong residual surface flows

12

ACCEPTED MANUSCRIPT

driven by wind stress are a very important contribution to the observed variability of surface currents. The statistical characteristics of the residual currents and winds will now be examined in greater detail, using the full data set. The aim here is to find out what wind

RI PT

directions are the most frequent, as well as what directions can be associated with high frequencies of occurrence of stronger winds and stronger residual currents. To this end, the frequency of occurrence of different wind directions has been plotted in Figure 6 and frequency of occurrence of different wind speed estimated for each wind direction. For

SC

this analysis all data have been pooled (from both 2014 and 2015). As can be seen in Figure 6a and b, winds blowing to the south (from the north) are the most frequent. Fairly

M AN U

common are also winds blowing to the east (from the west). Winds blowing westwards are the least common in the data sets. Winds blowing to the east (from the west) are associated with the highest occurrence of strong winds (see also Table 1). Figure 7 displays the frequency of occurrence of residual current speed and direction for a given interval of wind direction for pixels A3 (Figures 7 a-h) and A1 (Figures 7 i-p). Note that for wind directions with a higher occurrence of stronger winds (winds blowing to the east

TE D

from the west), higher residual current speeds were measured more often. The most obvious example is for a wind azimuth of 60 to 90 degrees. Table 1 summarizes the frequency of occurrence of residual currents > 0.25 ms-1 and > 0.35 ms-1 for wind directions recorded at least 4% of the time or more (when there was enough data for

EP

statistical analysis). Obviously, wind fetch also plays an important role: for example, for wind azimuths of -120 to -150 degrees and -90 to -120 degrees, current residuals >0.25

AC C

ms-1 and > 0.35 ms-1 were recorded more often at pixel A1 than at pixel A3, whereas the reverse applies to the wind azimuth from 0 to 60 degrees. For winds with the same azimuth the frequency of stronger residual currents was greater when higher wind speeds were measured more often. Table 2 shows that this is the case for wind azimuths 60-90 and 90-120 degrees. Note that the most frequent current residual directions are shifted to the right in comparison to the wind directions, as expected in the Northern hemisphere. However, this shift is smaller (about 10-30 degrees) than it would be according to the classical Ekman model (45 degrees, Ekman, 1905), and there is some spread of data in the histograms. This can be explained by the fact that the Porsangerfjord is influenced by

13

ACCEPTED MANUSCRIPT

very variable winds and that surface currents often do not reach stationary conditions. Moreover, ocean surface currents include Stokes drift (e.g., Wu, 1983; Polton et al., 2005; Tamura et al., 2012, McWilliams et al. 2012). Finally, winds at this study site vary spatially, as they are influenced by the proximity of the fjord’s boundaries and land

RI PT

topography. Different wind fetches, changing with wind direction and pixel position, contribute to the overall variability of the relationship between wind and currents. Therefore, one would not have expected the relationship between wind and currents to conform to the basic features of the Ekman model in the present study location, even if

SC

the area covered by HF radars had been situated in the outer part of the fjord, where the waters are not as sheltered by land topography as in the inner part of the fjord. It is likely

M AN U

that the influence of the coast is reflected in the bimodal distribution of residual currents (see Figure 7j). Note, however, that the number of observations for the wind azimuth of 0 to -30 degrees was relatively low when this bimodal distribution was observed, so these results need to be treated with caution. Finally, it should be borne in mind that residuals that include contributions from inertial and drift currents are being compared with winds recorded in Honningsvaag, and that wind speed and directions are somewhat variable

TE D

over the area of the fjord (Cieszynska and Stramska, 2017). Summarizing, stronger wind events were recorded when winds were blowing from west to east (azimuth 60 to 120 degrees) and residual current speeds of >0.25 ms-1 and > 0.35 ms-1 were more frequent with these wind directions than with others.

EP

In order to investigate more closely the temporal patterns in winds and currents, the frequency of occurrence of their speed and direction were estimated at different time

AC C

intervals: late spring, summer and early fall. This was partly motivated by the fact that there is a well-defined annual cycle of wind speed and direction if one looks into multiyear wind records (Svendsen et al., 1994; Cieszynska and Stramska, 2017). The annual cycle involves a seasonal shift in wind directions. Winds are directed from south to north in winter, whereas in summer approximately opposite directions prevail. The month of May is a transition period with more variable wind directions. In addition, summer conditions are characterized by lower average wind speeds and lower standard deviations than in winter conditions (Cieszynska and Stramska, 2017). Although the present surface current measurements did not cover the entire calendar year, it is

14

ACCEPTED MANUSCRIPT

worthwhile checking for any temporal changes in current characteristics corresponding to seasonal changes of winds from spring to early fall. The results of this statistical analysis of winds during given time intervals are presented as the cumulative frequency of wind speed, and frequency of occurrence (in %) of a given wind speed and wind azimuth in

RI PT

Figure 8, and as wind roses in Figure 9. The frequency distributions of the wind azimuth have different patterns in each time interval, and there is a significant difference between the same time intervals in different years. Figure 8 shows that stronger winds were more frequent in late spring (time interval #1 in 2014 and 2015) than in interval #2 (summer)

SC

of 2014 and 2015 (see also Table 3). Time interval #3 (early fall) in 2014 also had a larger percentage of stronger winds than time interval #2. Moreover, stronger winds (>

M AN U

7.5 m/s) were recorded more frequently during time interval #1 of 2015 than during any other time period. The wind azimuth between -150 and 150 prevailed in time period #1 of 2014 and in time period #2 of 2015. These differences in wind patterns during given time intervals are consistent with the more detailed analysis of the annual cycle in winds based on multiyear meteorological data sets presented by Cieszynska and Stramska (2017). The cumulative frequency and frequency of occurrence (in %) of residual surface

TE D

current speed, as well as the frequency of occurrence of different residual current directions for each of the time periods discussed in this paper are shown in Figures 8 g-l. The most striking observation is that there are more events with stronger residual current speeds during time interval #1 in 2015 than in any other time interval; this is during the

EP

same time period when the largest percentage of stronger winds was recorded (Table 3). In addition, current residuals with speeds > 0.25 ms-1 in this time interval were more

AC C

frequent at point A3 than at A1. This is probably to be expected, as the wind fetch at pixel A3 is longer for winds blowing from westerly directions. Note also that there is a shift in the most frequent residual directions between pixels A1 and A3. This may be due to the influence of the coastline, land topography and spatial variability of the wind.

3.3 Comparison of surface and subsurface currents Figure 10 compares surface currents recorded by the HF radar system and an ADCP current meter at pixel M in 2014. For this comparison both data sets were filtered with a 13-hour moving average filter in order to remove tide-related variability. For

15

ACCEPTED MANUSCRIPT

greater clarity, Figure 10 illustrates only data from selected depths. The results show that the current speed is larger in the surface layer (surface WERA data and 15-m depth ADCP data) than at greater depths. At 20 m and below, the current speed is significantly smaller than at the surface, and the variability patterns are well synchronized. The

RI PT

directions of the surface and 15-m currents are similar, and generally conform to the wind azimuth (Figure 10 b). Starting at 20 m, however, the temporal pattern in the current direction is quite different from that at shallower depths (often the reverse), and it is very similar at all greater depths. From these data, one can infer that wind influenced currents

SC

in a surface water layer at least 15 m deep, whereas a different forcing (pressure gradient) was dominant at depths of 20 m and below. These results need to be treated with caution,

M AN U

since ADCP time series are short and limited to only one location.

3.4 Surface water transport across transect A

Continuous HF radar observations of surface currents in 2014 and 2015 allowed to investigate the transport of surface water across transect A. The estimate is based on decomposing surface currents into components that are parallel and perpendicular to the

TE D

transect line. Current components perpendicular to the transect line at 7 pixels have been used to derive the daily average water transport across different sections of transect A as well as the total transport across the entire transect (positive out of the fjord). The results are displayed in Figure 11 a and b for the 2014 and 2015 time series, respectively. The

EP

estimates shown in Figure 11 a and b are based on the assumption that the water layer is 1 m deep. This assumption leads to an underestimation of the total surface water transport,

AC C

but since no information is available about currents with sufficient depth resolution, it is difficult to provide more precise information about the transport below. Assuming a constant eddy viscosity of 0.05 m2/sec, the Ekman layer depth at 70 oN can be estimated at ca 26 m, but in our coastal location we cannot really expect the Ekman model to apply (e.g., Csanady, 1982). Considering that at 15 m the current direction was similar to that at the surface, the estimate based on the 1-m layer should at a rough guess be multiplied by 5-10 in order to obtain a bulk estimate of the volume of water exported and originating from the surface mixed layer, which was about 15-20 m deep (Bialogrodzka et al., 2017).

16

ACCEPTED MANUSCRIPT

Figures 11 a and b show that the export of surface water from the fjord is often more efficient on its south-eastern side. This may be due to the amplification of currents related to the boundary condition on the coast and is consistent with Cushman-Roisin et al. (1994), who estimated that the Porsangerfjord is approximately three times as wide as

RI PT

the deformation radius. Accordingly, the circulation in this fjord is significantly affected by the Coriolis force. The influence of this force on currents in broad Norwegian fjords was also investigated by Wassmann et al. (1996). In addition, there are events when, during the day, more surface water is transported into than out of the fjord. For

SC

comparison, Figure 10 c shows time series of the south-north component of the wind velocity (VH, positive in the north direction) measured in 2014 and 2015 at

M AN U

Honningsvaag. It seems that events of greater water transport out of (or into) the fjord correspond to positive (or negative) values of VH, respectively. In particular, there were no significant water export events on days 190-205 in 2015, when VH had relatively low values.

To assess the relationship between VH and water transport, a scatter plot showing the values of the surface current component perpendicular to the transect line (ur) at point

TE D

A1 plotted as a function of VH is presented in Figure 11e. There is a significant correlation between the two quantities. Figure 11f shows a similar scatter plot, but for the total transport (across transect A) plotted as a function of VH. Again, there is a significant correlation between the two variables. Note that even if there are significant differences

EP

in meteorological conditions, including wind speeds and directions in different parts of the fjord (Cieszynska and Stramska, 2017), the time series of winds recorded at Lakselv

AC C

and Honningsvaag indicate that there is a very good correlation between the V component of the wind velocity at both stations (Figure 11d), despite the fact that they are located at opposite ends of the fjord. Finally, a comparison of the estimated water transport across transect A with

terrigenous water runoff would be informative. Estimated daily average water runoffs into the fjord (inner and middle zones of the fjord) are presented in Figure 11g. They include river and non-point runoffs estimated with the E-Hype model. (Water runoff is discussed in greater detail in Cieszynska and Stramska, 2017). Figure 11g shows that there is a significant discrepancy in the temporal patterns of freshwater water runoff and

17

ACCEPTED MANUSCRIPT

surface water export across transect A. Water runoff is subject to a strong seasonal cycle, whereas water export displays intense synoptic variability, which is wind-forced to a large degree. The average water runoff in days 120-190 in 2014 was 16.1 x 106 m3 per day. A similar estimate for the same time interval in 2015 was a little higher, about 18.1

RI PT

x 106 m3 per day. Water runoff after day 190 was quite low in both years. Total water transport across transect A, averaged over the time of this experiment in 2014 (based on a 1-m layer), was 32 x 106 m3 per day, whereas in 2015 it was only 0.6 x 106 m3 per day. The difference in the long-term average transport of surface waters in 2014 and 2015 is

SC

larger than the difference in water runoff from the land. This shows that water runoff does not have a direct relationship with water export patterns at a seasonal time scale. In addition, if the 1-m based approximation is multiplied by 5, the estimates of water

M AN U

transport will increase to about 160 x 106 m3 per day and 30 x 106 m3 per day for 2014 and 2015, respectively, figures that probably still underestimate the total transport of surface water. Thus, the total export of surface waters across transect A is significantly larger than the terrigenous water runoff.

The vertical profiles of salinity at three stations between the head and the mouth

TE D

of the fjord shown in Figure 12 indicate that water runoff significantly affects the functioning of the fjord. The water is essentially separated into two layers, with the upper one less saline than the lower one at each position along the fjord. The upper layer increases in salinity from lower values at the head to a value close to that of the open sea

EP

at the mouth of the Porsangerfjord. Fresher water flowing out over saltier open ocean water creates a strong halocline, highly stable in the vertical, that inhibits mixing with

AC C

water below the halocline. This can lead to higher temperatures in the surface layer during the warm season than would be the case if there were no halocline. Figure 11 shows that freshwater runoff into the Porsangerfjord has a pronounced seasonal variability, since it is fed by meltwater from snow. Therefore, runoff increases in the summer to many times the winter rate and lags behind the snowfall by several months. Variability in runoff results in seasonal and synoptic fluctuations of salinity in the inner part of the fjord (see also Bialogrodzka et al., 2017). Before it flows out to the open sea in the upper layer, terrigenous water mixes to some extent with the more saline water. This conclusion is drawn from the gradual

18

ACCEPTED MANUSCRIPT

increase in surface water salinity along the fjord. This mixing can be attributed to tides, wind and internal waves. Because of this process, the outflow of surface water at the fjord mouth exceeds the freshwater runoff from the land. This means that a compensatory inflow of seawater has to take place below the upper layer. The inflow and the outflow

RI PT

have to be dynamically associated if the average salinity of the waters inside the fjord is not to change over a longer time scale. In such a scenario, therefore, an increase in freshwater runoff would tend to reduce the salinity of fjord waters but would also accelerate the inflow of seawater, thereby increasing the salinity. This scenario is in

SC

agreement with our general understanding of the estuarine-like circulation in other highlatitude fjords (Cottier et al., 2010, Inall and Gillibrand, 2010), but our results provide

from the Porsangerfjord.

M AN U

quantitative and more detailed information regarding surface currents and water export

In summary, the net seaward (outward) flow in the upper water layer must be compensated by a net inward flow in deeper waters. This pattern in the fjord circulation is due to freshwater being supplied to the fjord by rivers and non-point runoff. Wind, and in particular its V component, plays an important part in governing temporal patterns of

TE D

surface water exchange between the fjord and the coastal zone of the Barents Sea. It is possible that if there were a large departure from the normal wind pattern, the hydrographic conditions inside this fjord would change significantly.

EP

5. Conclusions

This paper focuses on the Porsangerfjord, one of the largest fjords in northern

AC C

Norway and discusses surface currents and surface water transport out of it. This information is important to scientists interested in research on the advection of small marine organisms, as well as transport with surface waters of terrigenous material from the land to the ocean. The results can, for example, be used as a basis for interpreting the variability of bio-optical properties in surface waters and can help with the interpretation of satellite ocean color data in this region. The results of this work indicate that the most important tidal components in the surface currents are M2 and S2. However, only 10–30% of the variance in surface currents is attributable to tidal currents; the remainder (70–90% of variance) is due to the

19

ACCEPTED MANUSCRIPT

wind-induced and geostrophic currents. In particular, surface currents respond to winds. The inertial component is a significant contribution to residual currents, as documented by the shape of rotary spectra. Surface currents in the fjord are also significantly influenced by direct wind stress forcing. The statistical analysis shows that this

RI PT

relationship is quite complicated, owing to the overall variability of the wind field. Stronger wind events occurred when winds were blowing from the west to the east (azimuth 60 to 120 degrees) and residual currents exceeding 0.25 ms-1 and 0.35 ms-1 were more frequent in such situations in comparison to other wind directions.

SC

Temporal patterns in wind and current statistics were documented, although the present measurements did not cover an entire year. In spring (time interval #1 in 2014

M AN U

and 2015), stronger winds and residuals were more frequent than in summer (interval #2 of 2014 and 2015). In fall (time interval #3 in 2014), there was also a higher percentage of stronger winds than during summer (time interval #2). There were more events with stronger residual current speeds during time interval #1 in 2015 than during any other time interval. The fact that statistics for interval #1 in 2014 and 2015 were different reflects the response of currents to the interannual variability in winds during spring in

TE D

this region (Cieszynska and Stramska, 2017). Residuals with speeds > 0.25 ms-1 during time interval #1 were more frequent at point A3 than at A1, most likely because of the longer wind fetch at pixel A3 for winds blowing from westerly directions. Unfortunately, as no HF radar data were gathered in winter, it is impossible to describe the full annual

EP

cycle of surface currents. Analysis of meteorological data makes it clear that wind speeds and directions undergo a pronounced annual cycle (Svendsen, 1995; Cieszynska and

AC C

Stramska, 2017), so this is expected to be reflected in surface circulation patterns. Subsurface currents were recorded only at one location during a two-week time interval. These data show that subsurface currents (20 m depth and more) were weaker than surface currents and were not well correlated with surface currents, but data record is too short to carry out a more detailed analysis of subsurface currents. Water runoff significantly affects fjord hydrography. The surface water layer is less saline than the deeper layer at each position along the fjord. The upper layer increases in salinity from lower values at the head to a value close to that of the open sea at the mouth of the Porsangerfjord. A strong halocline induces a high vertical stability

20

ACCEPTED MANUSCRIPT

that inhibits mixing of surface and deep water. This was also reported earlier by Svendsen (1995). Freshwater runoff into the Porsangerfjord has a pronounced seasonal variability, increasing in the summer to many times the winter rate. HF radar observations of surface currents in 2014 and 2015 enabled the transport

RI PT

of surface water across transect A in the outer part of the fjord to be estimated. The estimate is based on decomposing surface currents into components that are parallel and perpendicular to the transect line. The conclusion was drawn that although there are events when more surface water is transported into than out of the fjord during a day, a

SC

net seaward (outward) flow in the upper layer prevails over extended time intervals. This flow is partly compensated by a net inward flow in the deeper layer. This general pattern

M AN U

in the fjord circulation is associated with freshwater supplied to the fjord by rivers and non-point runoff. Wind, and in particular its V component, plays an important role in governing temporal patterns of surface water exchange between the fjord and the coastal zone of the Barents Sea.

If a full understanding is to be acquired of how the Arctic functions, an environment that is extremely variable spatially, research needs to be expanded to include

TE D

a range of different local environments. In the future, this will enable knowledge to be extrapolated to larger scales with better precision. This paper has shown how remote sensing of surface currents with the HF radar system can be used to supply more information about fjords situated in harsh climatic conditions that make in situ

AC C

EP

experiments difficult to carry out.

21

ACCEPTED MANUSCRIPT

Acknowledgements We are grateful to Thomas Helzel and his staff from Helzel Messtechnik GmbH for their support with all aspects of HF radar system operations and Hans Kristian Strand from the

RI PT

Institute of Marine Research for his help with the field logistics. Many thanks go to Jagoda Bialogrodzka, Dariusz Ficek, Mateusz Jakowczyk, Daniel Materka, Roman Majchrowski, Roman Marks, Marek Swirgon, Marzena Wereszka and Tomasz Żmójdzin for their participation in the NORDFLUX fieldwork. The authors are grateful to all the

SC

persons involved in the programs providing free access to the sea level (the Norwegian Hydrographic Service, http://vannstand.no), water runoff (Swedish Meteorological and

M AN U

Hydrographical Institute, hypeweb.smhi.se) and meteorological data (Norwegian Meteorological Institute, http://www.yr.no) used in this study. Wind roses are based on Daniel Pereira’s free access Matlab code. Constructive comments and suggestions from the reviewers are greatly appreciated.

Funding

TE D

This work was funded by the Norway Grants through the Polish-Norwegian Research Program operated by the National Centre for Research and Development (NCBR contract No. 201985 entitled ‘Application of in situ observations, high frequency radars, and ocean color, to study suspended matter, particulate carbon, and dissolved organic carbon

EP

fluxes in coastal waters of the Barents Sea’). Partial funding for MS and AJ comes also

AC C

from the statutory funds at IO PAN.

22

ACCEPTED MANUSCRIPT

Tables Table 1. Summary of the results of the statistical analysis of winds and surface currents.

5.5

4.0

14.9

3.7

4.6 2.8 2.2 3.7 9.2 6.2 12.9 11.2 6.5 11.2

13.4 x x x 13.6 13.4 25.6 17.5 8.3 5.3

7.8 x x x 6.1 3.4 19.9 8.3 0.3 1.0

15.3 x x x 6.8 14.0 16.0 12.4 11.0 8.5

6.7 x x x 3.3 4.5 9.6 4.7 3.0 3.0

RI PT

8.9

6.7

0.2

8.6 x x x 8.6 14.2 13.3 9.6 11.8 9.5

3.8 x x x 6.7 9.3 10.6 3.8 5.0 3.0

SC

-180 to 150 -150 to 120 -120 to -90 -90 to -60 -60 to -30 -30 to 0 0 to 30 30 to 60 60 to 90 90 to 120 120 to 150 150 to 180

% % % Residual current % Residual current azimuth wind speed speed at A1 speed at A3 % >7.5ms-1 > 10ms-1 >0.25ms-1 >0.35ms-1 >0.25ms-1 >0.35ms-1 20.6 7.6 2.8 10.1 1.8 7.8 0.7

M AN U

Wind azimuth

Table 2. Frequency of occurrence of residual current speeds for selected wind conditions.

60 to 90

% Residual current % Residual current speed at A1 speed at A3 >0.25 ms-1 >0.35 ms-1 >0.25 ms-1 >0.35 ms-1 13.3 3.6 10.1 5.8 14.0 10.2 10.9 14.1 29.4 23.5 21.4 17.1

TE D

Wind direction Wind speed (degrees) (ms-1) <5 7.5 to 10 >10

90 to 120

5.0 14.9 41.9

4.2 3.2 14.0

7.6 12.6 12.0

3.3 7.4 8.0

AC C

EP

<5 7.5 to 10 >10

23

ACCEPTED MANUSCRIPT

Table 3. Statistics for winds and residual currents in different time intervals. A – average, M – median, 10th and 90th are the percentiles.

A 0.14 0.17 0.13 0.23 0.15

RI PT

#1/2014 #2/2014 #3/2014 #1/2015 #2/2015

SC

A 0.15 0.16 0.16 0.17 0.14

M AN U

#1/2014 #2/2014 #3/2014 #1/2015 #2/2015

Wind speed 10th 90th >6ms-1 >8ms-1 2.4 9.1 40.3% 15.4% 1.2 7.9 23.0% 9.1% 1.6 10.4 41.3% 22.2% 2.0 9.7 50.8% 26.0% 1.3 7.0 17.7% 5.3% A1 Residual current speed M 10th 90th >0.25ms-1 >0.35ms-1 0.14 0.06 0.28 12.0% 3.9% 0.14 0.05 0.31 17.6% 6.4% 0.14 0.05 0.29 17.8% 4.1% 0.15 0.06 0.32 23.2% 7.3% 0.12 0.04 0.27 12.8% 2.7% A3 Residual current speed M 10th 90th >0.25 ms-1 >0.35ms-1 0.12 0.04 0.27 12.6% 1.4% 0.15 0.06 0.30 17.1% 4.5% 0.12 0.04 0.22 5.5% 0.4% 0.20 0.07 0.42 42.0% 20.5% 0.13 0.06 0.26 11.9% 3.5% M 5.4 4.0 5.3 6.2 4.3

AC C

EP

TE D

#1/2014 #2/2014 #3/2014 #1/2015 #2/2015

A 5.6 4.3 5.8 6.0 4.3

24

ACCEPTED MANUSCRIPT

References

RI PT

Arnone, R., Vandermeulen, R., Ladner, S., Ondrusek, M., Kovach, C., Yang, H., Salisbury J., 2016. Diurnal changes in ocean color in coastal waters, 2016, Proc. SPIE 9827, Ocean Sensing and Monitoring VIII, 982711 (May 17, 2016); doi:10.1117/12.2241018, http://dx.doi.org/10.1117/12.2241018 Asplin, L., Johnsen, I. A., Sandvik, A. D., Albretsen, J., Sundfjord, V., Aure, J., Boxaspen, K. K., 2014. Dispersion of salmon lice in the Hardangerfjord, Marine Biology Research, 10, 3, 216-225, DOI: 10.1080/17451000.2013.810755

M AN U

SC

Barth, A., Alvera-Azcarate, A., Gurgel, K.-W., Staneva, J., Port, A., Beckers, J.-M., Stanev, E. V., 2010. Ensemble perturbation smoother for optimizing tidal boundary conditions by assimilation of High-Frequency radar surface currents – application to the German Bight. Ocean Science 6: 161-178, doi:10.5194/os-6161-2010 Bendat, J.S. and Piersol, A.G., 2011. Random Data: Analysis and Measurement Procedures. In: Wiley Series in Probability and Statistics. Wiley: 640 pp. Bloomfield P. 2000. Fourier Analysis of Time Series: an Introduction. Wiley, New York: 288 pp.

TE D

Białogrodzka, J., Stramska, M., Ficek, D., Wereszka M., 2017. Total suspended particulate matter in the Porsanger fjord (Norway) in the summer of 2014 and 2015. Oceanologia, https://doi.org/10.1016/j.oceano.2017.06.002. Broderstad, E.G., Eythórsson, E., 2014. Resilient communities? Collapse and recovery of a social-ecological system in Arctic Norway. Ecology and Society 19(3): 1.

EP

Cieszyńska A., Stramska M., 2017, Climate related trends and meteorological conditions in Porsanger fjord, Norway, Oceanologia, in press.

AC C

Cosoli S., M Gačić, A Mazzoldi, 2012, Surface current variability and wind influence in the northeastern Adriatic Sea as observed from high-frequency (HF) radar measurements, Continental Shelf Research, 33, 1–13. Cosoli S., M Gačić, A Mazzoldi, 2008, Variability of currents in front of the Venice Lagoon, Northern Adriatic Sea. Ann. Geophys., 26, 1–16. Cottier FR, Nilsen F, Skogseth R, Tverberg V, Svendsen H, Skarðhamar J (2010) Arctic fjords: A review of the oceanographic environment and dominant physical processes. Howe, J. A., Austin, W. E. N., Forwick, M. & Paetzel, M. (eds) Fjord Systems and Archives. Geological Society, London, Special Publications, 344, 35–50.

25

ACCEPTED MANUSCRIPT

Csanady G. T., 1982, Circulation in the coastal ocean, D. Reidel Pub., Chichester, UK, 279 pp. Cushman-Roisin B., Asplin L., Svendsen H., 1994. Upwelling in broad fjords. Continental Shelf Research 14: 1701-1721.

RI PT

Donnelly C., Andersson J.C.M. and Arheimer B., 2015. Using flow signatures and catchments similarities to evaluate the E-HYPE multi-basin model across Europe. Hydrological Sciences Journal 6667.

SC

Eilertsen H.C., Frantzen S., 2007. Phytoplankton from two sub-Arctic fjords in northern Norway 2002–2004: I. Seasonal variations in chlorophyll a and bloom dynamics. Marine Biology Research 3: 319–32.

M AN U

Ekman, V. W., 1905, On the influence of the earth’s rotation on ocean currents, Arch. Math. Astron. Phys (Arkiv for Matematik, Astronomi och Fysik), 2, 1-52. Foreman M.G.G., 1977. (on line version revised in 2004) Manual for tidal heights analysis and prediction. Pacific Marine Science Report 77-10, Institute of Ocean Sciences, Patricia Bay, Sidney, BC, 97 pp. Foreman M.G.G., 1978. (on line version revised in 2004) Manual for tidal currents analysis and prediction. Pacific Marine Science Report 78-6, Institute of Ocean Sciences, Patricia Bay, Sidney, BC, 57 pp.

TE D

Gonella J., 1972. Rotary component method for analyzing meteorological and oceanographic vector time series. Deep Sea Research 19, 833–846, doi:10.1016/0011-7471(72)90002-2

EP

Helzel T., Petersen L., Mariette V., Pavec M., 2011. Reliability of Coastal Radar WERA for Coastal Zone Management. Journal of Coastal Research, SI 64 (Proceedings of the 11th International Coastal Symposium), 1345 - 1347.

AC C

Helzel T., Kniephoff M., Petersen L., 2010. Oceanography radar system WERA: features, accuracy, reliability and limitations. Turkish Journal of Electrical Engineering and Computer Science, 18, 389- 397, doi: 10.3906/elk-1001-369 Hvingel, C., Kingsley, M. C. S., Sundet, J. H., 2012. Survey estimates of king crab (Paralithodes camtschaticus) abundance off Northern Norway using GLMs within a mixed generalized gamma-binomial model and Bayesian inference. ICES Journal of Marine Science, 69, 1416–1426. Inall, M. E.; Gillibrand, P. A. 2010, The physics of mid-latitude fjords: a review, Howe, J. A., Austin, W. E. N., Forwick, M. & Paetzel, M. (eds) Fjord Systems and Archives. Geological Society, London, Special Publications, 344, 17-33

26

ACCEPTED MANUSCRIPT

IPCC 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996 pp.

RI PT

Jakowczyk, M., Stramska, M., 2014. Spatial and temporal variability of satellite-derived sea surface temperature in the Barents Sea. International Journal of Remote Sensing 35: 6545–6560, doi: 10.1080/01431161.2014.958247

SC

Johnsen, I.A., Asplin, L.C., Sandvik, A.D., Serra-Llinares, R.M., 2016. Salmon lice dispersion in a northern Norwegian fjord system and the impact of vertical movements. Aquaculture Environment Interactions, 8, 99-116, doi:10.3354/aei00162

M AN U

Kokkini, Z., Potiris, M., Kalampokis, A., Zervakis, V., 2014. HF Radar observations of the Dardanelles outflow current in North Eastern Aegean using validated WERA HF radar data. Mediterranean Marine Science, 15 (SPECIAL ISSUE), 753-768, doi: 10.12681/mms.938 McWilliams, J. C., E. Huckle, J.-H. Liang, and P. P. Sullivan, 2012, The wavy Ekman layer: Langmuir circulations, breaking waves, and Reynolds stress. J. Phys. Oceanogr., 42, 1793–1816, doi:10.1175/JPO-D-12-07.1.

TE D

Myksvoll, M. S., Jung, K-M., Albretsen, J., Sundby S., 2013. Modeling dispersal of eggs and quantifying connectivity among Norwegian Coastal cod subpopulations. ICES Journal of Marine Science, doi:10.1093/icesjms/fst022.

EP

Myksvoll, M. S., Sundby, S., Ådlandsvik, B., Vikebø, F. B., 2011. Retention of coastal cod eggs in a fjord caused by interactions between egg buoyancy and circulation pattern. Marine and Coastal Fisheries, 3, 279-294, doi: 10.1080/19425120.2011.595258

AC C

Pedersen, O.P., Zhou, M., Tande, K.S, Edvardsen, A., 2005. Eddy formation on the coast of North Norway—evidenced by synoptic sampling. ICES J Mar Sci 62:615–628, doi: 10.1016/j.icesjms.2005.01.017 Polton. J. A., Lewis, D.M., Belcher, S. E., 2005, The role of wave-induced CoriolisStokes forcing on the wind-driven mixed layer, Journal of Physical Oceanography, 35, 444-457. Sivertsen, K., Bjørge, A., 2015. On the brink of the Arctic: Unusual intertidal sub-Arctic kelp associations in the Porsangerfjord, North Norway, Marine Biology Research, 11:4, 405-413, DOI: 10.1080/17451000.2014.940975

27

ACCEPTED MANUSCRIPT

Stramska, M, Jankowski, A., Cieszynska, A., 2016. Surface currents in the Porsanger fjord in northern Norway. Polish Polar Research, 37, 337–360, doi: 10.1515/popore-2016-0018

RI PT

Svendsen, H. 1995. Physical oceanography of coupled fjord-coast systems in northern Norway with special focus on frontal dynamics and tides. In: Skjoldal, H., Hopkins, C., Erikstad, K., Leinaas, H. (Eds.), Ecology of Fjords and Coastal Waters. Elsevier: 149-164.

SC

Skagseth, Ø., Drinkwater, K. F., Terrile, E., 2011. Wind‐ and buoyancy‐ induced transport of the Norwegian Coastal Current in the Barents Sea. Journal of Geophysical Research, 116, C08007, doi:10.1029/2011JC006996.

M AN U

Stanev, E. V., Ziemer, F., Schulz-Stellenfleth, J., Seemann, J., Staneva, J., Gurgel, K.-W., 2014. Blending Surface Currents from HF Radar Observations and Numerical Modeling: Tidal Hindcasts and Forecasts. Journal of Atmospheric and Oceanic Technology, 32, 256–281, doi: 10.1175/JTECH-D-13-00164.1. Tamura, H., Miyazawa, Y., Oey, L.-Y. 2012, The Stokes drift and wave induced mass flux in the North Pacific, Journal of Geophysical Research, 117, C08021, doi:10.1029/2012JC008113.

TE D

Wassmann, P., Svendsen, H., Keck, A., Reigstad M., 1996, Selected aspects of the physical oceanography and particle fluxes in fjords of northern Norway, Journal of Marine Systems, 8, 53-71,

EP

Windsland, K., Hvingel, C., Nilssen, E. M., Sundet, J. H., 2014. Dispersal of the introduced red king crab (Paralithodes camtschaticus) in Norwegian waters: a tag-recapture study. – ICES Journal of Marine Science, doi:10.1093/icesjms/fst241.

AC C

Wu, J., 1983, Sea-surface drift currents induced by winds and waves, Journal of Physical Oceanography, 13, 343-368.

28

ACCEPTED MANUSCRIPT

Figure Captions Figure 1. Location of the Porsangerfjord (indicated by the black rectangle) in the coastal region of the Barents Sea. Arrows indicate surface currents. The Norwegian Atlantic Current is shown by the solid lines, the Norwegian Coastal Current by dotted lines.

RI PT

Figure 2 a) Porsangerfjord bathymetry. HF radar locations are indicated by the letters A and B (the distance between A and B is ~ 27 km). Also shown are stations A1, A2, A3 and M discussed in the text. b) Topography of the land in the vicinity of the Porsangerfjord. Positions of meteorological stations (Lakselv and Honningsvaag) and the island of Tamsøya are indicated. The next fjord to the east is Laksefjord.

SC

Figure 3. Example maps of surface currents recorded by the HF radar system on June 27, 2015. More examples can be found in Stramska et al. (2016).

M AN U

Figure 4. Example time series of a) wind speed, b) wind azimuth, c) surface current speed, d) surface current direction, e) residual current speed, f) residual current direction. The data were collected during time interval #1 in 2015. Surface currents were obtained from the HF radar system at pixel A3 (see Figure 2 for the pixel position). Wind data are from the Honningsvaag station.

TE D

Figure 5. Example power spectra based on data collected in 2014 at location A1. Rotary spectra for a) and b) current velocities, c) and d) for current residuals. CCW and CW stand for counterclockwise and clockwise rotation, respectively. e) Normalized power spectra for sea level and current speed. g) Normalized power spectra for wind speed and current speed. Spectra in figure e and g have been normalized to the total variance in the respective time series. f) Squared coherence between sea level and current speed and h) between wind speed and current speed. Figure 6. Summary of wind conditions at Honningsvaag based on all the data collected during the experiments in 2014 and 2015: a) frequency distribution of wind azimuth, b) wind rose, c-f) frequency of occurrence of wind speed for different wind azimuths.

AC C

EP

Figure 7. Summary of residual surface currents based on data collected during the experiments in 2014 and 2015: a-h) data for pixel A3, i-p) data for pixel A1. Left-hand panel: frequency distribution of residual current speed for different wind azimuths; right-hand panel: frequency distribution of residual current directions for different wind azimuths. The wind azimuths are indicated in the legend. Figure 8. Comparison of frequency of occurrence of wind and residual current characteristics during different time intervals in 2014 and 2015. Plots a-f are for wind speed, plots g-l are for residual currents at pixel A3. The numbers 1-3 indicate the time intervals defined in the text. Figure 9. Summary of wind conditions. Each wind rose indicates wind azimuths and speed in a different time interval (indicated by the numbers 1-3 and defined in the text). Figure 10. Comparison of current speed (left-hand panel) and direction (right-hand panel) recorded at pixel M in 2014 by WERA radars (surface currents) and ADCP measurements (shown for selected depths). The wind direction recorded at Honningsvaag is also included in figure b.

29

ACCEPTED MANUSCRIPT

RI PT

Figure 11. a and b) Time series of daily averaged net transport of surface water (in m3 s-1) across transect A in 2014 and 2015 respectively (positive out of the fjord). The different numbers shown in the legend indicate different segments of transect. c) Time series of the V wind component (positive in the northward direction) recorded at Honningsvaag in 2014 and 2015. d) Comparison of the V component of the wind velocity recorded at Honningsvaag (VH) and Lakselv (VL). e) Velocity component of surface current (uR) at pixel A1 in the direction perpendicular to transect A, plotted as a function of the V component of wind velocity recorded in Honningsvaag. f) Daily averaged net transport of surface water (in m3 s-1) across transect A, plotted as a function of the V component of wind velocity recorded at Honningsvaag. g) Daily averaged water runoff estimated from the E-Hype model.

AC C

EP

TE D

M AN U

SC

Figure 12. a) Vertical profiles of water salinity in the inner, middle, and outer (station A2) parts of the fjord. b) Vertical profiles of water salinity recorded at station A2. Line numbers indicate dates: 1 – May 30, 2015; 2 – June 3, 2015; 3 – June 9, 2015; 4 – June 10, 2015; 5 – June 18, 2015. Lines 3 and 5 represent conditions when oceanic water inflow took place, whereas lines 1, 2 and 4 represent conditions when net water outflow prevailed.

30

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Figures

Figure 1. Location of the Porsangerfjord (indicated by the black rectangle) in the coastal region of the Barents Sea. Arrows indicate surface currents. The Norwegian Atlantic Current is shown by the solid lines, the Norwegian Coastal Current by dotted lines.

31

ACCEPTED MANUSCRIPT

b

AC C

EP

TE D

M AN U

SC

RI PT

a

Figure 2 a) Porsangerfjord bathymetry. HF radar locations are indicated by the letters A and B (the distance between A and B is ~ 27 km). Also shown are stations A1, A2, A3 and M discussed in the text. b) Topography of the land in the vicinity of the Porsangerfjord. Positions of meteorological stations (Lakselv and Honningsvaag) and the island of Tamsøya are indicated. The next fjord to the east is Laksefjord.

32

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

a

AC C

EP

TE D

b

Figure 3. Example maps of surface currents recorded by the HF radar system on June 27, 2015. More examples can be found in Stramska et al. (2016).

33

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

Figure 4. Example time series of a) wind speed, b) wind azimuth, c) surface current speed, d) surface current direction, e) residual current speed, f) residual current direction. The data were collected during time interval #1 in 2015. Surface currents were obtained from the HF radar system at pixel A3 (see Figure 2 for the pixel position). Wind data are from the Honningsvaag station.

34

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 5. Figure 5. Example power spectra based on data collected in 2014 at location A1. Rotary spectra for a) and b) current velocities, c) and d) for current residuals. CCW and CW stand for counterclockwise and clockwise rotation, respectively. e) Normalized power spectra for sea level and current speed. g) Normalized power spectra for wind speed and current speed. Spectra in figure e and g have been normalized to the total variance in the respective time series. f) Squared coherence between sea level and current speed and h) between wind speed and current speed.

35

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

b

Figure 6. Summary of wind conditions at Honningsvaag based on all the data collected during the experiments in 2014 and 2015: a) frequency distribution of wind azimuth, b) wind rose, c-f) frequency of occurrence of wind speed for different wind azimuths.

36

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 7. Summary of residual surface currents based on data collected during the experiments in 2014 and 2015: a-h) data for pixel A3, i-p) data for pixel A1. Left-hand panel: frequency distribution of residual current speed for different wind azimuths; right-hand panel: frequency distribution of residual current directions for different wind azimuths. The wind azimuths are indicated in the legend.

37

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 7. Continued

38

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

Figure 8. Comparison of frequency of occurrence of wind and residual current characteristics during different time intervals in 2014 and 2015. Plots a-f are for wind speed, plots g-l are for residual currents at pixel A3. The numbers 1-3 indicate the time intervals defined in the text.

39

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Figure 9. Summary of wind conditions. Each wind rose indicates wind azimuths and speed in a different time interval (indicated by the numbers 1-3 and defined in the text).

40

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 10. Comparison of current speed (left-hand panel) and direction (right-hand panel) recorded at pixel M in 2014 by WERA radars (surface currents) and ADCP measurements (shown for selected depths). The wind direction recorded at Honningsvaag is also included in figure b.

41

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 11. a and b) Time series of daily averaged net transport of surface water (in m3 s-1) across transect A in 2014 and 2015 respectively (positive out of the fjord). The different numbers shown in the legend indicate different segments of transect. c) Time series of the V wind component (positive in the northward direction) recorded at Honningsvaag in 2014 and 2015. d) Comparison of the V component of the wind velocity recorded at Honningsvaag (VH) and Lakselv (VL). e) Velocity component of surface current (uR) at pixel A1 in the direction perpendicular to transect A, plotted as a function of the V component of wind velocity recorded in Honningsvaag. f) Daily averaged net transport of surface water (in m3 s-1) across transect A, plotted as a function of the V component of wind velocity recorded at Honningsvaag. g) Daily averaged water runoff estimated from the E-Hype model.

42

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Figure 12. a) Vertical profiles of water salinity in the inner, middle, and outer (station A2) parts of the fjord. b) Vertical profiles of water salinity recorded at station A2. Line numbers indicate dates: 1 – May 30, 2015; 2 – June 3, 2015; 3 – June 9, 2015; 4 – June 10, 2015; 5 – June 18, 2015. Lines 3 and 5 represent conditions when oceanic water inflow took place, whereas lines 1, 2 and 4 represent conditions when net water outflow prevailed.

43