Investigation Hydrometeorological Regime of the Barents and White Seas

«Living Planet Symposium 2016» Prague, Czech Republic, 9-13 May 2016 Based on Satellite Altimetry Data Geophysical Center, Russian Academy of Sciences Space Research Institute, Russian Academy of Sciences Sergey A. Lebedev [email protected] [email protected]

G FO T Kanin Nos The White Sea is a semi-enclosed mediterranean sea (Fig. 1b). rac k 1 The sea border with the Barents Sea is a line joining Cape 65 E Calibration/Validation Wind Speed R The Barents and White seas S a l Svyatoy Nos (northeastern coast of Kola Peninsula) with Cape & u E s T N 68° r n E a V i Kanin Nos (northwestern extremity of Kanin Peninsula). The T c n R I r k S a S A e c 5 & 5 T P k The Barents Sea is one of the shelf Arctic seas. It is located on 8 northern part of the sea is called the Voronka (funnel). The E 816 n N Kola i

V n I the continental European shelf between northern coast of southern and central parts of the White Sea called the Basin are In the framework of the ALTICORE project, the consistency between altimeter- S Peninsula a A K T T A 5 4 Europe and three archipelagoes – Novaya Zemlya, Franz Josef the largest and deepest parts of the sea. There are also several 2 7 derived wind speed and measurements on coastal meteorological stations was S I 0 0 k K V k c a T an N r d ac T A Land and Spitsbergen (Fig. 1a). Its open water area is large and shallow bays in the area, namely the Dvinsky, Onega, al E r Mezen checked. Satellite-derived wind data used were obtained from the RADS data base. O S a T I 797 k & F Mozhovets Is. 2 3 sha G V k Bay S N B ac approximately 1,424,000 km and total volume is 322,000 km . R E r Mezen and Kandalaksha bays. The Gorlo (neck) is a narrow For the comparison, 30 MS on the Barents Sea coast and 21 MS on the White Sea a E y & T

S The Barents Sea shelf is rather deep. More than 50% of the strait connecting the Basin and Voronka. The total water coast were selected. As an example, we show results of the comparison for two MS R E 66° M 2 e z area has a depth of 200-500 m. The average depth is 222 m, the surface area is 90,873 km including islands, and the total – Kem’ Port (White Sea) and Kanin Nos (Barents Sea). G e FO n k 061 3 Tr rac R a & J1 T iv maximum depth in the Norwegian trench reaches 513 m and in volume is 6,000 km including also the Voronka area opening The ERS & ENVISAT, GFO and T/P & J1 satellite tracks, which were used when ck T/P e 4 r 25 Solovetsky the Franz Josef Land straits it exceeds 600 m. The Barents Sea to the Barents Sea. Thus the White Sea covers approximately obtaining satellite-derived wind speed data for Kem’ Port and Kanin Nos are shown Islands 2 Kem’-Port Dvinsky watershed area is 668,000 km large. The total river runoff is 6% of the total open water area of both seas and comprises only in Fig. 3. Examples of altimeter and observed (in-situ) data on wind speed and Bay 3 O n 163 km /yr, 80-90% of it falls on southeastern region of the 2% of the total volume of marine water, but it assumes more corresponding scatter plots for these MSs are given in Figs. 4, respectively. e g a sea. The largest river flow-ing into the Barents Sea is the than half of the river runoff in the region. The White Sea B In the cases of both Kem’ Port and Kanin Nos (Fig. 3) MS databases for 2000-2007 a Sev y erna

2 ya

D

Pechora River which has about a half of watershed area watershed area is 729,000 km (Atlas of Arctic 1985). The total v

were used. Altimeter wind speed values for Kem’ Port were in most cases i n

2 3 64° a 3 O 322,000 km ; its runoff is 130 km /yr. The river runoff is river runoff is 259 km /yr that is about 4% of the total amount n noticeably less than meteo ones and correlation between them was practically e g R R a i iv ve essentially re-flected in hydrological conditions only in a of the White Sea water volume. The main rivers are the absent (r = 0.16). Influence of land may be supposed as a possible reason of such er r southeastern part of the sea. Therefore this area is sometimes Severnaya Dvina, Onega and Mezen having runoff of 111, 18 poor correlation. Better correlation (r = 0.45) was observed in the Kanin Nos case. 35° 40° 45° referred to as the Pechora Sea. and 26 km3/yr correspondingly. Improvement in correlation between meteo and altimeter data on wind speed can be obtained when using decomposition of the winds in four quadrants according to Fig. 3. ERS & ENVISAT, GFO and T/P & J1 satellites wind direction relative to coastline direction. An example of its application for the tracks in the White and Barents seas which were used for Kanin Nos case is shown in Fig. 5. The reference line was chosen to be parallel to the comparative analysis between altimetric and MS in- the axis of this Peninsula. situ data. Red circles mark locations of MSs.

Cape Kanin Nos a) Franz Josef Land b)

Polar GMO Cape im. E.T.Krenkelja Svyatoy

a Nos l

u Spitsbergen 68° s n 79° i n Voronka e P

a) b) Barentsburg n Kola i

Great n Bank Peninsula a Barents Kandalaksha K

Russkaya Gavan K en Sea a Umba rg nd be al Mezen ts ak pi a Sosnovets Mozhovets Is. S nk sha Bay a Central y 75° B l B Bank m a o y rl Bear Is. e o N Z G o gh h M r ou c 66° tr e . T n w Is re z ear e B T n S l a e ra Basin e t y Kara g n R e i a a v i e C r a v Solovetsky n o Sea Islands

N Kem’-Port Dvinsky Hohhingsvarg 71° Solovki Bay O Vardo n e Bakaritsa g Pechora a Kolguev Is. Pechenga Kyslogybskaya B Severodvinsk Sea S Polyarnuy a ev y erna Teriberka Cape Bugrino ya

Murmansk Kanin Nos Tobseda D

Varandey v

Cape i Cape n

Fig. 4. Scatter plot of observed versus GFO altimeter data for Kem’ Port (a) and Kanin Nos (b).

Svyatoy Nos Onega a 64° a W l Segeysiy u Konstantinovskiy s

S n Kola i h Shar n O Iokanga e i e P n a te ra e n Peninsula i ho g R n Cape c r R a e e a i Indiga P iv i v K Mikulkin R v e er r 67° 20° 30° 40° 50° 60° 70° 35° 40° 45° Fig. 5. Scheme of decomposition of wind directions in the Calibration/Validation Sea Surface four quadrants (Q1-Q4) in relation to the orientation of the coastline for the Kanin Nos case. The appropriate Fig. 1. Maps of the Barents (a) and the White (b) seas. Dashed lines show boundaries of the seas and their internal parts. correlation coefficients between wind speed from altimetry Circles mark tide gauges location.. Height and in-situ MS records are shown in the scheme. Arrows show prevailing wind direction for each quadrant. The consistency between altimeter sea surface height (SSH) data acquired by the satellites T/P, ERS-1/2, ENVISAT, GFO and J1, and measurements of 18 TGs 0

located in the Barents Sea and 8 TGs – in the White Sea was investigated (Fig. 1). 4 )

Meteorological conditions in the Barents Sea are determined by of seasonal variations in the White Sea level is observed in the m Vardo TG Processing of satellite altimetry data included calculation of sea surface height c (

30 ERS-2 & ENVISAT y atmospheric cyclones that are formed in the North Atlantic and estuary zones of the Onega and Severnaya Dvina rivers and l anomaly (SSHA) with using all corrections without tidal and inverse barometer a m 20 move to the Barents Sea. In winter, southwesterly and southerly may reach 30-60 cm. o

correction. Then, for every cycle of the selected satellite passes, spatial coordinates n A

winds prevail in the southern part of the sea, whereas Sea level variations related to storm surge are formed by strong t

of the point nearest to the concrete TG location (within 15 miles) were determined. 10 h g i southeasterly and easterly winds are most frequent in the winds associated with passage of intense atmospheric cyclones. For the Barents Sea, correlation between SSHA derived from altimetry and tide e H 0

e

northern regions. Monthly mean wind speed in the southern The western and northwestern cyclones dominate and amount gauge data can be fairly high (> 0.85) when using ERS-1/2 and ENVISAT data. For c a f r

regions (in particular, nearby Kanin Nos) reaches about 10 m/s, to 88% from their total number. During the extreme situations u GFO data the correlations are noticeably lower, probably due to orbit errors and -10 S

a

to the north it decreases to 6 m/s. In summer the pressure the storm surge height can be comparable with the tidal height. inaccurate corrections. Correlations are extremely high (0.99) when ERS-1/2 and e S gradients are weakened. Homogeneous wind conditions, with The storm surge values are greatest in the southeastern part of ENVISAT data are used for the comparison with Vardo and Hohhingsvarg TGs for -20 predominance of winds of northern directions, dominate over the Barents Sea (up to 2 m). For the Kola coast and Spitsbergen which there are long and high quality time series. These high correlations might be 2002 2003 2004 2005 2006 2007 2008 2009 almost the whole sea. Monthly mean wind speed in this season area storm surge heights are about 1 m; the smaller values of explained by the tide effect, which is very significant in the Barents Sea (see Year is about 6 m/s, on the north – 5 m/s. 0.5 m are observed near the Novaya Zemlya coast. As concerns above). Moreover, these TGs are located in the area where nonlinear and residual In the White Sea, winds from the south, southwest and west the White Sea, on the average the storm surge is of 30-70 cm, tidal phenomena are not important. For TGs in Murmansk, Polyarny, Indiga and Fig. 6. Time variability of Vargo TG and ERS-2 & prevail from October to March, whereas in May - August winds but in the Onega Bay it can reach about 80 cm. others which are located in fjords or estuaries the correlations are small (< 0.4). ENVISAT SSHA . from north, northeast and east are most frequent. Southeasterly Tides play the key role in the coastal water dynamics in the Absence of 1Hz altimetric measurements in these areas or strong effect of nonlinear winds are frequently observed at the top of the bays (in ports Barents and White seas, and dominate in total sea level and residual tidal phenomena may be a cause of it. Details of SSHA comparison are 1000 1000

a)M2 b) 100 )

Mezen’, Kandalaksha, Onega). Monthly mean wind speed in variations. In the southern part of the Barents Sea the r h

shown in Fig. 6. hr) 2 2 m

( 10 100

y S2 t K the open sea and on islands is 7-10 m/s from September to semidiurnal tide reaches amplitudes up to 2 m which is the i 1 s

For the White Sea, a comparison between satellite and TG SSHA data also shows n 1 e M2 D

l a

April and 5-7 m/s from May to August. In the bays running largest in the Arctic seas. In the northern and north-eastern r t 0.1 10 significant correlations (>0.6) for all the satellites except GFO. Greater values of c e p S 0.01 (m Density Spectral deep in land, mean wind speed does not exceed 3-5 m/s during Barents Sea the tidal height decreases and for the Spitsbergen K S 1 correlation for TG Onega (0.76 for T/P and J1 data; 0.96 for ERS-1/2 and 2 0.001 1 whole year. coast equals to 1-2 m. For the southern coast of the Franz 2 4 6 8 10 12 14 16 18 20 22 24 26 2 4 6 8 101214161820222426 ENVISAT data) and TG Severodvinsk (0.97 and 0.98, respectively) are Period (hr) Period (hr) The general circulation and sea level variations in the Barents Joseph Land it is of 40-50 cm only. It is explained by the 1000 1000 condinioned by their location. Both TGs are situated in river estuaries, thus river c) d) )

r M 2 M2 h

hr) 2 and White seas are formed under cumulative effect of wind peculiarities of the bottom relief, the coastline configuration, 2 m

runoff has a strong influence on the hydrological regime in these parts of the White ( 100 100

y t i s n

forcing, water interaction and ex-change between surrounding and the interference of tidal waves, which come from the North e

Sea. Correlation minimum (0.66 and 0.61, respectively) is observed at TG Kem’ D

l a

r K1 t 10 10 c S2 seas, strong tides, peculiarities of bottom topography, seasonal Atlantic and Arc-tic oceans. The tidal heights in the White Sea e Port where nonlinear and residual tidal phenomena are important on shallow water. p S Spectral Density (m Density Spectral K1 S variability of river runoff, precipitation and ice cover, and other vary from 8 m in the Mezen Bay to 1 m in the Dvinsky Bay . 2 According to the results obtained, satellite altimetry SSHA data are in good 1 1 2 4 6 8 10 12 14 16 18 20 22 24 26 2 4 6 8 101214161820222426 factors. Thus, sea level variations in the Barents and White seas As a result of nonlinear tidal phenomena and nonlinear consistency with TG SSHA. Period (hr) Period (hr) have a complex nature and are characterized by a significant interaction of the main tidal components (M2, S2, N2, K2, K1, spatial and temporal variability. O1, P1, Q1) there is a set of additional tidal harmonics. The seasonal variations of the sea level in the Barents and According to the results of numerical simulation an amplitude Fig. 7. Spectral density of the SSHA time series at the TG White seas are caused by an impact of atmospheric pressure of nonlinear harmonics in the average is about 5% of the total Solovki (a) and the superposition of main tidal constituents M2, S2 and K2 calculated with tidal models: and wind, temperature and salinity, river runoff, precipitation, tide height for the Barents Sea and about 50% for the White Calibration/Validation Tidal Models (b) – CSR4.0, (c) – GOT00.2 and (d) – HMRC. and ice cover. For the Barents Sea the maximum of seasonal Sea with the exception of the Voronka (10-25%). The sea level variation is observed in November-December and its maximum value (more than 25%) is observed between the minimum is registered in May-June, that is in accordance with Spitsbergen Bank and Central Bank in the Barents Sea, and in Peculiarities of the tidal regime in the Barents and White seas make inaccurate use of global tidal models for a processing (tidal correction) of the atmospheric pressure effect upon the sea surface. For the Pechora Sea. In the White Sea the maximum value is about satellite altimetry data. Therefore, to improve altimetry data processing in these regions, it is very important to use regional tidal model. example, the difference between maximum and minimum mean 100% for the Severnaya Dvina and Kandalaksha bays (Fig. 2). Global tidal models At present there are more than 11 global tidal ocean models. They differ by spatial resolution and procedures of sea level in Murmansk can amount to 40-50 cm. A range assimilation of TG and satellite altimetry data. A variety of tests indicates that all these tidal models agree within 2–3 cm in the deep ocean, and they represent a significant improvement over the classical Schwiderski 1980 model. But global tidal models ignore nonlinear, residual and other tidal phenomena which inherent to coastal and shallow water area of marginal and semi-enclosed seas. In our investigation global tidal models CSR4.0 and GOT00.2 were used. Regional tidal model Three dimensional baroclinic model with free surface was developed in Laboratory of Sea Applied Research at HMRC

of Russian Federation. For the Barents and White seas, only M2 S2 and K2 main tidal components were used in calculation. Mean temperature 1

0 and salinity fields, as well as atmospheric fields from the NCEP/NCAR Reanalysis project were used in calculation of the Barents Sea level a) 5 Franz Josef Land b) and currents for 1948-2008. In contrast to global models results of the HMRC model represent fields of total tide height for each time Polar GMO im. E.T.Krenkelja

a moment with a given time step. l Kola u Spitsbergen 68° s n

79° i For comparison of global and regional tidal models spectral density of sea level was calculated for all TGs and for different main tidal

Peninsula 1 n

0 e 5 P constituents – M , S and K with time step of 1 hour. Analysis of SSHA was based on the results of harmonic analysis of temporal series. 2 2 2 Barentsburg n i 5 2.5 5 10 25 50 75 100 n

a Spectral density of TG Solovki (Fig. 7) includes main tidal constituents (M2, S2 and K2), their nonlinearly interference (see peaks with period

0 K

1 Kandalaksha 2

5 less than 8.5 hours) and periods related to synoptic dynamics. 1

0

5 Russkaya 2 5 0 Gavan 10 Umba

1 5 7

0 Mozhovets Is.

a Sosnovets

y 5 0 75° l m

e Bear Is. 0

25 Z 5 75

0 M 5 66° 5 Long-Term Variability of Sea Level e 1 z 0 e

10 1 n a

5

0

0 y 0 R i

a v

e

1 r 79° v

0 5 Solovetsky 7 o

0 Islands 5

5 N

1

Kem’-Port 0 Hohhingsvarg 0 The trend of the Barents Sea level estimated from TG observational data is

71° Solovki 2 Vardo 5 -0.05 ± 0.05 cm/yr after all corrections. It may be supposed that negative value of 5 Bakaritsa

25 0 1

Kolguev Is. 2 0 5 Severodvinsk Pechenga Kyslogybskaya the Barents Sea level trend is a result of a slow rise of Fennoscandia and Novaya

Se

Polyarnuy ver 0 n

Teriberka Bugrino 5 aya

Murmansk 0 Tobseda D

5 Zemlya. Coast of Kola Peninsula of the Barents and White seas uplifts with a rate

Varandey v

2 i

Cape n 75°

5 Onega a 64° a W l Segeysiy u Konstantinovskiy s

S n h i Shar n O Iokanga e ite e of 4-8 mm/yr. Rate of ascend of Earth’s crust in the region of Novaya Zemlya P n a ra e n i ho g R n Cape c r R Kola a e e a i Indiga P iv i v 2.5 5 10 25 50 75 100 K Mikulkin R v e Peninsula er r 67° Island is 1–2 mm/yr. For interpretation of climatic sea level change it is desirable 20° 30° 40° 50° 60° 70° 35° 40° 45° to keep in mind that the Barents Sea shelf is sagging. With the end of investigating long-term variability of sea level, SLA (sea level anomaly) was calculated with regional tide HMRC model in three points (Fig. 8). 71° A Fig. 2. The ratio (%) of the total nonlinear harmonics amplitudes to the total main tidal components amplitudes for the Barents (a) and White (b) seas. Points A and B are located near Scandinavia and Kola Peninsula. Their position B coincides with isoline 2 cm of climatic dynamic topography field obtained with hydrodynamic model HMRC without account for tide-generating force. Third point C is located on the boundary between the Barents and White seas. Altimetric C data from the satellites ERS/ENVISAT and GFO-1 were used for processing. 67° Analysis of temporal SLA variations in the point A shows that SLA increases in 20° 30° 40° 50° 60° 70° Interannual Variability of Sea Ice Edge March 1998 – May 1999, May 2001 – August 2001 and June 2005 – September 2007 with a rate of 4.58 cm/yr, 9.62 cm/yr and 2.00 cm/yr, respectively (Fig. 9). 79° Ice cover in these time periods is less than in the following years. So it is possible Fig. 8. Location of points A, B and C superimposed on Position that this phenomenon was associated with intensification of the Norwegian Coastal climatic dynamic topography field obtained with Current. hydrodynamic HMRC model for time period 1958-2008. Traditional research of Arctic ice cover and ice edge location are based on infrared and microwaves radiometry data. But satellite altimetry data may be 75° also used to estimate ice cover extent or rather position of ice edge. 0 Acknowledgements 0 For this purpose 34 descending tracks of the satellites ERS-2 and ENVISAT Point A ) ) m

This study was supported by a m k have been chosen (Fig. 10). These tracks are located under the optimal angle to c ( (

k

series of grants of the Russian y c l a mean climatic sea ice edge position. Significant SSH records along the satellite a r m T

Science Foundation (No 14-17- 400 600 80 o g

tracks over the sea surface and over the ice fields strongly differ. This point of n 5 10152 n A o

00555) l 71° l a e

the signal change located on the track was identified as an ice edge. Some v e e c L n temporal variations of ice edge position along tracks are shown in Fig. 11. a a t e s i From 1995 to 2008 sea ice edge didn’t reach its mean climatic position along S D 988 track. For 932 track sea ice edge observed in the winters of 1996, 2003 and Track 932

2004 was displaced southwestward by 60 km, 40 km and 53 km, respectively, -10 -5 0 -200 0 200 67° from its mean climatic position. For both tracks the 2000 and 2007 winters 1994 1996 1998 2000 2002 2004 2006 2008 2010 1996 1998 2000 2002 2004 2006 2008 2010 Year Year 20° 30° 40° 50° 60° 70° were very warm. According to our calculations, interannual trends of sea ice position northeast- Fig. 10. Ground tracks of 34 descending ERS-2 and ENVISAT passes which are used for analysis of ward along 988 and 932 tracks were 13.2±0.6 km/yr and 10.7±0.4 km/yr, Fig. 11. Temporal variability of sea ice edge position Fig. 9. Variability of SLA in points A based on ERS and alongtrack sea ice edge position. Red line is mean respectively. For summer months, similar trends were 24.8±1.2 km/yr and along 932 (descending ERS-2 and ENVISAT tracks. ENVISAT data. climatic sea ice edge position. Green line shows position 8.2±0.5 km/yr, respectively. Averaged velocity of the ice edge along track Dashed line shows interannual trend. Zero corresponds to of 988 track and purple line – of 932 track. displacement for all tracks was 11.8±0.5 km/yr. the mean climatic sea ice edge position.