Tidal and non-tidal sea level variability along the coastalNova Scotia from satellite altimetry

GUOQI HAN*†, JOHN LODER‡ and BRIAN PETRIE‡

†Northwest Atlantic Fisheries Centre, Fisheries and Oceans Canada,

St John’s, NL, Canada

‡Bedford Institute of Oceanography, Fisheries and Oceans Canada,

Dartmouth, NS, Canada

TOPEX/Poseidon (T/P) sea level data are used to investigate tidal and non-tidal

sea level variability over the Scotian Shelf. Eight major diurnal and semi-diurnal

tidal constituents are derived from the September 1992 to December 2001 T/P

data. The present altimetric tides agree generally with in situ observations and

show an error reduction of over 30� in phase over Han et al.’s [Han, G., Ikeda, M.

and Smith, P.C., 1996, Oceanic tides on the Newfoundland and Scotian Shelves

from TOPEX/POSEIDON altimetry. Atmosphere–Ocean, 34, pp. 589–604] results

for the diurnal tides. De-tided T/P sea level data are used to provide geo-referenced

seasonal-mean sea surface topography during the period September 1992 to July

2002. The T/P data are explored to estimate seasonal-mean sea level along the

Nova Scotia coast. Comparisons with sea level measurements from coastal tide

gauges are carried out to evaluate the potential of altimetry data for coastal sea

level monitoring. Seasonally the sea level is highest along the coast in late autumn/

early winter (associated with the peak of the Nova Scotian Current), and over the

Scotian Slope in summer. On the interannual scale it was low in the early 1990s,

high in the mid 1990s, and close to normal in the late 1990s and early 2000s.

1. Introduction

The Scotian Shelf off Nova Scotia (figure 1) is subject to strong downstream influ-

ences of the Labrador Current from the north and the outflow of the Gulf of St

Lawrence. The influence of the Labrador Current of Arctic origin can continue

equatorward along the upper continental slope and can intrude onto the Scotian

Shelf through cross-shelf channels (Petrie and Drinkwater 1993, Petrie 2007). The

intruding water is a mixture of the Labrador Slope Water and the warmer and more

saline Warm Slope Water offshore (Sutcliffe et al. 1976). Offshore over the ScotianSlope, the sea level and circulation are subject to influences of the Gulf Stream

meanders and rings (e.g. Han 2007).

Earlier studies using water level and hydrographic data suggest that local and non-

local forcings play significant roles in sea level variability on the Scotian Shelf and

Slope. Thompson (1986) examined sea level data off northeast America for the period

from 1950 to 1975 and found that wind stresses and air pressure were only small

contributors to the interannual changes of the coastal sea level. Umoh and Thompson

(1994) showed that interannual temperature variability on the Scotian Shelf could notbe accounted for by the atmospheric heat flux variability. The Nova Scotian Current

*Corresponding author. Email: Guoqi.Han@dfo-mpo.gc.ca

International Journal of Remote SensingISSN 0143-1161 print/ISSN 1366-5901 online # 2010 Government of Canada Crown Copyright

http://www.tandf.co.uk/journalsDOI: 10.1080/01431161.2010.485152

International Journal of Remote Sensing

Vol. 31, Nos. 17–18, September 2010, 4791–4806

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

is usually strong in winter (Anderson and Smith 1989), associated with the large

onshore sea surface slope. The dominant alongshore wind in summer results in coastal

upwelling (Petrie et al. 1987). Steric height variability over the Scotian Shelf was

studied based on the temperature and salinity climatology (e.g. Loder et al. 1997),showing general seasonal increases of 5–10 cm from winter to summer, with an

exception that the winter steric height is higher on the inner Scotian Shelf. Tidal

information was mainly from scattered tide-gauge data (Godin 1980) and hydrody-

namic models (de Margerie and Lank 1986, Han and Loder 2003).

Satellite altimetry has been widely used for studying oceanic tides and mesoscale

ocean circulation in the deep ocean (e.g. Fu and Cazenave 2001). Its synoptic mea-

surements with near global coverage of the world’s ocean make it an essential tool in

ocean and climate research and monitoring. Satellite altimetry provides reliablemeasurements of offshore sea level variations off Atlantic Canada, particularly for

those with large amplitudes such as tides (Han et al. 1996), major ocean currents

(Hakkinen and Rhines 2004), and mesoscale eddies (Han 2004).

Until recently the majority of oceanographic applications of satellite altimetry have

simply excluded coastal and shelf regions. Estimation of associated sea level varia-

tions at coastal sites from altimetry is more challenging and the results may be subject

to larger uncertainty, generally related to data degradation, sparsity, and errors in

various environmental corrections near land. Appropriate re-tracking methods mustbe implemented and evaluated to ensure accuracy. Environmental corrections such as

water vapour need to be improved. Accurate coastal tide models and methods to

integrate them with the global deep ocean tide models for de-tiding are required.

−68 −66 −64 −62 −60 −5841

42

43

44

45

46

47

012

050088

12607

1

10903

3037

020

Longitude°W

Latit

ude°

N

Nova Scotia

Halifax

Scotian Shelf SEC

GeorgesBank

SIB

LCNorth Sydney

YarmouthLunenburg

Canso Harbour

Lockeport

Murphy’s Coveo

o

o

o

o

o

o

Figure 1. Map showing the Scotian Shelf and adjacent oceans. The numerical labelled lines(thin dashed) are the selected T/P ground tracks on which the analysis is performed. The 200-,1000-, 2000-, 3000- and 4000-m isobaths (grey lines) are also shown. LC: Laurentian Channel;SIB: Sable Island Bank; SEC: Shelf-edge current.

4792 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

There are only three (figure 1) long-term tide gauges along the nearly 1000-km

coastline of Nova Scotia and less than a dozen along the approximately 10 000-km

Atlantic (including the Gulf of St Lawrence) coastline of Canada. Therefore, there is

high potential that satellite altimetry can contribute to coastal sea level monitoring,

particularly in remote areas and ones with spatially-varying sea level changes. Thealtimetric data available since the 1990s allow for the extraction of important seasonal

and interannual sea level variations which are of significance to coastline variability,

climate and other societal issues in Atlantic Canada.

In this study we used TOPEX/Poseidon (T/P) altimetry data for the period from

1992 to 2002 to study tidal and non-tidal sea level variability over the Scotian Shelf.

Our objectives are: (1) to provide improved observational estimates of the tidal

variations in sea surface topography over the Scotian Shelf; (2) to provide geo-

referenced seasonal-mean sea surface topography over the Scotian Shelf and adjacentregions by season and year; and (3) to compare seasonal and interannual variations in

coastal sea level from T/P altimetry with those from coastal tide gauges along Nova

Scotia, in order to evaluate the potential of satellite altimetry for coastal sea level

monitoring.

This paper consists of five sections. Section 2 briefly describes the datasets, proces-

sing techniques, and methods for tidal and non-tidal analyses. Tidal results are

presented in section 3. Section 4 discusses non-tidal results including seasonal and

interannual variations. A summary is given in section 5.

2. Methodology

2.1 T/P data processing and analyses

T/P sea-surface height data were obtained from the NASA Pathfinder Project. The

satellite has a nominal repeat cycle of 10 days, and ideally there are 360 observations

at each location. There are ascending (SW–NE) and descending (NW–SE) tracks with

spacing of about 200 km on the Scotian Shelf (figure 1). The along-track resolution is

about 6 km. The data were corrected based primarily on the principles in Benada

(1997) for various atmospheric and oceanographic effects, including:

1. wet troposphere delay measured by the T/P microwave radiometer;2. dry troposphere delay determined from the European Centre for Medium-

Range Weather Forecasts surface pressure model;

3. ionosphere delay based on the dual frequency altimeter measurement;

4. electromagnetic bias (due to ocean wave influences) using 2% of the significant

wave height;

5. solid Earth tide; and

6. inverse barometric effect.

The standard NASA GSFC (Goddard Space Flight Center) precise orbit based on the

Joint Gravity Model-3 (JGM-3) was used to produce the sea surface height data

relative to the T/P reference ellipsoid with equatorial radius of 6378.1363 km and a

flattening coefficient of 1/298.257.

2.1.1 Analysis for tides. The T/P data used for tidal analysis are for the period from

August 1992 to December 2001. They are not corrected for ocean, load, or pole tides

henceforth called the tide-in T/P data. The 10-year T/P data are sufficient to resolve

PORSEC 2008 4793

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

the eight major semi-diurnal (M2, S2, N2, K2) and diurnal (K1, O1, P1, Q1) tidal

constituents well. Little improvement can be achieved with a longer dataset.

A modified response analysis method (Han et al. 1996) was used to extract eight

major tide constituents from the tide-in T/P data. The orthogonal response analysis

method to retrieve tides from satellite altimetry was described and discussed in greatdetail by Cartwright and Ray (1990). In essence, the time-varying portion of the tide-

generating potential for each species is expressed as a sequence of nearly orthogonal

functions. A least squares regression is performed to determine the convolution

coefficients from the observed data. These coefficients are used to calculate the

admittance to the tide-generating potential at a frequency for each species. We also

included the annual cycle in the least squares analysis. As a result, the regression has a

14-parameter description, with two sets of six parameters for the semi-diurnal and

diurnal species and two (cosine and sine) remaining parameters for the annual cycle.The tidal admittance is used to calculate the amplitude and Greenwich phase lag for

each constituent.

2.1.2 Analysis for seasonal-mean anomalies. T/P sea-surface height data forseasonal-mean anomalies are for the period from mid 1992 to May 2002. Ocean,

load and pole tides were all corrected.

An along-track digital filter with a filter width of about 36 km (seven points) is

applied to the altimetric sea surface height data. The results presented are based on the

smoothed height data unless indicated otherwise. A time-mean sea surface was con-

structed from the T/P data. We then calculate the sea surface height anomalies relative

to the mean sea surface. Both the marine geoid and mean oceanic topography are

removed by this procedure.

2.2 Tide-gauge data and processing

Water level data at three permanent tide-gauge sites along the Nova Scotia coast

(North Sydney, Halifax and Yarmouth; see figure 1) were obtained from Marine

Environmental Data Service (MEDS) of Fisheries and Oceans Canada. One-hour

water level records from 1992 to 2002 are quality controlled, and then analysed by a

harmonic analysis method (Foreman and Henry 1989) to extract major tidal consti-tuents. The 2002 data from North Sydney were suspect and not included in this

analysis. The 2002 data for the two other sites received no or only minimal quality

control by MEDS but appeared to be of good quality.

A linear trend was removed from the tide-gauge data, to approximately account for

the effect of the ground motion, mainly due to post-glacial rebound. Note that any

long-term sea level trend is also removed by the de-trending. This approximation is

reasonable since the tidal and non-tidal variabilities of interest in this paper have

much shorter timescales than those of the Glacial Isostatic Adjustment (Peltier 2004).The detided sea level data are filtered using the Godin 25_24_24 filter (a boxcar

filter with 24, 24 and 25 h windows successively applied; Godin 1972) and averaged to

produce monthly mean sea levels. The inverse barometer effect is corrected using local

atmospheric pressure data at nearby meteorological stations of Environment Canada.

Tidal harmonic constituents at other coastal tide gauge stations and bottom

pressure gauge sites (see figure 2) are also used for evaluating the altimetric tides.

The in situ data are obtained from Han et al. (1996).

4794 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

3. Major tidal constituents

We have derived eight major semi-diurnal (M2, S2, N2, K2) and diurnal (K1, O1, P1,

Q1) tidal constituents from the tide-in T/P data using the modified response analysis

method outlined in section 2.1.1. The major features are consistent with previous

results (e.g. Godin 1980, Han et al. 1996). For example, the semidiurnal has small

amplitude and phase change from the east to west, except for the southwestern

Scotian Shelf. The diurnal features amphidromic points near the eastern Scotian

Shelf. See Han et al. (1996) for the general patterns of semi-diurnal (M2, S2, N2)

and diurnal (K1, O1) co-tidal charts.We have compared the present altimetry estimates with in situ observations from

coastal tide gauges at the three permanent tide-gauge stations (table 1). For Halifax

and North Sydney, the altimetric tides on the ascending and descending tracks are

interpolated/extrapolated onto the tide-gauge site using an optimal linear interpola-

tion method with a correlation scale of 3� and 3� in the longitudinal and latitudinal

directions. For Yarmouth, a linear extrapolation scheme is used, with the altimetric

tidal data on Track 033 only.

There is good agreement between tide-gauge data and T/P estimates at Yarmouth.The largest amplitude (phase) difference is 0.023 m for K2, the smallest semi-diurnal

constituent (36� for Q1, the smallest diurnal constituent) with root mean square

(RMS) differences over eight constituents of 0.012 m (9�). This suggests the T/P

altimetry is able to provide accurate information for coastal tides.

−68 −66 −64 −62 −60 −5841

42

43

44

45

46

47

012

050088

126

071

10903

3

037

020

Longitude°W

Latit

ude°

N oooo

oo

o

oo

o

o

ooo

o

o

oo

Figure 2. Locations of coastal tide gauges and pelagic bottom pressure gauges where T/Pestimates are compared with in situ observations and with Han et al.’s (1996) results (seetable 2). Dashed lines are satellite ground tracks.

PORSEC 2008 4795

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

At Halifax, the altimetric M2 amplitude is lower than the tide-gauge result by about

0.05 m, while the S2 and N2 amplitude differences are within 0.001 m. The altimetric

K1 and O1 amplitudes underestimate by about 0.01 and 0.02 m, respectively. The

overall RMS amplitude difference for all eight constituents is 0.021 m. There is

generally good agreement in phase between altimetric and tide-gauge data for the

semi-diurnal constituents, K1 and P1; however, the phases for O1 and Q1 differsubstantially (by 33 and 153�).

At North Sydney, the altimetric tide amplitude is significantly higher than the tide-

gauge magnitude by 0.14 m for M2, lower by 0.03 m for S2 and N2 and higher by 0.04

and 0.05 m for K1 and O1. The overall RMS amplitude difference for all eight

constituents is 0.06 m. There are also larger phase differences at this site than at

Yarmouth or Halifax. As we can see, the nearest altimetric data points are far away

from North Sydney (figure 1). For the diurnals, the presence of nearby amphidromic

points further limits the accuracy of extrapolation of altimetric tides to the coastalsite. A dynamic approach that assimilates along-track altimetric tides into a coastal

tidal model may be useful (e.g. Han et al. 2000, Egbert and Erofeeva 2002).

It is usually more difficult to observe weak constituents such as K2, P1, and Q1 due

to low signal-to-noise ratio, which were excluded in Han et al. (1996). The present

results from longer altimetric time series show approximate agreement with tide-

gauge estimates, except for Q1 at Halifax (table 1). Nevertheless, it is not surprising

to see large discrepancy for the Q1 phase, since the Q1 amplitude is nearly zero.

The altimetric tidal results are further evaluated against other coastal tide-gaugedata and pelagic bottom pressure gauge measurements (see figure 2 for location) off

Nova Scotia for the five major constituents. A comparison of the present altimetric

tide solutions with Han et al.’s (1996) altimetric estimates indicates overall improve-

ment for the diurnal tides (table 2), with an error reduction of over 30� in phase. The

nearly doubled record length is now sufficient to get the aliased K1 tide (which has an

aliasing period of 172.3 days) separated from the semi-annual cycle, which improves

the estimation of the tidal admittance at each frequency of the diurnal species.

The comparison also indicates little improvement for the semi-diurnal constituents.

Table 1. Harmonic constants from T/P estimates and tide-gauge observations (in parentheses)for eight major semi-diurnal and diurnal constituents at North Sydney, Halifax and Yarmouth.

Amplitudes are in metres and Greenwich phase lags are in degrees.

Yarmouth Halifax North Sydney

Amplitude Phase Amplitude Phase Amplitude Phase(m) (�) (m) (�) (m) (�)

M2 1.658 (1.652) 62 (61) 0.584 (0.633) 349 (351) 0.510 (0.368) 353 (349)S2 0.272 (0.287) 96 (89) 0.139 (0.139) 17 (23) 0.109 (0.143) 37 (24)N2 0.351 (0.336) 34 (35) 0.139 (0.14) 333 (327) 0.076 (0.106) 330 (320)K2 0.082 (0.059) 98 (90) 0.035 (0.039) 19 (16) 0.03 (0.035) 32 (27)K1 0.14 (0.137) 183 (183) 0.088 (0.109) 123 (123) 0.077 (0.043) 325 (27)O1 0.108 (0.124) 165 (169) 0.018 (0.04) 127 (94) 0.082 (0.029) 287 (332)P1 0.048 (0.050) 182 (183) 0.024 (0.032) 122 (124) 0.026 (0.015) 315 (29)Q1 0.016 (0.019) 160 (124) 0.014 (0.006) 128 (335) 0.011 (0.012) 250 (286)

4796 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

4. Non-tidal sea levels

4.1 Altimetric mean surface topography

Mean sea levels at T/P track locations were calculated over the period from 1992 to

2002 relative to the T/P ellipsoid. We also bi-linearly interpolated the 20 by 20 GSFC00

mean surface topography (Wang 2001) onto the track locations. The differencesbetween the T/P and GSFC00 mean sea levels (the former minus the latter) were

computed location by location. We then mapped the differences onto a 0.2� by 0.2�

grid using an optimal linear interpolation method with correlation scales of 3� and 3�

in the longitudinal and latitudinal directions. The mapped differences were added

onto the GSFC00 mean sea surface. As expected, the T/P mean sea surface topogra-

phy (figure 3) is predominated by the spatial pattern in the marine geoid.

Interpretation for mean sea surface topography due to ocean dynamics are subject

Figure 3. Mean sea surface height (colour band, m) relative to the T/P ellipsoid from the T/Pdata for the period from August 1992 to June 2002. The 200-m isobath (grey lines) and T/Ptracks (white lines) are also displayed.

Table 2. The RMS differences between the altimetric tides and in situ observations. The valuesin parentheses are from Han et al.’s (1996) solutions. See figure 2 for observation locations.

Coastal Pelagic

Amplitude Phase Amplitude Phase(m) (�) (m) (�)

M2 0.031 (0.040) 11 (7) 0.012 (0.008) 4 (4)S2 0.017 (0.031) 13 (15) 0.015 (0.016) 10 (9)N2 0.014 (0.015) 17 (17) 0.012 (0.009) 8 (8)K1 0.014 (0.017) 15 (48) 0.013 (0.014) 10 (17)O1 0.019 (0.031) 29 (69) 0.018 (0.017) 21 (28)

PORSEC 2008 4797

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

to uncertainty in the geoid. For coastal and shelf oceans, this is true even with recent

state-of-the-art mean dynamic topography (Foreman et al. 2008). The geoid from

GOCE (Gravity field and steady-state Ocean Circulation Explorer Mission)

(Drinkwater et al. 2007) is expected to significantly enhance the mean dynamical

topography on the shelf scale.

4.2 Anomalies of altimetric seasonal-mean sea-surface topography

Long-term, seasonal-mean, sea-surface height anomalies averaged from 1992 to 2002

were constructed using the optimal linear interpolation method with correlation

scales of 3� and 3� in the longitudinal and latitudinal directions (figure 4). We can

see the coastal sea level west of Halifax is highest in autumn and lowest in spring,

which can be well accounted for by the steric height variability. Over the open shelf,the sea level is higher in summer and autumn and lower in spring and winter, mainly in

response to seasonal atmospheric heating and cooling (Han et al. 2002). The seasonal

variability is enhanced over the continental slope, associated with shifting and/or

meandering of the Gulf Stream and rings pinched off from it. The annual sea level

range varies from 10 cm near the coast to 20 cm over the lower slope. Relatively low

sea level nearshore in summer could be associated with coastal upwelling under

alongshore wind forcing (Petrie et al. 1987) and relatively high sea level off the

southwestern Nova Scotia in winter could be associated with the passage of thepeak Nova Scotian Current (Anderson and Smith 1989, Han et al. 1997).

Figure 4. Long-term seasonal-mean sea surface height anomalies (colour band, m) from T/Pdata. The 200-m isobath (grey lines) and T/P tracks (white lines) are also displayed.

4798 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

Using the optimal linear interpolation method, we also constructed seasonal-

mean sea-surface topography anomalies by season for selected years to show inter-

annual variability (figure 5). The sea level over the Scotian Shelf was low in 1994,

high in 1997, and in between in 2000, consistent with Han’s (2002) earlier analysis

from the descending-track data only. Overall, the interannual variability over theScotian Slope is nearly out of phase with that along the coast and over the Scotian

Shelf. Han (2007) showed that the mean sea level profile was lowest near the 3000-m

isobath, rising towards the Scotian Shelf and coast associated with the southwest-

ward shelf edge current and the Nova Scotian Current and towards offshore

associated with the northeastward slope current and the Gulf Stream. They illu-

strated that a southward shift of the Gulf Stream position would result in a sea level

rise near the coast and a sea level fall over the lower continental slope, and vice versa.

Gulf Stream rings can also be seen over the Scotian Slope, e.g. in spring and autumnof 1994 (figure 5(a)).

4.3 Seasonal and interannual sea level anomalies at tide-gauge stations and offshore

An important aspect of this paper is to evaluate accuracy of altimetric estimates of

coastal sea levels. Seasonal-mean sea level estimates at selected tide-gauge sites along

the Nova Scotia coast, extrapolated from the altimetry data using the aforementionedoptimal linear interpolation method, indicate significant interannual variations at all

the sites (figure 6). In general, the sea level was low in the early 1990s, high in mid

1990s and close to normal in the late 1990s and early 2000s. The range is about 10 cm.

Altimetric estimates are compared with tide-gauge measurements at the three

permanent tide-gauge sites (North Sydney, Halifax and Yarmouth) (figure 7). The

correlation coefficients are 0.53–0.64. The RMS sea level differences between records

are 3.5–4.5 cm, comparable to the RMS values of individual records. The removal of

the annual and semiannual cycles substantially enhances the agreement at Halifax(the RMS difference: 1.5 cm), and but not at the other two sites (the RMS difference:

2.3, 1.9 cm) (figure 8). At all three sites, the satellite sensor data appear to lag the

coastal tide gauge data for the period 1992–1997. Furthermore, at North Sydney,

there also appears to be some systematic difference between the tide gauge and

satellite observations, i.e. for the first half of the record the difference (satellite

minus gauge) is generally negative, and for the second half mostly positive. The lowest

correlation and largest RMS difference at this location are presumably associated

with the poorest T/P coverage. A dynamic approach that assimilates altimetricobservations into a coastal circulation model would be useful for improving the fit.

To provide further information on the along-shelf and cross-shelf structure of sea

level anomalies, four along-track bands were selected based on bathymetry: (i) coast to

mid-shelf, (ii) mid-shelf to 200 m, (iii) 200–2000 m and (iv) 2000–4000 m. For Tracks

050, 088 and 126, we average sea level anomalies band by band every cycle, to produce

band-averaged time series of sea level anomalies. To derive the annual cycle from the

sea level anomalies, we fit an annual harmonic to the data using least squares.

Over the inner shelf, the amplitude of the annual cycle is about 5 cm off central NovaScotia, decreasing towards both the northeastern and southwestern ends of the Scotian

Shelf (table 3). The phase pattern shows a ,2 month lag from east to west, suggesting a

southwestward propagation of the sea level signal at a rate of about 6 cm s-1.

The amplitude of the annual cycle is larger over the continental slope. The sea level is

PORSEC 2008 4799

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

Figure 5. Seasonal-mean sea surface height anomalies (colour band, m) for (a) 1994, (b) 1997,and (c) 2000. The 200-m isobath (grey lines) and T/P tracks (white lines) are also displayed.

4800 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

highest in September/October, mainly associated with steric effects due to thermal

expansion of the upper ocean in response to the surface heating and cooling. The

outer Scotian Shelf and the shelf edge show transitional features between those of the

slope and the inner shelf.The RMS sea level values (table 4) indicate the higher-frequency variability dom-

inates over the annual and interannual cycles over the Scotian Shelf and Slope. Off the

central (Track 088) and western (Track 050) Nova Scotia, the interannual variability

shows some increase towards the deep ocean (table 4), presumably due to the proxi-

mity of the Gulf Stream.

5. Summary

We have derived major tidal constituents and seasonal-mean sea level anomalies over

the Scotian Shelf from T/P data. The sea level along the Nova Scotia coast is highest in

autumn and lowest in spring. Over the open shelf, the sea level is higher in summer and

autumn and lower in winter and spring. The sea level along the Nova Scotia coast and

over the Scotian Shelf was low in the early 1990s, high in the mid 1990s, and close to

normal in the late 1990s and early 2000s. The correlation coefficients between alti-

metric and tide-gauge measurements at North Sydney, Halifax and Yarmouth are

Figure 5. (Continued.)

PORSEC 2008 4801

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

0.53 (0.48), 0.64 (0.82), and 0.63 (0.62), respectively, without (with) the seasonal cycle

being removed. The correlation analysis indicates fair agreement at the seasonal and

interannual scales, except at Halifax where there is good agreement at the interannual

scale.

Eight major semi-diurnal and diurnal constituents are derived from the tide-in T/Pdata. Comparisons with tide-gauge data and bottom pressure gauge data indicate

overall good agreement. The present tidal solutions are in better agreement with

0

North Sydney

Canso Harbour

Murphy’s Cove

SS

H A

nom

aly

(m)

Halifax

Lunenburg

Lockeport

−0.06

0.06

−0.06

0

0.06

−0.06

0

0.06

−0.06

0

0.06

−0.06

0

0.06

−0.06

0

0.06

−0.06

0

0.06

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002Year

Yarmouth

Figure 6. Seasonal-mean sea level anomalies (solid lines) at sites along the Nova Scotia coastestimated from T/P data. The dashed lines are the results with the annual and semi-annualcycles removed and subsequently smoothed with a five-point moving filter. SSH, sea surfaceheight.

4802 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

tide-gauge data than Han et al.’s (1996) for the K1 and O1 tides with an improvement

of 30� in phase, but nearly the same for M2, S2 and N2. Detailed comparisons for thethree permanent tide-gauge stations indicate good agreement at Yarmouth and

Halifax, but with notable discrepancies at North Sydney. The RMS amplitude

differences for all eight tidal constituents are 0.012, 0.021 and 0.06 m at Yarmouth,

Halifax and North Sydney, respectively.

For estimating coastal sea levels (both tidal and non-tidal) from satellite altimetry,

higher spatial resolution is demanded. It is apparent that the close agreement between

altimetric estimates and tide-gauge measurements at Yarmouth and Halifax can be

−0.20

−0.10

0.00

0.10

0.20

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

SS

H A

nom

aly

(m)

North Sydney

RMSD = 4.47 cm

Corr. = 0.53

T/P RMS = 4.49 cm Tide−gauge RMS = 4.75 cm

−0.20

−0.10

0.00

0.10

0.20

SS

H A

nom

aly

(m)

Halifax

RMSD = 3.58 cm

Corr. = 0.64

T/P RMS= 4.59 cm Tide−gauge RMS= 3.62 cm

−0.20

−0.10

0.00

0.10

0.20

SS

H A

nom

aly

(m)

Year

Yarmouth

RMSD = 3.49 cm

Corr. = 0.63

T/P RMS= 4.43 cm Tide−gauge RMS= 3.32 cm

Figure 7. Comparison of T/P estimates and tide-gauge observations of seasonal-mean seasurface height (SSH) anomalies.

PORSEC 2008 4803

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

−0.08

−0.04

0.00

0.04

0.08

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

SS

H A

nom

aly

(m)

North Sydney

RMSD = 2.28 cm

Corr. = 0.48

T/P RMS = 2.48 cm Tide−gauge RMS = 1.94 cm

−0.08

−0.04

0.00

0.04

0.08

SS

H A

nom

aly

(m)

Halifax

RMSD = 1.47 cm

Corr. = 0.82

T/P RMS = 2.51 cm Tide−gauge RMS = 2.23 cm

−0.08

−0.04

0.00

0.04

0.08

SS

H A

nom

aly

(m)

Year

Yarmouth

RMSD = 1.89 cm

Corr. = 0.62

T/P RMS = 2.08 cm Tide−gauge RMS = 2.26 cm

Figure 8. Same as figure 7, except for annual and semi-annual cycles being removed andsubsequently being smoothed with a five-point moving filter.

Table 3. Annual cycles of sea level derived from the segment-averaged T/P data. Amplitudesare in metres and phases indicate the year day of the maximum sea level.

050 088 126

Amplitude Amplitude AmplitudeSegment (m) Phase (m) Phase (m) Phase

Shore to mid-shelf 0.037 8 0.049 314 0.035 290Mid-shelf to 200 m 0.013 343 0.052 300 0.050 297200–2000 m 0.046 252 0.055 272 0.048 2802000–4000 m 0.052 266 0.063 275 0.071 264

4804 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

partly attributed to better data coverage near the site. It is therefore expected that the

coastal sea levels can be improved using the data from the concurrent multi-mission

satellite altimetry missions. The improvement may also be achieved by a dynamic

approach which assimilates altimetric observations into a coastal ocean model.

Acknowledgements

We thank J. Li and M. Newhook for assistance in data processing and analysis.

Helpful comments and suggestions were received from the two anonymous reviewers.The project was funded by the Canadian Space Agency, the Climate Change Impacts

on the Energy Sector (CCIES) subprogram of the federal Program on Energy,

Research and Development (PERD), and Fisheries and Oceans Canada. The T/P

altimeter data were obtained from the NASA Pathfinder Project.

References

ANDERSON, C. and SMITH, P.C., 1989, Oceanographic observations on the Scotian Shelf during

CASP. Atmosphere–Ocean, 27, pp. 130–156.

BENADA, R., 1997, Merged GDR (TOPEX/Poseidon) User Handbook. JPL D-11007,

(Pasadena, CA: Jet Propulsion Laboratory).

EGBERT, G.D. and EROFEEVA, S., 2002, Efficient inverse modeling of barotropic ocean tides.

Journal of Atmospheric and Oceanic Technology, 19, pp. 183–204.

CARTWRIGHT, D.E. and RAY, R.D., 1990, Oceanic tides from Geosat altimetry. Journal of

Geophysical Research, 95, pp. 3069–3090.

DE MARGERIE, S. and LANK K., 1986, Tidal circulation of the Scotian Shelf and Grand Banks.

Technical report, Contract Report 08SC.FD901-5-X515, Department of Fisheries and

Oceans, Ottawa, Ontario, Canada.

DRINKWATER, M.R., HAAGMANS, R., MUZI, D., POPESCU, A., FLOBERGHAGEN, R., KERN, M. and

FEHRINGER, M., 2007, The GOCE gravity mission: ESA’s first core earth explorer. In

Proceedings of 3rd International GOCE User Workshop, 6–8 November 2006, Frascati,

Italy, ESA SP-627 (Noordwijk, The Netherlands: ESA), pp. 1–8.

FOREMAN, M.G.G., CRAWFORD, W.R., CHERNIAWSKY, J.Y. and GALBRAITH, J., 2008, Dynamic

ocean topography for the northeast Pacific and its continental margins, Geophysical

Research Letters, 35, L22606, doi: 10.1029/2008GL035152.

FOREMAN, M.C.G. and HENRY, R.F., 1989, The harmonic analysis of tidal model time series.

Advances in Water Resources, 12, pp. 109–120.

Fu, L.L. and Cazenave, A. (Eds), 2001, Satellite Altimetry and Earth Sciences, A Handbook of

Technique and Applications (London: Academic Press).

GODIN, G. 1972, The Analysis of Tides, p. xxi (Toronto: University of Toronto Press).

Table 4. The RMS values (m) of (a) the raw 10-day sea level data, (b) with the annual cycleremoved, and (c) with an additional 37-point moving filter applied.

RMS values (m)

050 088 126

(a) (b) (c) (a) (b) (c) (a) (b) (c)

Shore to mid-shelf 0.118 0.115 0.022 0.116 0.110 0.028 0.108 0.104 0.030Mid-shelf to 200 m 0.097 0.097 0.014 0.107 0.100 0.021 0.113 0.106 0.032200–2000 m 0.107 0.101 0.019 0.114 0.107 0.027 0.109 0.104 0.0292000–3000 m 0.125 0.119 0.040 0.127 0.118 0.037 0.117 0.105 0.024

PORSEC 2008 4805

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14

GODIN, G. 1980, Cotidal charts for Canada. Marine Science Directorate, Manuscript Report

Series, 55, 93 pp., Department of Fisheries and Oceans, Ottawa, Ontario, Canada.

HAKKINEN, S. and RHINES, P.B., 2004, Decline of subpolar North Atlantic circulation during the

1990s. Science, 304, pp. 555–559.

HAN, G., 2002, Interannual sea level variations in the Scotia–Maine region in the 1990s.

Canadian Journal of Remote Sensing, 28, pp. 581–587.

HAN, G., 2004, Scotian Slope circulation and eddy variability from TOPEX/Poseidon and

frontal analysis data. Journal of Geophysical Research, 109, C03028, doi: 10.1029/

2003JC002046.

HAN, G., 2007, Satellite observations of seasonal and interannual changes of sea level and

currents over the Scotian Slope. Journal of Physical Oceanography, 37, pp. 1051–1065.

HAN, G. and LODER, J.W., 2003, Three-dimensional seasonal-mean circulation and hydrogra-

phy over the eastern Scotian Shelf. Journal of Geophysical Research, 108, pp. 3136, doi:

10.1029/2002JC001463.

HAN, G., IKEDA, M. and SMITH, P.C., 1996, Oceanic tides on the Newfoundland and Scotian

Shelves from TOPEX/POSEIDON altimetry. Atmosphere–Ocean, 34, pp. 589–604.

HAN, G., HANNAH, C.G., SMITH, P.C. and LODER, J.W., 1997, Seasonal variation of the three-

dimensional circulation over the Scotian Shelf. Journal of Geophysical Research, 102,

pp. 1011–1025.

HAN, G., HENDRY, R. and IKEDA, M., 2000, Assimilating TOPEX/POSEIDON derived tides in

a primitive equation model over the Newfoundland Shelf. Continental Shelf Research,

20, pp. 84–108.

HAN, G., TANG, C.L. and SMITH, P.C., 2002, Annual variations of sea surface elevations and

currents over the Scotian Shelf and Slope. Journal of Physical Oceanography, 32,

pp. 1794–1810.

LODER, J.W., HAN, G., HANNAH, C.G., GREENBERG, D.A. and SMITH, P.C., 1997, Hydrography

and baroclinic circulation in the Scotian Shelf region: winter vs summer. Canadian

Journal of Fisheries and Aquatic Sciences, 54 (suppl. 1), pp. 40–56.

PELTIER, W.R., 2004, Global glacial isostasy and the surface of the Ice-Age Earth: the ICE-

5G(VM2) model and GRACE. Annual Review of Earth and Planetary Sciences, 32,

pp. 111–149.

PETRIE, B., 2007, Does the North Atlantic Oscillation affect hydrographic properties on the

Canadian Atlantic Continental Shelf? Atmosphere–Ocean, 45, pp. 141–151.

PETRIE, B.D. and DRINKWATER, K., 1993, Temperature and salinity variability on the Scotian

Shelf and in the Gulf of Maine 1945–1990. Journal of Geophysical Research, 98, pp.

20 079–20 089.

PETRIE, B., TOPLISS, B.J. and WRIGHT, D.G., 1987, Coastal upwelling and eddy development off

Nova Scotia. Journal of Geophysical Research, 92, pp. 12 979–12 991.

SUTCLIFFE, W.H., LOUCKS, R.H. and DRINKWATER, K.F., 1976, Coastal circulation and physical

oceanography of the Scotian Shelf and Gulf of Maine. Journal of the Fisheries Research

Board of Canada, 33, pp. 98–115.

THOMPSON, K.R., 1986, North Atlantic sea-level and circulation. Geophysical Journal of the

Royal Astronomical Society, 87, pp. 15–32.

UMOH, J.U. and THOMPSON K.R., 1994, Surface heat flux, horizontal advection, and the

seasonal evolution of water temperature on the Scotian Shelf. Journal of Geophysical

Research, 99, pp. 20 403–20 416.

WANG, Y.M., 2001, GSFC00 mean sea surface, gravity anomaly, and vertical gravity gradient

from satellite altimeter data. Journal of Geophysical Research, 106, pp. 31 167–31 174.

4806 G. Han et al.

Dow

nloa

ded

by [

Fish

erie

s an

d O

cean

s C

anad

a] a

t 08:

48 2

2 Ja

nuar

y 20

14