Quality assessment of a satellite altimetry data product ... · Received 29 August 2011; revised 24 January 2012; accepted 25 January 2012; published 17 March 2012. [1] Satellite

Quality assessment of a satellite altimetry data productin the Nordic, Barents, and Kara seas

Denis L. Volkov1 and M.-Isabelle Pujol2

Received 29 August 2011; revised 24 January 2012; accepted 25 January 2012; published 17 March 2012.

[1] Satellite altimetry provides high-quality sea surface height data that have beensuccessfully used to study the variability of sea level and surface geostrophic circulationat different spatial and temporal scales. However, the high-latitude regions havetraditionally been avoided due to the persistent sea ice cover. Most of the validation studieshave focused on the areas below the polar circles. In this paper we examine the qualityand performance of a gridded satellite altimetry product in the Nordic, Barents, andKara seas. The altimetric sea level in coastal areas is validated using available tide gaugerecords. We show that at most locations in the Nordic seas the altimetry and tide gaugemeasurements are in a good agreement in terms of the root-mean square differences and theamplitudes and phases of the seasonal cycle. The agreement deteriorates in the shallowareas of the Barents and Kara seas subject to the seasonal presence of sea ice, and where thealtimetry data are contaminated by the residual aliasing of unresolved high-frequencysignals. The comparison of linear trends at the locations of tide gauges revealsdiscrepancies that need to be taken into account when interpreting long-term changes of sealevel in the region. Away from the coast the altimetry data are compared to driftertrajectories, corrected for Ekman currents. The drifter trajectories are found consistent withthe mesoscale variability of the altimetric sea level. This study provides the firstcomprehensive validation of a gridded satellite altimetry data product in the high-latitudeseas.

Citation: Volkov, D. L., and M.-I. Pujol (2012), Quality assessment of a satellite altimetry data product in the Nordic, Barents,and Kara seas, J. Geophys. Res., 117, C03025, doi:10.1029/2011JC007557.

1. Introduction

[2] Sea level is a natural integral indicator of climate vari-ability. Satellite altimetry providing continuous, nearly global,and high-accuracy measurements of sea surface height (SSH)has become the main tool to study the variability of sea level atdifferent spatial and temporal scales [Fu and Cazenave, 2001].However, the use of altimetry data in the polar oceans has beenlimited due to the seasonal presence of sea ice and temporallysparse altimetry measurements.[3] Owing to the 66� orbital inclinations of the former

TOPEX/POSEIDON (TP) and present Jason-1 and OSTM/Jason-2 missions, radar altimeters on these satellites do notacquire measurements above the polar circles. The EuropeanSpace Agency's (ESA's) former ERS-1 and ERS-2 and presentEnvisat satellites with 98� inclination orbits have sampled thepolar oceans up to latitudes of 82� starting from July 1991. TheU.S. Navy Geosat Follow-On (GFO) satellite sampling extendsto 72� latitude. The National Aeronautics and Space

Administration's (NASA's) ICESat satellite was providing thesea surface topography measurements up to 86�N from 2003 to2010, based on which Kwok and Morison [2011] constructedthe mean dynamic topography over the ice-covered areas of theArctic Ocean. The ESA’s Cryosat-2 altimetry satellite waslaunched on 8 April 2010 with an inclination of 92�. TheFrench-Indian SARAL/AltiKa mission is scheduled for launchin early 2012 and will fly on ERS/Envisat orbit to insure thecontinuity of altimetry observations in the long-term.[4] Although much attention is being paid to polar regions at

the present time, most of satellite altimetry data products in thepolar oceans are based on measurements by one particularsatellite mission at a time, while at lower latitudes the combi-nation of simultaneous measurements from several satelliteshas provided a better representation of the mesoscale variability[Le Traon et al., 1998; Le Traon and Dibarboure, 1999; Ducetet al., 2000; Pascual et al., 2006]. In this study we assess thedata that above the polar circle originate either from ERS-1/2 orEnvisat satellites. These satellites have flown on the same orbitthus providing the longest continuous record. Although ICESatdata are not part of the product we use, we acknowledge thatthe combination of ICESat data and, when sufficiently longrecords become available, Cryosat-2 data with ERS-1/2 andEnvisat measurements may improve the quality of futurealtimetry products. For example, recently Farrell et al. [2012]

1Joint Institute for Regional Earth System Science and Engineering,University of California, Los Angeles, California, USA.

2Space Oceanography Division, Collecte Localisation Satellites,Ramonville Saint-Agne, France.

Copyright 2012 by the American Geophysical Union.0148-0227/12/2011JC007557

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, C03025, doi:10.1029/2011JC007557, 2012

C03025 1 of 18

http://dx.doi.org/10.1029/2011JC007557

combined ICESat and Envisat altimetry data to construct thehigh-resolution mean sea surface of the Arctic Ocean and usedit to derive mean dynamic topography.[5] ERS-1/2 and Envisat have flown on a 35-day repeat

cycle orbit meaning that signals at one particular locationalong a satellite track with periods shorter than 70 days are notresolved. The unresolved signals, mainly tides, the barotropicvariability of sea level forced by high-frequency winds, andpart of the mesoscale variability, alias into lower frequenciesand corrupt altimetry measurements [Fukumori et al., 1998;Stammer et al., 2000]. The tidal aliasing is routinely correctedfor by using tidal models that assimilate altimetry data[Le Provost, 2001]. The aliasing due to the high-frequency bar-otropic variability is largely suppressed by subtracting model-simulated high-frequency variability [Carrere and Lyard, 2003;Volkov et al., 2007]. Nevertheless, it is well known that ERS-1/2and Envisat measurements are not optimal for recovering oceantides [Andersen, 1999, 1994]. The dynamic atmospheric cor-rection applied to altimetry data reduces aliasing from signalswith periods less than 20 days, which is optimal for TOPEX/POSEIDON, Jason-1, and Jason-2/OSTMmissions, but not forERS-1/2, Envisat, and GFO satellites.[6] The above considerations indicate that the quality of

satellite altimetry products in high-latitude regions remainsuncertain. Tide gauges have routinely been used to validatealtimetry measurements [Mitchum, 1998; Volkov et al., 2007;Pascual et al., 2009; Vinogradov and Ponte, 2010]. However,none of the similar validation studies known to the authors hasfocused on the northern subpolar regions. Recent advances inthe processing of satellite altimetry geophysical data records[Dorandeu et al., 2010; Faugere et al., 2006; Faugere et al.,2010] are making available more SSH measurements in high-latitude areas for the periods free of sea ice. In this paper wepresent a comparison between a state-of-the-art gridded satellitealtimetry product and available tide gauge and surface driftermeasurements in the Nordic (Greenland, Icelandic, and Nor-wegian), Barents, and Kara seas. The domain of our studycovers an area from 40�W to 100�E and from 60�N to 80�N(Figure 1). We thus aim to complement previous studies byextending the validation of altimetry data further north andprovide the first objective assessment of the quality and per-formance of satellite altimetry in this important region.

[7] The choice of the regional domain for this study is dic-tated by the availability of observational records (discussed inthe next section) and its particular importance in the regionaland global climate system. The Nordic seas connect theAtlantic and the Arctic oceans and play a prominent role in theEarth’s climate system. This region is particularly important fordeep-water formation and the meridional overturning circula-tion. Here, warm Atlantic waters lose heat to atmosphere,subduct to intermediate depths, and transport residual heatfurther into the Arctic Ocean. Multidecadal variability in thetemperature of the Atlantic Water core affecting the upperlayers of the Arctic Ocean has been reported by Polyakov et al.[2003, 2005]. As a result of global warming, long-term declinein the Arctic sea-ice cover has been observed over the last30 years [Comiso et al., 2008] with a record minimum seen in2007 [Stroeve et al., 2008]. Because of this, the study region, inparticular, the Kara Sea is experiencing an increase in openwater area during summer-early fall and thus more altimeterobservations are becoming available.

2. Data and Methods

2.1. Altimetry Data

[8] In this paper we assess the quality of the maps of sealevel anomaly (MSLA) from 14 October 1992 to 1 December2010, produced by SSALTO/DUACS and distributed byAVISO with support from Centre National d’Etudes Spaciales(CNES; http://www.aviso.oceanobs.com/duacs/). We use thestate-of-the-art delayed-time global “reference” MSLA prod-uct, generated by merging SSH measurements from twosatellites at a time: TP/ERS-1/2, Jason-1/Envisat, and Jason-2/Envisat. This means that SSH data over most of the domain ofour study (above 66�N) originates from either the ERS-1/2 orEnvisat mission.[9] Prior to the production of MSLA, the along-track SSH

data are corrected for instrumental noise, orbit determinationerror, atmospheric attenuation (wet and dry tropospheric andionospheric effects), sea state bias, and tidal influence. A globalcrossover adjustment using the TP/Jason-1/Jason-2 orbit as areference is performed to correct for ERS-1/2 and Envisat orbiterrors [Le Traon and Ogor, 1998]. The adjustment uses cubicspline functions that reduce SSH differences at crossovers.Le Traon et al. [1998] and Ducet et al. [2000] have

Figure 1. Map of the study region with bottom topography and the locations of tide gauges.

VOLKOV AND PUJOL: ALTIMETRY DATA IN HIGH-LATITUDE SEAS C03025C03025

2 of 18

demonstrated the efficiency of this method in the open ocean.A mean profile calculated for the 1993–1999 period is sub-tracted from SSH measurements to obtain sea level anomalies(SLA).[10] A dynamic atmospheric correction is applied to account

for the dynamic response of the ocean to atmospheric pressureand wind forcing [Carrere and Lyard, 2003]. This correctioncombines the high frequencies of the Modèle d’Onde deGravité à 2 Dimensions (MOG2D) barotropic model [Lynchand Gray, 1979], forced by the ERA-Interim pressure andwind reanalysis data provided by the European Centre forMedium-Range Weather Forecasts (ECMWF), with the lowfrequencies of the inverted barometer correction. This signifi-cantly reduces the aliasing of the high-frequency sea levelvariability in altimetry data, especially in coastal regions[Carrere and Lyard, 2003; Volkov et al., 2007].[11] The mapping procedure to produce MSLA relies on

optimal interpolation with realistic correlation functions[Le Traon et al., 1998, 2003]. The optimal interpolationscheme is designed to reduce long wavelength errors.These errors are seen as a bias between neighboring tracks,induced by geographically correlated residual errors (e.g.,residual orbit error, instrumental errors) and residualaliasing from large-scale high-frequency signals (tides andhigh-frequency response to atmospheric forcing). Maps areproduced every week on a 1/3� Mercator projection grid.Thus there are 947 maps for the period of study (14 October1992 to 1 December 2010).[12] The maps of absolute dynamic topography (MADT) are

obtained by adding the MSLA referenced to the 1993–1999time period to the MDT_CNES-CLS09 mean dynamic topog-raphy (MDT) product for the 1993–1999 period, based on

4.5 years of GRACE data, and 15 years of altimetry and in situ(hydrography and drifter) measurements [Rio et al., 2011]. Analternative MDT can also be used, e.g., DNSC08 MDT byAndersen and Knudsen [2009]. The MADT (MSLA+MDT)are used to compute the absolute geostrophic velocity compo-nents (Ug and Vg) that will be used for comparison with surfacedrifter velocities: Ug = �(g/f )(∂h/∂y) and Vg = (g/f )(∂h/∂x),where g is gravity, f is the Coriolis parameter, and h is the SSH.[13] Typical mapping errors as a percentage of signal vari-

ance given by the optimal interpolation scheme are displayed inFigure 2a (for details on mapping error estimation, see LeTraon et al. [1998] and Ducet et al. [2000]). The errors arerelatively small (typically below 10% of the signal variance)south of 66�N due to the combination of two satellites. TheERS-1/2 and Envisat error patterns are more complex thanthose of TP, Jason-1, and Jason-2/OSTM, showing both spaceand time variability [Ducet et al., 2000]. North of 66�N, wherethe sampling is carried out by one satellite at a time, errorestimation fields exhibit a typical “diamond structure” withmaxima occurring at the centers of intertrack spaces and oftenreaching 40–50% of the signal variance (Figure 2a). Themapping error is large along the eastern coast of Greenland, inthe Fram Strait, northeast of Spitsbergen, and in the eastern partof the Kara Sea. On the other hand, the mapping error in thewestern and central parts of the Kara Sea is rather low, gener-ally below 20%.[14] Because of the seasonal presence of sea ice, the altim-

etry data in the study region have gaps. The ice flag in ERS/Envisat data is set if one of the following criteria is true: (1) thenumber of valid 20 Hz data is less than 17, (2) the absolutevalue of the difference between the radiometer and theECMWF model wet tropospheric correction is greater than

Figure 2. (a) Mapping error field (percentage of the signal variance) on 28 August 2008 given by theoptimal interpolation of Jason-2/OSTM and Envisat measurements. (b) Portion (%) of missing weeklyaltimetry records.


3 of 18

10 cm, and (3) the waveform peakiness parameter [Laxon,1994] exceeds a 2.0 threshold (for details on ice detection,see Faugere et al. [2003]).[15] The portion of missing data relative to the total number

of maps is displayed in Figure 2b. There are either no missingdata or missing data constitutes less than 10% in the areas thatare free of sea ice all year round. In these areas some data aremissing in January–March 1994 when the ERS-1 satellite wasflying on a 3-day repeat orbit for calibration and sea iceobservation. The percentage of missing data due to sea iceexceeds 20% along the eastern coast of Greenland and in thenortheastern part of the Barents Sea. In the Kara Sea the per-centage of missing data increases from about 20% in thesouthwest to about 80% in the northeast, so that over most partsof the sea only SSH data from 3 to 5 summer-autumn months(July through November) are available.

2.2. Tide Gauge Data

[16] The tide gauge data, used in this study, have been dis-tributed by the Permanent Service for Mean Sea Level(PSMSL, www.psmsl.org). We used the monthly averaged sealevel records from 23 tide gauges located along the Eurasiancoast and on islands (Figure 1). Many tide gauges in the studyregion were closed in 1990s and only a few remained opera-tional through 1990s and 2000s [Proshutisky et al., 2007]. Thelist of the tide gauges with the record intervals used in thisstudy and the number of gaps in the monthly records aresummarized in Table 1. The records from the majority of tidegauges have substantial gaps. Several tide gauges (Bugrino,Malye Karmakuly, Morzhovaia, Isachenko, Uedinenia, andRusskii) have no or a very few records overlapping withaltimetry measurements. In order to validate the seasonal cycleof the altimetry SSH at these locations, we considered longertide gauge records starting from January 1958 (January 1965for Bugrino).

[17] The monthly tide gauge records were corrected for theinverted barometer effect. The inverted barometer correction(IB) is given by IB = (Pa � Pref)/rg, where Pa is the sea levelpressure (SLP), Pref is the SLP averaged over the entire ocean,r is the seawater density, and g is gravity. A change of Pa by1 mbar corresponds to approximately 1 cm change in sea level.For the tide gauges with the record intervals (shown in Table 1)starting in January 1992 we used the monthly averages ofERA-Interim SLP (available from January 1989 onwards),while for those with the record intervals starting in January1958 (January 1965 for Bugrino) we used the monthly avera-ges of ERA-40 SLP (available from September 1957 to August2002). The average reference pressure Pref is time varying witha standard deviation of about 0.5 mbar in the ERA-Interim and0.6 mbar in the ERA-40 data. Because we used monthlyaveraged tide gauge data, the high-frequency sea level fluc-tuations (tides, storm surges, etc.) had been filtered out.

2.3. Glacial Isostatic Adjustment

[18] To account for the vertical crustal motions due to post-glacial rebound, the tide gauge records were also corrected forglacial isostatic adjustment (GIA) using the ICE5G_VM4_L90GIA model by R. Peltier obtained from the PSMSL (www.psmsl.org/train_and_info/geo_signals/gia/peltier) (for detailsabout previous versions, see Peltier [1998, 2004]). This modelprovides the present-day rate of change of relative sea level,that is the sea surface relative to a nearby benchmark connectedto the solid Earth (Figure 3). The tide gauge records are thuscorrected by subtracting a linear trend with a slope inferredfrom the GIA model. Although Scandinavia is experiencing arather strong postglacial rebound accompanied with a decreasein relative sea level at the maximum rate of about 1 mm/yearcentered in the Gulf of Bothnia in the Baltic Sea, it appears thatthe absolute value of the relative sea level change at most tidegauges in the Norwegian, Barents, and Kara seas is an order ofmagnitude smaller (Figure 3).

Table 1. The List of Tide Gauges, Their Locations, Record Intervals Used in This Study, and the Number of Gaps in Monthly Data

Tide Gauge Station Location

Coordinates

Record Interval Used in the Study Number of Gaps in Monthly DataLongitude (�E) Latitude (�N)

Torshavn Norwegian Sea 353.23 62.02 Jan 1992 to Dec 2006 26Lerwick North Sea 358.87 60.15 Jan 1992 to Dec 2009 48Kristiansund Norwegian Sea 7.73 63.12 Jan 1992 to Dec 2009 1Rorvik Norwegian Sea 11.25 64.87 Jan 1992 to Dec 2009 0Andenes Norwegian Sea 16.15 69.32 Jan 1992 to Dec 2009 0Hammerfest Barents Sea 23.67 70.67 Jan 1992 to Dec 2009 2Honningsvag Barents Sea 25.98 70.98 Jan 1992 to Dec 2009 0Barentsburg Greenland Sea 14.25 78.07 Jan 1992 to Dec 2009 35Vardo Barents Sea 31.1 70.33 Jan 1992 to Dec 2009 13Murmansk Barents Sea 33.05 68.97 Jan 1992 to Dec 2009 14Bugrino Barents Sea 49.33 68.8 Jan 1965 to Dec 1990 7Malye Karmakuly Barents Sea 52.7 72.37 Jan 1958 to Dec 1980 18Zhelania Kara Sea 68.55 76.95 Jan 1992 to Feb 1996 12Bolvanskii Nos Kara Sea 59.08 70.45 Jan 1992 to Oct 1993 0Amderma Kara Sea 61.7 69.75 Jan 1992 to Dec 2009 13Morzhovaya Kara Sea 67.58 71.42 Jan 1958 to Sep 1994 7Dikson Kara Sea 80.4 73.5 Jan 1958 to Feb 1997 6Izvestia CIK Kara Sea 82.95 75.95 Jan 1992 to Dec 2009 0Isachenko Kara Sea 89.2 77.15 Jan 1958 to Nov 1993 8Uedinenia Kara Sea 82.2 77.5 Jan 1958 to Mar 1995 5Russkii Kara Sea 96.43 77.17 Jan 1958 to Jul 1989 28Vise Kara Sea 76.98 79.5 Jan 1992 to Dec 2009 20Golomianyi Kara Sea 90.62 79.55 Jan 1992 to Jul 2009 11


4 of 18

[19] It should be noted that ideally satellite altimetry datamust also be corrected for GIA. Crustal deformations due topostglacial rebound are increasing the volume of the oceanbasins [Tamisiea and Mitrovica, 2011]. Assuming that thevolume of water remains the same, the global average SSHrelative to a reference ellipsoid (as measured by space-bornealtimeters) must decrease. The associated change of the glob-ally averaged SSH is �0.3 mm/year [Peltier, 2001]. Recently,Tamisiea and Mitrovica [2011] derived a close value of�0.25 mm/year. This means that this correction must be sub-tracted from the altimetric globally averaged trend of SSH toaccount for crustal deformations over the ocean.[20] When analyzing local (regional) measurements of SSH,

the application of the globally averaged GIA correction is notappropriate, because the impact of GIA on the ocean is notuniform due to a nonuniform associated change in gravity. Thegeographical distribution of the GIA correction over the ocean,presented by Tamisiea and Mitrovica [2011] (their Figure 3b),shows that the absolute value of the GIA impact on SSH alongthe Norwegian and Russian coasts is close to zero and does notexceed 0.1 mm/year. Unfortunately, these data are not availableto us, and therefore we do not correct altimetry data for GIA.As will be shown later in the manuscript, the potential GIAimpact on the local SSH is very small and absolutely not suf-ficient to explain the differences in SSH trends inferred fromtide gauges and altimetry.

2.4. Surface Drifter Data

[21] Satellite-tracked drifters have proven to provide valu-able observations for validating altimetry velocities in the opensea [e.g., Ducet et al., 2000; Pascual et al., 2006, 2009]. Thedrifter data are provided by the Global Drifter Program DataAssembly Center of the Atlantic Oceanographic and Meteo-rological Laboratory (http://www.aoml.noaa.gov/phod/dac/).The raw data are optimally interpolated to uniform 6-h inter-vals. The zonal and meridional components of velocity arecalculated via centered finite differencing over 1/2-day dis-placements. The high-frequency ageostrophic phenomena,such as inertial oscillations, tidal currents, internal waves,coastal upwelling, cyclostropic waves, and others, are reducedby applying a 3-day low-pass filter [Rio and Hernandez, 2004].To obtain the near-surface geostrophic currents, the wind-driven ageostrophic flow (Ekman currents) must be subtractedfrom the drifter velocities. The estimation of Ekman currents is

performed by fitting a two-parameter (angle and amplitude)model to wind stress data and high-frequency ageostrophiccurrents derived from drifting buoys (for details, see Rio andHernandez [2003]).[22] We used information from 252 drifters that were present

in the study region from January 1993 to December 2010.Displayed in Figure 4 are the trajectories of these drifters andthe density of drifter data in 1� � 1� bins. One can see thatmost of drifter records, sometimes exceeding 500 per 1� � 1�bin, are concentrated southwest of Iceland and in the Norwe-gian Sea (Figure 4b). In these areas the drifter trajectories arehighly variable. Poleward from about 70�N some drifters werecarried toward the Fram Strait, while some were advected intothe Barents Sea (Figure 4a). From the first group, a few driftersreached the interior of the Greenland Sea around 5�W and75�N. The East Greenland Current carried one drifter all theway south toward the Denmark Strait. From the second group,relatively large number of drifters reached the interior of theBarents Sea. The others followed the coastal current with onedrifter entering the Kara Sea through the Karskie Vorota Strait.

3. Quality Assessment

3.1. Comparison With Tide Gauge Records

[23] For comparison with tide gauge records the data fromMSLA were bilinearly interpolated to the locations of the tidegauges. The errors that might be introduced by the interpolationare small because the spatial scales resolved by the MSLA arelarger than the grid resolution [e.g., Pascual et al., 2009]. Forexample, the standard deviation of the difference between SLAaveraged over 0.5� � 0.5� around Andenes tide gauge andSLA interpolated at Andenes is 0.3 cm, which is much smallerthan about 7 cm standard deviation of SLA near Andenes.[24] The time means, computed over the overlapping

records, were subtracted from both the altimetry and tide gaugetime series. The time series of four characteristic tide gaugerecords corrected for IB and GIA and the corresponding timeseries of altimetry data interpolated to the locations of tidegauges are presented in Figure 5. The degree of consistencybetween the altimetry and tide gauge data is summarized inTable 2, which presents the standard deviations of the timeseries and root-mean-square (RMS) differences between therecords. The comparison is presented only for the tide gaugesthat have overlapping records with altimetry. In general, the

Figure 3. Numerical prediction of the present-day impact of glacial isostatic adjustment on relative sealevel measured by tide gauges (mm/year) from the ICE5G_VM4_L90 model by R. Peltier and the loca-tions of tide gauges used in this study.


5 of 18

application of the IB correction reduces the standard deviationof the tide gauge data bringing it closer to the standard devia-tion of the altimetry data, and it reduces the RMS differencebetween the tide gauge and altimetry data. The static effect ofthe atmospheric pressure (IB effect) is thus an important con-tributor to the sea level variability in the region and needs to becorrected for before carrying out the comparison with altimetrydata. The impact of the GIA correction on the RMS differenceis much smaller, on the order of 1 mm. At some locations, theapplication of the GIA correction reduces, at some increases,while at the majority of locations it has no impact on the RMSdifference.[25] The lowest RMS difference between the tide gauge and

altimetry data down to about 2 cm is estimated at Torshavn andLerwick, where the altimetry data are based on the combinationof two satellites. The tide gauge and altimetry records in theNorwegian and Barents seas are also close (Table 2). Thenumber of overlapping monthly records here is sufficientlylarge (around 200). The average RMS difference at the Nor-wegian tide gauges (Andenes, Hammerfest, Honningsvag,Vardo), located in the areas where either ERS-1/2 or Envisatdata are available, is about 3 cm. Displayed in Figure 5 (topleft) are the time series of the tide gauge and altimetry sea levelat Andenes manifesting a good agreement between the datasets in terms of the seasonal and interannual variability. TheRMS difference between them is 3.1 cm when GIA correctionis applied and 2.9 cm without GIA.[26] The agreement between the altimetry and tide gauge

data deteriorates in the Barents and Kara seas (Table 2). Here,

at most tide gauges the RMS difference is between 5 and 7 cm.At Murmansk tide gauge the RMS difference exceeds 6 cm.This is expected because the tide gauge is located inland in afjord and altimetry data near Murmansk are more representa-tive of the nearby open sea region. At Zhelania and Dikson tidegauges the agreement is poor with the RMS difference ofnearly 11 cm. It should be noted that at several locations(Bolvanskii Nos, Morzhovaia, Isachenko, Uedinenia) the rela-tively low RMS difference is calculated from a small numberof overlaps. Neglecting errors in the tide gauge records, thegreater RMS difference in the Kara Sea is probably because thealtimetry signal is contaminated by the presence of sea ice, it ismore influenced by land, and the residual aliasing from high-frequency signals is large in the shallow areas.[27] The reliability of records obtained from two tide gauges,

used in this study, is suspect. Namely, by analyzing the timeseries of the Barentsburg tide gauge (Figure 5), one can notice acontrast between the measurements prior and after 2002. Priorto 2002, the tide gauge record poorly agrees with altimetry.During this time it is considerably more variable than after2002 and has gaps. If only the period after 2002 is considered,the RMS difference between the tide gauge and altimetry dataat Barentsburg is reduced to 4 cm as opposed to 8 cm for the1992–2009 period (Table 2). On the basis of this comparisonand on the station documentation provided by the PSMSL(www.psmsl.org), we conclude that the quality of the Bare-ntsburg tide gauge data between 1992 and 2002 is uncertain.Although Amderma is one of the oldest and key observatorieson the Russian Arctic coast, its time series from 1992 to 2009

Figure 4. (a) The trajectories of surface drifters present in the region from January 1993 to December2010. (b) The number of drifter records in 1� � 1� bins.


6 of 18

Figure 5. Sea level anomaly (centimeters) from the records at four tide gauges (black curves) and fromthe MSLA interpolated to the locations of the tide gauges (red curves).

Table 2. Comparison Between the Monthly Tide Gauge Records and Altimeter MSLA Interpolated to the Location of Tide Gauges inTerms of Standard Deviations and Root Mean Square Differencesa

Tide Gauge Station Number of Overlapping Records

Standard Deviation of SLA RMS of SLA Differences

ALT TG TG-IB TG-IB-GIA ALT-TG ALT-(TG-IB) ALT-(TG-IB-GIA)

Torshavn 144 5.5 8.1 6.0 5.8 6.9 2.3 2.2Lerwick 157 6.2 9.6 6.9 6.8 6.2 1.9 1.9Kristiansund 204 5.9 12.5 8.9 8.9 8.8 3.9 3.9Rorvik 205 7.0 12.5 8.5 8.5 8.1 3.3 3.1Andenes 200 7.0 12.4 7.5 7.4 8.0 2.9 3.1Hammerfest 198 6.1 12.4 7.7 7.7 8.2 3.2 3.2Honningsvag 200 6.5 11.9 7.0 7.0 7.4 2.9 2.8Barentsburg 168 4.3 11.8 9.5 9.6 11.3 8.0 8.0Barentsburgb 91 4.8 8.7 6.6 6.5 7.3 4.0 4.0Vardo 187 6.8 11.8 7.5 7.4 7.6 3.1 3.1Murmansk 186 6.2 12.0 9.2 9.2 9.1 6.6 6.6Zhelania 17 4.6 13.6 11.8 11.8 11.8 10.9 10.8Bolvanskii Nos 12 7.4 15.0 8.8 8.8 6.7 4.7 4.7Amderma 142 8.0 14.2 11.7 11.8 9.2 7.4 7.5Amdermac 142 8.0 13.0 10.1 10.1 8.0 5.8 5.8Morzhovaia 6 10.4 21.8 22.7 22.7 6.4 6.0 6.0Dikson 16 11.6 16.0 13.8 13.8 11.1 10.7 10.7Izvestia CIK 85 8.1 12.8 10.1 10.1 6.8 5.3 5.3Isachenko 5 7.1 14.3 10.8 10.8 4.4 4.6 4.6Uedinenia 8 6.4 16.0 12.6 12.5 8.8 6.3 6.3Vise 81 4.3 12.7 10.6 10.6 9.1 6.8 6.9Golomianyi 60 4.2 9.7 6.3 6.3 7.7 6.2 6.3

aThe comparison is conducted only for the tide gauges that have overlapping records with altimetry. The IB correction for the tide gauge data wascalculated using the ERA-Interim SLP. The time mean sea level, computed over the overlapping records, was subtracted from both the tide gauge andaltimeter data. Abbreviations: ALT is altimeter data, TG is raw tide gauge records, (TG-IB) is tide gauge records corrected for IB, (TG-IB-GIA) is tidegauge records corrected for IB and GIA.

bOnly data from 2002 to 2009 inclusive are considered.cThe 1992–2010 linear trend has been removed from the tide gauge data.


7 of 18

even after applying the GIA correction have a strong positivelinear trend of 1.2 cm/year (Figure 5). The trend estimated fromavailable altimetry data at Amderma is only 0.28 cm/year. Wedoubt that the trend observed at the tide gauge is realistic andwill discuss this issue in more detail in a section dedicated tolinear trends. By removing the 1.2 cm/year linear trend fromthe Amderma tide gauge record, the RMS of the differencebetween the tide gauge and altimetry data is reduced from7.5 cm to 5.8 cm.[28] As has been mentioned earlier, residual aliasing of tidal

and other high-frequency signals in altimetry data remains oneof the concerns for the time series analysis [Volkov et al.,2007]. The ERS-1/2 and Envisat altimeters due to their 35-day repeat cycles have problematic alias periods and they arenot suitable for recovering shallow-water tides [Andersen,1994, 1999]. In addition, this imposes the aliasing of K1 andP1 tidal constituents at 365.24 days period, which is not sepa-rable from the seasonal cycle [Andersen, 1999; Le Provost,2001]. Unfortunately, because of substantial data gaps, we arenot able to analyze the spectrum of the tide gauge and altimetryrecords at all locations in order to understand to what degreethe aliasing affects the altimetry data. Therefore, here we esti-mated the power spectral densities (PSD) only at three stationsthat have continuous records (no gaps) with sufficient duration:Andenes in the Norwegian Sea and Honningsvag and Bugrinoin the Barents Sea. At Andenes and Honningsvag, the tidegauge (altimetry) records are from January 1992 to December2009 (from April 1994 to November 2010). Since there are nooverlapping records at Bugrino, we used the tide gauge datafrom the earlier June 1977 to May 1986 time interval and thealtimetry data from April 2001 to February 2010.[29] The PSD estimates for both the tide gauge (black

curves) and the altimetry data (red curves) are presented inFigure 6. The strongest signal is due to the seasonal cycle(frequency = 1 cpy (cycle/year)). Note that the PSD estimatesof the seasonal cycle from the tide gauge and altimetry data arealmost of the same power at Andenes and Honningsvag(Figures 6a and 6b). This means that the seasonal cycle at theselocations is well observed by altimetry and it is not stronglycorrupted by possible aliasing of K1 and P1 tidal constituents.At Bugrino, the seasonal cycle observed by altimetry has morepower than the one observed by the tide gauge (Figure 6c).Assuming that the seasonal cycle has been stationary over theperiod of observations, the PSD offset at the annual frequencyis probably due to the quality of altimetry measurements at thislocation. The Bugrino tide gauge is situated in the southern partof Kolguyev Island on the shelf of the Barents Sea (Figure 1). Itis a shallow region subject to shallow water tides and stormsurges. The amplitude of the most energetic diurnal constituentK1 here exceeds 10 cm [e.g., Killett et al., 2011; Kowalik andProshutinsky, 1993]. Thus, in the southeastern part of theBarents Sea the aliasing from the residual tidal and other high-frequency signals can represent a bigger problem than for thetwo former locations.[30] At higher frequencies the altimetry data appear to

underestimate the variability. The PSD of the altimetry data issystematically lower than the PSD of the tide gauge data. Thetide gauge records at the three locations have PSD maxima at asemiannual periodicity (2 cpy). The altimetry data also show asemiannual periodicity at Bugrino, but it is not as well observedat Andenes and Honningsvag.

3.2. Assessment of the Seasonal Variability

[31] As we have shown, the seasonal cycle is the major sealevel signal in the monthly tide gauge and altimetry records.Therefore, comparing the seasonal variability of sea levelobserved by tide gauges and altimeters can be regarded as adiagnostic of altimetry products. In this section we estimate theamplitude and phase of the seasonal cycle by fitting a harmonicfunction (in a least squares sense) with an annual frequency tothe monthly averages of the MSLA and tide gauge records.[32] The amplitude and phase of the seasonal cycle estimated

from the MSLA (Figures 7a and 7b) are possibly influenced bybottom topography. In the Norwegian and Greenland seas,there are two open-sea seasonal amplitude maxima located inthe deep parts (depths exceed 3000 m) of the Lofoten Basinaround 2�E and 70�N (amplitudes up to 10 cm) and GreenlandBasin around 5�W and 75�N (amplitudes about 7 cm). Theamplitude is also large, sometimes exceeding 10 cm, along theNorwegian and Russian coasts. In the Barents Sea, the maxi-mum amplitude is concentrated in the southeastern part overthe area with depths less than 250 m. The maximum amplitudein the Kara Sea is observed over the coastal area with depthsless than 50 m, however, because of gaps in data, the accuracyof these estimates is uncertain and will be discussed below.[33] Over the deep parts of the Greenland, Icelandic, and

Norwegian seas, the annual maximum of the sea level isobserved in September (Figure 7b). In the shallower areas ofthe study region, where depths are generally less than 1000 m,the maximum sea level occurs later, from October to Decem-ber. It is observed in October over the Greenland–Iceland andFaeroe–Iceland ridges, along the continental shelf margin, andin the northern parts of the Barents and Kara seas. In thesouthern part of the Barents Sea and over most of the Kara Seathe annual maximum sea level occurs in November. TheDecember maximum is observed over the shallow southeasternparts of the Barents Sea and along the western coast of NovayaZemlya archipelago (depths less than 250 m), and in thesouthwestern and eastern parts of the Kara Sea.[34] Mork and Skagseth [2005] have analyzed the seasonal

variability of SSH in the Nordic seas and found that the stericeffect explains only 40% of the total SSH variability and it isprimarily caused by the net surface heat fluxes. The seasonalvariations in bottom pressure (ocean mass component) havebeen found dominant. In our study, for the first time we alsopresent the altimetry-based spatial distribution of the seasonalamplitude and phase in the Barents and Kara seas. We havenoted that over these relatively shallow seas the seasonalmaximum SSH occurs 1–3 months later than over the deepNordic seas (Figure 7b). In the shallow Barents and Kara seasthe steric effect is expected to be smaller than in the deepNordic seas. Therefore it is reasonable to assume that the bulkof the seasonal SSH variability in these seas, corrected for thestatic effect of atmospheric pressure, is barotropic and wind-induced. River runoff, which is maximum in June, is influentialnear estuaries [Proshutisky et al., 2007]. It is likely that wind isalso responsible for the phase difference between the deep andshallow parts of the study region, but this suggestion needs tobe confirmed in a future study.[35] While the estimates of the seasonal amplitude and phase

are robust in most areas of the Nordic seas, they may containsignificant errors in the Kara Sea and in some areas of theGreenland and Barents seas where sea ice is present in winter.


8 of 18

Figure 6. Power spectral density (cm2/cpy) of records from three tide gauges in the Norwegian andBarents seas (black curves) and the MSLA interpolated at the locations of tide gauges (red curves). Thealtimetry (tide gauge) records used at (a) Andenes and (b) Honningsvag are from April 1994 to November2010 (January 1992 to December 2009). The altimetry (tide gauge) records used at (c) Bugrino are fromApril 2001 to February 2010 (June 1977 to May 1986).


9 of 18

For example, in the areas where the annual maximum and/orminimum sea level occurs at the time when sea ice is present, aleast squares fit of the annual harmonic function can be a poorapproximation of the seasonal cycle. Substantial river dischargeinto the Kara Sea may also complicate accurate estimation ofthe seasonal cycle.[36] In order to estimate the associated error, we used the

output of a high-resolution global-ocean and sea-ice data syn-thesis at eddy-permitting resolution from the Estimating theCirculation and Climate of the Ocean, Phase II (ECCO2)project (www.ecco2.org). An ECCO2 data synthesis isobtained by least squares fit of a global full-depth-ocean andsea-ice configuration of the Massachusetts Institute of Tech-nology general circulation model (for details, see Marshallet al. [1997] and Menemenlis et al. [2005]). Although, themodel adequately reproduces the amplitude of the seasonalcycle, the phase of the seasonal cycle (not shown) significantlydiffers from Figure 7b and does not seem to be realistic.Because the error estimate can be sensitive to the phase

information, we combined the high frequencies (periods lessthan 4 months) of the ECCO2 model sea level, hhf

m (x, y, t), withthe seasonal cycle estimated from the altimetry data, hsc

a (x, y, t):hc(x, y, t) = hhf

m (x, y, t) + hsca (x, y, t). We then imposed the same

gaps, as in the altimetry data, to the combined data set,hc(x, y, t), to generate ~hc x; y; tð Þ. After that, the seasonalamplitude and phase were estimated from both hc(x, y, t) and~hc x; y; tð Þ, and used to reconstruct the seasonal cycle fieldshscrec(x, y, t) and ~hrecsc x; y; tð Þ. The error ɛ(x, y) was then calcu-lated as the percentage of the RMS difference between hsc

rec and~hrecsc to the amplitude of the seasonal cycle (asc) computed from

hc: ɛ x; yð Þ ¼ 100%asc

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1N

PNi¼1

hrecsc � ~hrecsc

� �2r, where N is the num-

ber of maps.[37] The obtained error estimate is presented in Figure 7c.

Over most areas of the Greenland, Icelandic, Norwegian, andBarents seas, where the number of gaps in data is small(Figure 2b), the error is generally below 5%. In the northern

Figure 7. (a) Amplitude (centimeters) of the seasonal cycle. (b) Phase of the seasonal cycle (month of theannual maximum). (c) Error (%) on the determination of the seasonal cycle. The bottom topography (inFigures 7a and 7b) is shown for 50, 250, 500, 1000, 2000, and 3000 m.


10 of 18

part of the Barents Sea the error reaches 20%. As expected, thebiggest error is estimated in the Kara Sea. It reaches 50% insome locations in the eastern part of the Kara Sea. Neverthe-less, over large areas in the western and central Kara Sea theerror is rather low and generally does not exceed 20%.[38] The amplitudes and phases of the seasonal cycle esti-

mated from the tide gauge records and from the MSLA inter-polated to the locations of the tide gauges are compared inTable 3. At the locations, where there are no or a very few(with gaps) tide gauge records after 1992 (at Bugrino, MalyeKarmakuly, Morzhovaia, Dikson, Isachenko, Uedinenia, and

Russkii), we also used records prior to 1992 (Table 1),assuming that the seasonal cycle was stationary over the periodof observations. One can see that the overall agreement is sat-isfactory: the amplitudes are similar and the phases are thesame at most locations (along the Norwegian coast and atseveral islands in the Kara Sea); at other locations the phasedifference does not exceed 1 month.[39] It should be noted that the semiannual signal is also

present in both the altimetry and tide gauge records (notshown). The amplitude of the semiannual signal is relativelysmall, generally below 2 cm in the Norwegian and Barentsseas. However, the semiannual signal is significant at somelocations in the Kara Sea due to the impact of summer melt andriver runoff. For example, at the Dikson tide gauge, theamplitude of the semiannual signal (5.5 cm) exceeds theamplitude of the seasonal cycle (4 cm). The amplitude ofthe semiannual signal at the Izvestia CIK tide gauge (4 cm)is already about twice smaller than the amplitude of the sea-sonal cycle (9 cm). The semiannual signal can thus influencethe estimates of the seasonal cycle from available altimetrydata, in particular, near estuaries. This is probably why theworst agreement between the amplitudes of the seasonal cycleis observed at Dikson, while the other locations in the Kara Seaare not heavily affected (Table 3). Unfortunately, the data gapsdue to seasonal ice cover make the accurate estimation of thesemiannual signal in the altimetry data problematic. Therefore,we do not compare the estimates of the semiannual signal inthis study.

3.3. Comparison of Linear Trends

[40] One of the major climate-related concerns for the soci-ety is the rise of sea level. Comparisons of linear trends derivedfrom two independent observational systems, i.e., tide gaugesand altimetry, may shed some light on the quality of observa-tions. The differences between the trends can be regarded aserror bars. The MSLA were used to estimate the linear trends inthe Nordic and Barents seas from December 1992 to Novem-ber 2010 (Figure 8). In the Kara Sea, because of sea ice, thelinear trends in altimetry data were estimated using Julythrough September (JAS) averages of MSLA (Figure 9).

Table 3. Comparison of Amplitudes and Phases of the SeasonalCycle Estimated From the Altimetry and Tide Gauge Dataa

Tide Gauge Station

Altimetry Tide Gauge

A φ A φ

Torshavn 6 Sep 6 SepLerwick 7 Nov 8 NovKristiansund 7 Nov 10 NovRorvik 8 Nov 9 NovAndenes 8 Nov 8 NovHammerfest 7 Nov 8 NovHonningsvag 7 Nov 8 NovBarentsburg 4 Sep 6 OctVardo 7 Nov 8 NovMurmansk 7 Nov 7 OctBugrinob 8 Dec 6 NovMalye Karmakulyb 6 Nov 4 OctZhelania 4 Nov 5 DecBolvanskii Nos 7 Dec 11 DecAmderma 10 Nov 11 OctMorzhovayab 11 Nov 9 OctDiksonb 7 Oct 4 SepIzvestia CIK 9 Oct 9 NovIsachenkob 6 Nov 5 NovUedineniab 7 Nov 6 NovRusskiib 5 Nov 3 NovVise 4 Oct 6 OctGolomianyi 2 Nov 3 Nov

aAmplitude is measured in centimeters and phases are months of theannual maxima.

bThe amplitude and phase of the seasonal cycle in tide gauge data arecomputed using the time interval shown in Table 1.

Figure 8. Linear trend of sea level (mm/year) from December 1992 to November 2010 measured by sat-ellite altimetry in the Nordic and Barents seas. Red crosses indicate the locations of tide gauges, whoserecords were used for comparison.


11 of 18

[41] As one can see, the Norwegian and Greenland seas areexperiencing the biggest sea level rise in the region reaching8 mm/year and 5 mm/year, respectively (Figure 8). The sealevel is also rising in the southern part of the Barents Sea with alittle change in the northern areas. In the Kara Sea, the changeof the JAS sea level is much smaller and it is nonuniform,meaning that the error on the determination of trends possiblyhas a greater relative impact (Figure 9).[42] A comparison of trends obtained from tide gauge

records and MSLA interpolated to the locations of tide gauges

is presented in Table 4. In the Kara Sea, the linear trends in thetide gauge data were estimated using both the entire recordsfrom January 1993 and the JAS averages. Only the tide gaugeswith relatively long records in 1993–2009 (at least 15 years) areconsidered. Although the trends agree in sign, the differencebetween them at most locations is substantial. The best agree-ment is observed at Kristiansund, Hammerfest, and Hon-ningsvag tide gauges in the Norwegian Sea, where thedifference between trends does not exceed 1 mm/year. Theworst agreement is observed at Barentsburg and Amderma.

Figure 9. Linear trend of July through September (JAS) average sea level (mm/year) from JAS 1993 toJAS 2010 measured by satellite altimetry in the Kara Sea. Red crosses indicate the locations of tidegauges, whose records were used for comparison.

Table 4. Comparison of Linear Trends Estimated From the Altimetry and Tide Gauge Dataa

Tide Gauge Station Time Interval Used Trend in Altimetry Data Trend in Tide Gauge Data

Torshavn Jan 1993 to Dec 2007 3.8 6.5Kristiansund Jan 1993 to Dec 2009 3.5 3.5Rorvik Jan 1993 to Dec 2009 4.4 1.8Andenes Jan 1993 to Dec 2009 4.3 2.3Hammerfest Jan 1993 to Dec 2009 4.0 3.2Honningsvag Jan 1993 to Dec 2009 4.0 2.9Barentsburg Jan 1993 to Dec 2009 1.8 4.8Barentsburg Jan 2002 to Dec 2009 �0.7 �4.1Vardo Jan 1993 to Dec 2009 4.1 2.5Murmansk Jan 1993 to Dec 2009 3.8 5.2Amderma Jan 1993 to Dec 2009 – 12.8Amdermab JAS 1993 to JAS 2009 3.3 13.9Izvestia CIK Jan 1993 to Dec 2009 – 3.1Izvestia CIKb JAS 1993 to JAS 2008 1.7 4.2Vise Jan 1993 to Dec 2008 – 0Vise2 JAS 1996 to JAS 2007 0 2.2Golomianyi Jan 1993 to Dec 2008 – 1.6Golomianyib JAS 1993 to JAS 2008 1.6 5.7

aLinear trends are measured in millimeters per year.bThe linear trend is computed using July through September averages.


12 of 18

Although after 2002 the RMS difference between the tidegauge and altimetry data at Barentsburg is reduced to 4 cm(Table 2, Figure 4), the trends from January 2002 to December2009 differ almost by a factor of 6. The time series of theAmderma tide gauge from January 1993 to December 2009(corrected for IB and GIA) have a strong positive linear trendof approximately 12 mm/year (Figure 4). The trend estimatedfor JAS averages from 1993 to 2009 is about 14 mm/year(Table 4). The corresponding trend obtained using JAS avera-ges of MSLA at Amderma is approximately 3 mm/year.Although the latter trend is seasonally biased due to winter icecover, it is closer to the 1.5 mm/year trend estimated at theAmderma tide gauge from 1954 to 1989 [see Proshutinsky etal., 2004]. It seems to be unlikely that in 1992–2010 thetrend increased by almost an order of magnitude, and it raisesconcerns about the quality of the tide gauge record. The localGIA correction may contain errors and/or the geodetic bench-marks may have been unstable during this period due to pos-sible changes in permafrost.[43] In the Kara Sea, the linear trends, estimated from the

tide gauges located on islands (Izvestia CIK, Vise, andGolomianyi), are not uniform (Table 4). Sea level was rising atthe rate of about 2–3 mm/year in the vicinity of mainland(Izvestia CIK and Golomianyi), but it did not change in theopen sea (Vise). The trends inferred from the JAS averages ofthe altimetry data also show a sea level rise of about 2 mm/yearat Izvestia CIK and Golomianyi and no change at Vise. The

JAS trends obtained from the tide gauge data significantlydiffer from those estimated using the entire records. This meansthat the JAS trends may be not representative.[44] The comparison of trends suggests that while the sea

level rise by 3–4 mm/year in the Norwegian Sea is probablyrobust, the present rates of sea level change in the north of theGreenland Sea, and in the coastal areas of the Barents and Karaseas still remain uncertain. It should be noted that the omissionof the GIA correction in altimetry data does not explain thedifferences in trends. The absolute value of the GIA impact onsea level of 0.1 mm/year, documented by Tamisiea andMitrovica [2011], is at least an order of magnitude smallerthan any observed difference (Table 4).

3.4. Comparison With Drifter Trajectories

[45] In this section we investigate the ability of the high-latitude MADT and MSLA to adequately resolve the oceancirculation, in particular, the mesoscale eddy variability. Themapping errors over the areas sampled by one satellite at a timeare not homogeneous (Figure 2a) and this should be taken intoaccount in the interpretation. It has been documented that atlower latitudes the combination of TP and ERS-1/2 reduces themapping error by a factor of 4 compared to the TP alone and bya factor of 2 compared to ERS-1/2 alone [Le Traon andMorrow, 2001]. Because the computation of the geostrophicvelocities involves the differentiation of SLA, the U and V

Figure 10. The time mean U and V components of the (top) drifter and (middle) altimetry velocitiesaveraged over 1� � 1� bins, and (bottom) the difference between them. The drifter velocities were cor-rected for Ekman currents. The altimetry data were interpolated at the trajectories of drifters.


13 of 18

mapping errors are 2 to 4 times larger than the SLA mappingerror [Le Traon and Morrow, 2001].[46] In this section we compare the drifter velocities cor-

rected for Ekman currents with the absolute geostrophicvelocities derived from the MADT by interpolating the altim-eter velocity fields to the positions and times of the drifter data.Displayed in Figure 10 are the drifter and altimetry geostrophicvelocity observations averaged over all measurements fallinginto 1� � 1� bins and the difference between them. There is agood agreement in spatial patterns of the velocities (Figure 10,top and middle). One can observe the Norwegian Current thatsplits into the North Cape Current that enters the Barents Seaand the West Spitsbergen Current that heads toward the FramStrait following the bottom topography. The East GreenlandCurrent is also resolved by both the altimetry and the drifterobservations. In some places altimetry appears to underestimatethe velocities (Figure 10, bottom). In particular, the altimetry Vvelocity is about 10–15 cm/s smaller than the drifter V velocityalong the East Greenland Current. The altimetry U velocity isfrom 5 to 15 cm/s smaller than the drifter U velocity in thesouthern part of the Barents Sea and in the Karskiye VorotaStrait. The altimetry V velocity is about 5 cm/s smaller than thedrifter V velocity along the West Spitsbergen Current.[47] The RMS differences between the drifter (corrected for

Ekman currents) and altimetry velocities are presented inFigure 11. Depending on local ocean dynamics the values inmost areas range between 7 and 15 cm/s. In the areas of lowvariability the RMS differences usually do not exceed 7 cm/s.

In energetic regions, such as the Lofoten Basin in the Norwe-gian Sea (5�W–10�E, 68�N–72�N) and the East GreenlandCurrent, the RMS differences may reach 20 cm/s. These esti-mates are comparable to those estimated by Pascual et al.[2006, 2009] at lower latitudes.[48] To visualize how well the altimetry data along a par-

ticular drifter trajectory matches the drifter measurements, wepresent the time series of the altimetry and drifter records alongfour characteristic trajectories (Figure 12). One drifter(SD32531) floated all the way from Iceland to the Barents Sea.Another drifter (SD35368) entered the eddy-rich Lofoten Basinand circulated around the basin for a while, eventually con-tinuing to follow the Norwegian Current. The third drifter(SD63663) was carried by the current from the Lofoten Basineastward, entered the Kara Sea, and reached the eastern coast ofthe northern island of Novaya Zemlya archipelago. The forthdrifter (SD63945) was circulating around an anticyclonic eddyin the Lofoten Basin for several weeks and then movednorthward to the Fram Strait being carried by the West Spits-bergen Current. The time series of the U and V components(Figure 12) demonstrate a rather good agreement between thealtimetry data and the drifters. Even the short-term variability isadequately resolved by the altimetry measurements. However,the drifter data manifest stronger variability, probably becausethey are not as much filtered as the altimetry measurements.The agreement between the time series deteriorates in thesoutheastern part of the Barents Sea, in the Kara Sea, and in theGreenland Sea (drifters SD63663 and SD63945). The RMS

Figure 11. The RMS of the differences (cm/s) between the altimetry and drifter (a) U and (b) V velocitycomponents computed over 1� � 1� bins. The altimetry data were interpolated at the trajectories of drif-ters. The drifter velocities were corrected for Ekman currents.


14 of 18

differences for the four considered drifters are on the order of10 cm/s.[49] Displayed in Figure 13 are the trajectories of the drifter

SD63945 from 1 August 2009 to 29 August 2009 and thecorresponding altimetry MADT and geostrophic velocities inthe region between 1.5�W–6�W and 69.2�N–70.6�N. Duringthis time interval the drifter was located in the eddy-rich part ofthe Norwegian Sea (Lofoten Basin) and, therefore, represents agood example of the relationship between the drifter trajectoryand SSH field. The drifter was initially entrained in an anticy-clonic eddy located at about 2.6�W and 69.9�W. This eddy iswell resolved by the MADT. It was stationary until 12 Augustand then it started to slowly propagate northwestward. On26 August the center of the eddy moved to approximately2�W and 70.2�N. During the considered time interval anotheranticyclonic eddy entered the region from the east. It is inter-esting to note how the second eddy eventually entrained thedrifter on 22 August. This particular example demonstrates agood agreement between the altimetry and drifter data. Over-all, the MADT and MSLA in the Nordic seas seem to provideenough detail to adequately resolve the mesoscale variability.

4. Summary and Conclusions

[50] The use of altimetry data in the subpolar and polarregions has been challenging due to the persistent sea ice cover.Up to the date, most of the validation studies have been carriedout for the low and midlatitudes. Therefore the quality of thehigh-latitude altimetry records has largely remained uncertain.Recent advances in the processing of geophysical data records,obtained from altimetry satellites, have made available moreSSH data at high latitudes. In the work, presented in this paper,we have carried out the first validation of the state-of-the-artgridded satellite altimetry product (distributed by AVISO,www.aviso.oceanobs.com) in the Nordic and Kara seas. Wehave assessed the quality of the product by comparing altimetrymeasurements with those collected by tide gauges and surfacedrifters.[51] The comparison with tide gauge records corrected for

the IB and GIA has demonstrated that the altimetry data arerobust in most coastal areas of the study region. The IB cor-rection has a prominent effect on the variability of the tidegauge time series, bringing them closer to the corresponding

Figure 12. The along-trajectory drifter (color curves) and altimetry (black curves) U and V velocities forfour drifters. The color in the drifter trajectories and velocities is used to highlight different time intervals.The drifter velocities were corrected for Ekman currents. The RMS of the difference between the altimetryand drifter velocity components is shown.


15 of 18

SLA time series. The impact of the GIA correction on the RMSdifference between the altimetry and tide gauge records isfound small.[52] In the Norwegian and Barents seas, the RMS difference

between the altimetry and tide gauge measurements is gener-ally about 3 cm. This estimate is similar or lower than thosereported in the earlier studies focused on lower latitudes [e.g.,Pascual et al., 2009]. The discrepancy between the altimetryand tide gauge records increases in the southeast Barents Seaand in the Kara Sea. At the majority of tide gauges located inthese seas the RMS difference is about 5–7 cm. Because sat-ellite altimetry is more accurate away from the coast, asexpected, the best match in the Kara Sea is usually observed atthe tide gauges located on small islands.[53] We found that the quality of two tide gauge records is

suspect. Namely, the quality of the Barentsburg tide gaugebetween 1992 and 2002 is probably poor, which is confirmedby the available station documentation. The Amderma tide

gauge records from 1992 to 2010 demonstrate an unrealisticrate of sea level rise. Therefore one needs to be cautious wheninterpreting the tide gauge time series in the region.[54] The PSD estimates obtained from the altimetry and tide

gauge time series at two locations in the Norwegian (Andenes)and Barents (Honningsvag) seas have shown almost the samepower at an annual frequency. This indicates that the seasonalcycle at these locations is well observed by altimetry and it isnot strongly corrupted by possible aliasing from the residualtidal and high frequency signals. On the other hand, at Bugrino,located in the shallow part of the southeastern Barents Sea, thealtimetry seasonal cycle has a greater power than the tide gaugeseasonal cycle. This suggests that the altimetry data in thisshallow region can be contaminated by the aliased signals.[55] The seasonal cycle is the major signal in the monthly

altimetry and tide gauge records. The annual maximum sealevel variability with amplitude from about 7 to over 10 cm isobserved over the deep parts of the Norwegian and Greenland

Figure 13. The MADTs (centimeters), the altimetry geostrophic velocities, and the trajectory of the sur-face drifter SD63945 from 1 August 2009 to 29 August 2009 in the Norwegian Sea. The trajectories,corresponding to the weeks approximately centered at the times of the MADT are highlighted by magenta.


16 of 18

seas and along the Norwegian and Russian coasts. The annualmaximum sea level over most areas of the Greenland, Icelan-dic, and Norwegian seas occurs mainly in September. In thenorthern part of the Barents Sea and in the northwestern andcentral parts of the Kara Seas the annual maximum is observedin October, while in the southern part of the Barents Sea and inthe western and eastern parts of the Kara Sea it is observed inNovember–December. We have compared the amplitudes andphases of the seasonal cycle of sea level estimated from thealtimetry and tide gauge data. The amplitudes and phases arefound similar at most locations. The difference between theannual phases does not exceed 1 month. Analyzing the sea-sonal cycle in the altimetry and tide gauge records at lowerlatitudes, Vinogradov and Ponte [2010] have reported the samedifference. Along the Norwegian coast the annual phases areidentical.[56] The comparison of linear trends has revealed substantial

differences at most locations that can be regarded as a measureof uncertainty. On the basis of the results of this study, weconclude that the sea level rise by 3–4 mm/year along theNorwegian coast is probably robust with an uncertainty ofabout 1–2 mm/year. On the other hand, the present rates of sealevel change in the north of the Greenland Sea and in thecoastal areas of the Barents and Kara seas remain largelyuncertain because the signal-to-noise ratio is often less than 1.[57] The comparison with surface drifter data has provided

support for the realistic representation of surface circulation bysatellite altimetry in the region. The general patterns of circu-lation observed by altimeters and drifters are similar. In mostenergetic regions the short-term variability is adequatelyresolved by the altimetry measurements. However, altimetryappears to somewhat underestimate the surface velocities.This is probably because the drifter data are not as muchfiltered as the altimetry data. The RMS differences betweenthe drifter and altimetry velocities are comparable toearlier estimates for the World Ocean. They range fromabout 7 cm/s in the low variability regions to about 15–20 cm/s in energetic regions. We have demonstrated thatdrifter trajectories are in a good agreement with altimetricSSH and its mesoscale variability.[58] In summary, in this paper we have provided a rather

strong validation of a modern satellite altimetry gridded prod-uct in the Nordic seas. This means that this product can besuccessfully used to study the variability of sea level and sur-face circulation in the region. However, some limitations haveto be taken into account. We have shown that in some coastalareas of the Kara Sea the agreement between the altimetry andtide gauge data is relatively poor. This can be partly due to theresidual aliasing of the high-frequency sea level variability inaltimetry data and needs to be further investigated. Caution isalso required when analyzing linear trends obtained from boththe coastal altimetry and tide gauge data. Regional (ArcticOcean) tide models, as opposed to the global models used inthe AVISOMSLA products, may possibly better correct for thetidal aliasing in altimetry data. Their potential use needs to beexplored in the near future.

[59] Acknowledgments. This work was carried out at Jet PropulsionLaboratory, California Institute of Technology, within the framework of theproject “Investigating the variability of sea level in the sub-Arctic and Arc-tic seas,” sponsored by the NASA Physical Oceanography program (awardNNX11AE27G). The ERA-Interim and ERA-40 sea level pressure data areprovided by the European Centre for Medium Range Weather Forecast

(www.ecmwf.int). The authors thank C.K. Shum, an anonymous reviewer,and the editor A. Proshutinsky for their comments and suggestions.

ReferencesAndersen, O. B. (1994), Ocean tides in the northern North Atlantic andadjacent seas from ERS-1 altimetry, J. Geophys. Res., 99(C11),22,557–22,573, doi:10.1029/94JC01433.

Andersen, O. B. (1999), Shallow water tides in the northwest Europeanshelf region from TOPEX/POSEIDON altimetry, J. Geophys. Res.,104(C4), 7729–7741, doi:10.1029/1998JC900112.

Andersen, O. B., and P. Knudsen (2009), DNSC08 mean sea surface andmean dynamic topography models, J. Geophys. Res., 114, C11001,doi:10.1029/2008JC005179.

Carrere, L., and F. Lyard (2003), Modeling the barotropic response of theglobal ocean to atmospheric wind and pressure forcing: Comparisonswith observations, Geophys. Res. Lett., 30(6), 1275, doi:10.1029/2002GL016473.

Comiso, J. C., C. L. Parkinson, R. Gersten, and L. Stock (2008), Acceler-ated decline in the Arctic sea ice cover, Geophys. Res. Lett., 35,L01703, doi:10.1029/2007GL031972.

Dorandeu, J., G. Dibarboure, M. Pujol, and E. Bronner (2010), AltimeterL3–L4 products evolution during the last 17 years, Eos Trans. AGU,91(26), Ocean Sci. Meet. Suppl., Abstract PO45S-17.

Ducet, N., P.-Y. Le Traon, and G. Reverdin (2000), Global high resolutionmapping of ocean circulation from TOPEX/POSEIDON and ERS-1/2,J. Geophys. Res., 105(C8), 19,477–19,498, doi:10.1029/2000JC900063.

Farrell, S. L., D. C. McAdoo, S. W. Laxon, H. J. Zwally, D. Yi, A. Ridout,and K. Giles (2012), Mean dynamic topography of the Arctic Ocean,Geophys. Res. Lett., 39, L01601, doi:10.1029/2011GL050052.

Faugere, Y., F. Mertz, and J. Dorandeu (2003), Envisat validation and crosscalibration activities during the verification phase, Rep. 16243/02/NL/FFWP6, Eur. Space Res. and Technol. Cent., Noordwijk, Netherlands.[Available at http://earth.eo.esa.int/pcs/envisat/ra2/articles/ENVISAT_Verif_Phase_CLS.pdf]

Faugere, Y., J. Dorandeu, F. Lefevre, N. Picot, and P. Femenias (2006),Envisat ocean altimetry performance assessment and cross-calibration,Sensors, 6, 100–130, doi:10.3390/s6030100.

Faugere, Y., M. Ablain, E. Bronner, and N. Picot (2010), Recent and com-ing improvements of the performance of open ocean altimetry products,Eos Trans. AGU, 91(26), Ocean Sci. Meet. Suppl., Abstract PO45S-19.

Fu, L.-L., and A. Cazenave (2001), Satellite Altimetry and Earth Sciences:A Handbook of Techniques and Applications, Int. Geophys. Ser., vol. 69,Academic, San Diego, Calif.

Fukumori, I., R. Raghunath, and L.-L. Fu (1998), Nature of the global largescale sea level variability in relation to atmospheric forcing: A modelingstudy, J. Geophys. Res., 103(C3), 5493–5512, doi:10.1029/97JC02907.

Killett, B., J. Wahr, S. Desai, D. Yuan, and M. Watkins (2011), ArcticOcean tides from GRACE satellite accelerations, J. Geophys. Res., 116,C11005, doi:10.1029/2011JC007111.

Kowalik, Z., and A. Y. Proshutinsky (1993), Diurnal tides in the ArcticOcean, J. Geophys. Res., 98(C9), 16,449–16,468, doi:10.1029/93JC01363.

Kwok, R., and J. Morison (2011), Dynamic topography of the ice-coveredArctic Ocean from ICESat, Geophys. Res. Lett., 38, L02501,doi:10.1029/2010GL046063.

Laxon, S. (1994), Sea ice extent mapping using the ERS-1 radar altimeter,EARSeL Adv. Remote Sens., 3, 112–116.

Le Provost, C. (2001), Ocean tides, in Satellite Altimetry and EarthSciences: A Handbook of Techniques and Applications, Int. Geophys.Ser., vol. 69, edited by L.-L. Fu and A. Cazenave, pp. 267–304, Aca-demic, San Diego, Calif.

Le Traon, P.-Y., and G. Dibarboure (1999), Mesoscale mapping capabilitiesof multisatellite altimeter missions, J. Atmos. Oceanic Technol., 16,1208–1223, doi:10.1175/1520-0426(1999)016<1208:MMCOMS>2.0.CO;2.

Le Traon, P.-Y., and R. Morrow (2001), Ocean currents and eddies, in Sat-ellite Altimetry and Earth Sciences: A Handbook of Techniques andApplications, edited by L.-L. Fu and A. Cazenave, Int. Geophys. Ser.,vol. 69, pp. 171–216, Elsevier, New York.

Le Traon, P.-Y., and F. Ogor (1998), ERS-1/2 orbit improvement usingTOPEX/POSEIDON: The 2 cm challenge, J. Geophys. Res., 103,8045–8057, doi:10.1029/97JC01917.

Le Traon, P.-Y., F. Nadal, and N. Ducet (1998), An improved mappingmethod of multisatellite altimeter data, J. Atmos. Oceanic Technol., 15,522–534, doi:10.1175/1520-0426(1998)015<0522:AIMMOM>2.0.CO;2.

Le Traon, P.-Y., Y. Faugere, F. Hernandez, J. Dorandeu, F. Mertz, andM. Ablain (2003), Can we merge Geosat Follow-on with TOPEX/POSEIDON and ERS-2 for an improved description of the ocean


17 of 18

circulation?, J. Atmos. Oceanic Technol., 20, 889–895, doi:10.1175/1520-0426(2003)020<0889:CWMGFW>2.0.CO;2.

Lynch, D. R., and W. G. Gray (1979), A wave equation model for finite ele-ment tidal computations, Comput. Fluids, 7, 207–228, doi:10.1016/0045-7930(79)90037-9.

Marshall, J., A. Adcroft, C. Hill, L. Perelman, and C. Heisey (1997), Afinite volume, incompressible Navier-Stokes model for studies of theocean on parallel computers, J. Geophys. Res., 102, 5753–5766,doi:10.1029/96JC02775.

Menemenlis, D., I. Fukumori, and T. Lee (2005), Using Green’s functionsto calibrate an Ocean General Circulation model, Mon. Weather Rev.,133, 1224–1240, doi:10.1175/MWR2912.1.

Mitchum, G. T. (1998), Monitoring the stability of satellite altimeters withtide gauges, J. Atmos. Oceanic Technol., 15, 721–730, doi:10.1175/1520-0426(1998)015<0721:MTSOSA>2.0.CO;2.

Mork, K. A., and O. Skagseth (2005), Annual sea surface height variabilityin the Nordic seas, in The Nordic Seas: An Integrated Perspective, Geo-phys. Monogr. Ser., vol. 158, edited by H. Drange et al., pp. 51–64, AGU,Washington, D. C., doi:10.1029/158GM05.

Pascual, A., Y. Faugere, G. Larnicol, and P.-Y. Le Traon (2006), Improveddescription of the ocean mesoscale variability by combining four satellitealtimeters, Geophys. Res. Lett., 33, L02611, doi:10.1029/2005GL024633.

Pascual, A., C. Boone, G. Larnicol, and P.-Y. Le Traon (2009), On the qual-ity of real-time altimeter gridded fields: Comparison with in situ data, J.Atmos. Oceanic Technol., 26, 556–569, doi:10.1175/2008JTECHO556.1.

Peltier, W. R. (1998), Postglacial variations in the level of the sea: Implica-tions for climate dynamics and solid-earth geophysics, Rev. Geophys.,36(4), 603–689, doi:10.1029/98RG02638.

Peltier, W. R. (2001), Global glacial isostatic adjustment and modern instru-mental records of relative sea level history, in Sea Level Rise: History andConsequences, Int. Geophys. Ser., vol. 75, edited by B. C. Douglas, M. S.Kearney, and S. P. Leatherman, pp. 65–95, Academic, San Diego, Calif.

Peltier, W. R. (2004), Global glacial isostasy and the surface of the Ice-AgeEarth: The ICE-5G(VM2) model and GRACE, Annu. Rev. Earth Planet.Sci., 32, 111–149, doi:10.1146/annurev.earth.32.082503.144359.

Polyakov, I., D. Walsh, I. Dmitrenko, R. L. Colony, and L. A. Timokhov(2003), Arctic Ocean variability derived from historical observations,Geophys. Res. Lett., 30(6), 1298, doi:10.1029/2002GL016441.

Polyakov, I., et al. (2005), One more step toward a warmer Arctic, Geophys.Res. Lett., 32, L17605, doi:10.1029/2005GL023740.

Proshutinsky, A., I. M. Ashik, E. N. Dvorkin, S. Hakkinen, R. A. Krishfield,and W. R. Peltier (2004), Secular sea level change in the Russian sector

of the Arctic Ocean, J. Geophys. Res., 109, C03042, doi:10.1029/2003JC002007.

Proshutinsky, A., I. Ashik, S. Häkkinen, E. Hunke, R. Krishfield, M. Maltrud,W. Maslowski, and J. Zhang (2007), Sea level variability in the Arctic Oceanfrom AOIMP models, J. Geophys. Res., 112, C04S08, doi:10.1029/2006JC003916.

Rio, M.-H., and F. Hernandez (2003), High-frequency response of wind-driven currents measured by drifting buoys and altimetry over the worldocean, J. Geophys. Res., 108(C8), 3283, doi:10.1029/2002JC001655.

Rio, M.-H., and F. Hernandez (2004), A mean dynamic topography com-puted over the world ocean from altimetry, in situ measurements, and ageoid model, J. Geophys. Res., 109, C12032, doi:10.1029/2003JC002226.

Rio, M. H., S. Guinehut, and G. Larnicol (2011), New CNES-CLS09 globalmean dynamic topography computed from the combination of GRACEdata, altimetry, and in situ measurements, J. Geophys. Res., 116,C07018, doi:10.1029/2010JC006505.

Stammer, D., C. Wunsch, and R. M. Ponte (2000), De-aliasing of globalhigh frequency barotropic motions in altimeter observations, Geophys.Res. Lett., 27(8), 1175–1178, doi:10.1029/1999GL011263.

Stroeve, J., M. Serreze, S. Drobot, S. Gearheard, M. Holland, J. Maslanik,W. Meier, and T. Scambos (2008), Arctic sea ice extent plummets in2007, Eos Trans. AGU, 89(2), 13–14, doi:10.1029/2008EO020001.

Tamisiea, M. E., and J. X. Mitrovica (2011), The moving boundaries of sealevel change: Understanding the origins of geographical variability,Oceanography, 24(2), 24–39, doi:10.5670/oceanog.2011.25.

Vinogradov, S. V., and R. M. Ponte (2010), Annual cycle in coastal sealevel from tide gauges and altimetry, J. Geophys. Res., 115, C04021,doi:10.1029/2009JC005767.

Volkov, D. L., G. Larnicol, and J. Dorandeu (2007), Improving the qualityof satellite altimetry data over continental shelves, J. Geophys. Res., 112,C06020, doi:10.1029/2006JC003765.

M.-I. Pujol, Space Oceanography Division, Collecte LocalisationSatellites, 8-10 Rue Hermes, Parc Technologique du Canal, F-31520Ramonville Saint-Agne, France.D. L. Volkov, Jet Propulsion Laboratory, California Institute of

Technology, 4800 Oak Grove Dr., MS 300-323, Pasadena, CA 91109,USA. ([email protected])


18 of 18

Documents

Quality assessment of a satellite altimetry data product ... · Received 29 August 2011; revised 24 January 2012; accepted 25 January 2012; published 17 March 2012. [1] Satellite