Sea Level Analysis Based on ERS-1 Altimeter Data
sea level observations are necessary to answer the urgent questions about
climate changes and their impact on the socio-economy. At GFZ/D-PAF, reprocessed
ERS-1 altimeter OPR02 is used to generate a time series of monthly sea
surface height models from April 1992 to March 1995. The reprocessing consists
of D-PAF's improved satellite ephemerides, the inclusion of Grenoble tidal
model, the removal of a time bias and application of different range corrections.
The three year time series was taken to estimate the rate of change of
global mean sea level. A +2 mm sea level rise per year was estimated, which
is far below the assumed 30 to 50 cm in the next century. Regional trends,
however, show extreme differences in the sea level variations: A sea level
rise in the tropics and the Indian ocean (locally up to 10 cm/year), but
a sea level fall in the eastern Pacific and higher latitudes. For the quality
assessment, comparisons to tide gauge measurements were performed, indicating
a high coherence to the ERS-1 result. The relation of sea level variations
and climate change was examined in the global system ocean-atmosphere,
which was represented by sea surface temperatures, wind speeds and wave
heights. It was demonstrated that the sea level change can be attributed
to interannual variability and El Niño.
Level, Altimetry, ERS-1, Ocean-Atmosphere, D-PAF
Urgent questions about possible climate changes may be answered by means
of secular rates of change of the sea level. Trends in the sea level are
considered as indicators of a global temperature rise caused by the increase
of greenhouse gases. Between 1o and 4o Celsius can
be expected for the next century. If this happens, then the sea level will
rise 30 to 50 cm (Church et al.,1991; Houghton
and Woodwell, 1989), caused by the melting of glaciers, polar ice caps
and thermal expansion of the oceans (Church et al.,
1991; Meier, 1984). Such a scenario would lead to
socio-economic consequences that cannot be predicted (Broecker,1996).
The presented results (see also: Anzenhofer and Gruber,
1996; Nerem,1995), however, show, that the sea level
rise has regional characteristics, i.e. there are areas, which are much
more affected than others.
Before the advent of satellite altimetry the sea level could only be observed
through tide gauges. This data, however, has major disadvantages:
- The global distribution is uneven due to their location on the shores
of continents, and islands exclusively. Thus a global trend estimation
of the sea level is impossible.
- Tide gauge measurements may systematically be shifted due to post-glacial
rebound or tectonic uplift towards other stations.
The advantage of altimetry is that the sea surface can be monitored
in a continuous and repeated manner in a common reference datum. The required
accuracy for studying the sea level, however, only fulfills data of the
European satellites ERS-1 and ERS-2 and the US/French satellite TOPEX/POSEIDON.
Due to the insufficient knowledge of altimeter drifts, other systematics
and missing overlaps, measurements of SEASAT, GEOS-3 and GEOSAT (Allan,
1983; Bonavito et al., 1975; Cheney
et al., 1991; Horai, 1982) cannot be used for
a sea level study.
The paper at hand will show, that the secular changes of the sea surface
are in the millimeter range, which is far below the accuracy of satellite
orbits, altimeter range and altimeter corrections. Thus, a very careful
data preparation and calibration is needed.
If the assumption of a sea level rise caused by global warming is true,
then other parameters of the global system ocean-atmosphere must
also exhibit changes in their temporal behaviour. For this purpose, sea
surface temperatures, wind speeds and wave heights were analyzed in the
same manner as sea surface heights. The changes obtained, then were correlated
and interpreted within the framework of the ocean-atmosphere system.
The following chapter describes the different data used for the sea
level study. The altimeter measurements is the base of the sea level study.
Thus the data and its upgrade is described in detail.
Sea Surface Heights from ERS-1 Altimeter
Precise altimeter products (OPR) from the European Space Agency's ERS-1
mission (CERSAT, 1995) are used in this study as
a baseline. For the sea level study consistently processed ERS-1 data for
the period between launch and August 1995 were available. Due to the limited
spatial resolution of the 3 day repeat cycle periods, however, only data
from April 15, 1992 until March 20, 1995 was used. During the sea level
study the other available data of the second multidisciplinary phase were
not used, because they didn't cover a complete year. The annual cycle causes
periodic sea surface changes whose power is much higher than that of possible
secular sea level change. Thus, incomplete years of altimeter data cause
systematic artifacts in the analysis result. Furthermore, OPR data starting
with the second multidisciplinary phase have been processed with a new
software version. A mixing of these versions could cause systematics in
the range measurement and in other parameters, such as the microwave radiometer
Earlier investigations and the ERS cross calibration (Anzenhofer
et al., 1996; Benveniste, 1996a, 1996b) have shown
that the original OPR cannot be used for a sea level study. Additional
corrections, the exchange of up-to-date geophysical corrections and the
merging of consistent orbit ephemerides were necessary. They are listed
in the following paragraphs.
Wave Heights from ERS-1 Altimeter
The slope of the altimeter return signal is highly correlated with the
ocean wave heights. If the altimeter pulse is reflected by a calm sea surface
then the backscattered energy curve plotted versus the elapsed time is
steep. A turbulent sea surface with high waves results in multiple reflections
of the altimeter pulse and thus a flat backscattered energy curve. The
coherence in turn is used to refer the wave heights to the steepness of
the backscattered altimeter signal (Hancock et al.,
Wind Speed from ERS-1 Altimeter
Just as the wave heights, the wind speed has an influence on the altimeter
return signal. The radar backscatter returned to the satellite is modified
by wind-driven ripples on the sea surface and, since the energy of these
ripples increases with wind velocity, backscatter increases with wind velocity.
Sea Surface Temperatures from NMC
The sea surface temperatures used for the study were not taken from
ERS-1's ATSR. Instead the global sea surface temperature grids from the
U.S. National Meteorological Center (NMC), Washington were retrieved (Reynolds,
1987; Reynolds and Smith,1993). NMC mixes in-situ
measurements (ships and buoys) and AVHRR data of the NOAA satellites. The
base of the mixed models are in-situ data. The satellite measurements then
are used to complement areas with sparse data distribution. By means of
a regression analysis, a transformation of the satellite's skin to bulk
temperatures is performed. By an optimum interpolation technique, weekly
sea surface temperature grids are interpolated. The spatial resolution
of the global grids is 1ox1o degree (Reynolds
and Smith, 1993). Monthly grid models are also available.
The tide gauges used for the sea level study are obtained from the PSMSL
Public Access Directory, where the files of the permanent service for mean
sea level are stored. The PSMSL data set comprises monthly and annual means
of sea level (Pugh, 1987; Woodworth,
1996) measured at tide gauge stations.
The sea level investigations are performed by comparing monthly grid
models with a long-term mean. The transition to relative measures enables
the usage of small values. For the long-term mean a simple averaging of
the 36 monthly (3 years) grid models is performed. Monthly solved grid
models of sea surface heights means, that the data of one month are equally
weighted and interpolated to one grid model. The spatial resolution is
always 1 degree. The monthly grids come from different instruments and
sources. Further on, for the generation of grid models different editing
and quality criteria are used. This implies grid models with undefined
nodes that are not common to other grid models. Investigations in the global
system ocean-atmosphere have demonstrated (Fu and
Cheney, 1995), that there exist significant seasonal differences between
northern and southern hemisphere. This means that if grid models with different
numbers of defined and undefined grid nodes are intercompared, then systematic
shifts may happen. In order to avoid this error source, a common masking
for all grid models was performed.
For sea surface height, sea surface temperature, wind speed, wave height
and altimeter corrections (for quality assessment) two analyses have been
performed to extract variations within the investigation time period:
- Global Rate of Change. The first step of this method is to subtract
the monthly grid models from the long-term mean (3-year model). Then the
differences in the grid nodes are simply averaged to one mean. The obtained
value now represents the mean deviation from a stationary state. The successive
processing of all monthly grids leads to a time series which in turn is
used to estimate the rate of change. This is done by a regression analysis.
The regression coefficient (and its standard deviation) defines the global
rate of change for the abovementioned parameters, like sea surface height,
sea surface temperature, wind speed, wave height and altimeter corrections.
- Local Rates of Change. With the availability of the global rate
of change value the question is raised, whether the trend is evenly distributed
over the oceans or if there are regional characteristics. Therefore, a
regression analysis was performed for each grid node of the monthly models.
The estimated regression coefficients (and standard deviations) or local
rates of change then were visualized by raster plots.
Result and Quality Assessment
Due to the large number of different investigations, table representations
comprise the results.
Figure 2: Sea Level Trend
|Sea Level Trend: The
global rate of change indicates a sea level rise of globally +2.0±1.9
mm/yr. Regional characteristics, however, are evident, a sea level
rise in the tropics and around the western boundary currents and a sea
level fall in the eastern Pacific and in higher latitudes. But,the global
trend is far below the anticipated 30 to 50 cm for the next century. The
value coincides with former tide gauge investigations (Fu
and Cheney, 1995; Trupin and Wahr, 1990).
Figure 3: Sea Surface Temperature Trend
|Sea Surface Temperature Trend:
For the sea surface temperatures the global rate of change marks a slight
global warming of +0.042±0.032 K/yr.
A significant disproportion between northern and southern hemisphere trends
is evident, a warming in the southern and a cooling in the northern hemisphere.
A coherence between the sea level and the sea surface temperature trends,
however, can be found around the western boundary currents and the tropics.
It is physically impossible that the obtained disproportion has stationary
character. Thus an interannual variability with periods greater than 3
years is more realistic, e.g. decadal period (Latif et
al., 1995; Groetzner et al., 1996).
Figure 4: Wind Speed Trend
|Wind Speed Trend: The
global rate of change indicates also a positive trend of +64±29
mm/s/yr. The anticorrelation to the sea surface temperature is clearly
visible. This fact excludes data and processing errors of the sea surface
temperatures. The increase of wind speeds in the northern and decrease
in the southern hemisphere confirms an interannual variability mentioned
Figure 5: Wave Height Trend
|Wave Height Trend: The
expected high coherence with wind speed trends, as shown above, is evident,
a positive trend of the wind speed leads to an increase of corresponding
wave heights and vice versa. The global trend is positive +11±12
For the quality assessment of the sea level result, the altimeter corrections
were examined in the same manner as sea surface temperatures or wind speeds:
the corrections were gridded, filtered by the common mask and global/local
rates of change were estimated.
Figure 6: Wet Tropospheric Trend
Figure 7: Dry Tropospheric Trend
|On the sea surface the wet tropospheric correction has a
global trend of +0.8±0.5 mm/yr. This
value points to a markable influence on the sea level result. The coherence
with the sea surface temperatures, as seen in figure 3, reflects an internal
consistency and thus data quality. The correction derived from ERS-1 radiometer
is believed to be accurate.
||With a global trend of -0.1±0.3
mm/yr the dry troposphere has almost no influence on the sea level
result. Only small regional structures are visible. The correction is believed
to be very accurate.
Figure 8: EM-Bias Trend
Figure 9: Inv. Barometer Eff. Trend
|Similar to the dry tropospheric correction, the EM-bias
has almost no influence on the sea level estimate. The global trend is
only +0.1±0.2 mm/yr. Due to its dependence
on wave heights, the regional structures of wave heights are visible.
||With a global trend of -0.4±1.3
mm/yr the inverse barometer effect influences the sea level estimate.
The trends, however, have regional structures. These are concentrated around
the main pressure cells of the atmosphere. The correction is critical,
because pressure data are mainly derived from model runs and not from real
Figure 10: Ionospheric Trend
Figure 11: Tide Gauges Trend
|The global trend of the ionospheric correction (Bent model)
on the sea level is -8.6±1.7 mm/yr,
which is four times the sea level trend estimate. The regional influence
looks like a global systematic shift. The huge value itself demonstrates
that the ionospheric correction is the most critical point of the sea level
||As described above, the tide gauges data are retrieved from
the PSMSL archive. To show the excellent coherence between the sea level
trend and tide gauges measurements, only tide gauges are investigated,
which cover the time period between April, 1992 and March, 1995. Then a
regression analysis of the tide gauge time series was performed. The trend
values were gridded with a large influence circle to get a filled raster
plot for intercomparison.
By overlaying the sea level trend image (figure 2) and the image above
almost all positive and negative sea level trends are confirmed by tide
The ERS-1 altimeter is a single-frequency instrument. This means that
the ionospheric correction cannot be retrieved from path delays as it is
done for TOPEX/POSEIDON dual-frequency measurements. Thus, the ERS-1 ionospheric
correction must be computed from ionospheric models, like Bent, IRI90 or
IRI95. These models strongly depend on solar activity. Anomalies like sun
storms cannot be modeled. Thus, the ionospheric correction based on models
can never be as accurate as dual-frequency derived corrections. In order
to find the appropriate model for the sea level study, the Bent and IRI95
models are compared with TOPEX ionospheric corrections. Therefore,
the differences of Bent and TOPEX corrections (IRI95 and TOPEX corrections
as well) were computed and displayed as a time series. Then both time series
passed through a regression analysis:
- Trend of (Bent - TOPEX): -0.7 ± 0.9 mm/yr
- Trend of (IRI95 - TOPEX): -2.8 ± 1.3 mm/yr
It seems that the Bent model matches the real ionosphere much better
than IRI95. The estimated values demonstrate again that the ionospheric
correction is the most critical point of the sea level study. A change
to IRI95, for example, leads to a global sea level fall of -1
The investigation of upgraded ERS-1 altimeter data yielded a global
sea level rise of +2 mm/year within April 1992 and March 1995. For the
same time period wind speeds, wave heights and sea surface temperatures
were analyzed. Global positive trends were found as well. Besides global
trends, local rates of change were estimated, too. A sea level rise was
detected in the tropics and in the Indian ocean and a sea level fall in
higher latitudes, respectively. The local rates of change of wind speeds,
wave heights and sea surface temperatures exhibited a disproportion between
variations on the northern and southern hemisphere. Correlation to the
sea level change could be identified. A high coherence between parameters
of the global system ocean-atmosphere was evident and could be demonstrated.
Thus, it can be supposed that the variations can be attributed to periods
larger than the investigated 3 years (Hastenrath,
1984; Houghton and Woodwell, 1991; Latif
et al., 1995). The small amounts of global and local rates of change
raise the question about their significance. Therefore, the altimeter corrections
were individually processed and analyzed. It was found that deficiencies
with respect to their quality and long-term behaviour exist. Especially
the ERS-1 ionospheric correction, which is derived from a model, is very
critical. Depending on the model adopted for the correction, either a sea
level fall or a sea level rise can be produced. Investigations, however,
have shown, that the Bent model matches much better the ionospheric correction
derived from TOPEX dual-frequency measurements than the IRI95 model. Thus
a sea level rise is more likely.
The question about the long-term behaviour of the sea level trend is
attempted to be answered by a 14 years time series of sea surface temperatures.
Figure 12: Analysis of Sea Surface Temperatures and Amplitude Spectrum
The low-passed sea surface temperature time series shows that longer
periods of global warming and cooling happened in the past. Volcanic eruptions,
El Niño/ENSO and other phenomena (decadal period) can be attributed
as reasons for the temperature changes (Keeling et al.,
1989; Latif et al., 1995). Due to the relationship
between the sea level and sea surface temperatures (Stammer
and Wunsch, 1994), it can be concluded that longer periods of sea level
rise and fall happened in the past, too. Thus, the analysis time period
of 3 years is much too short for long-term predictions. During the sea
level study it became clear that the altimeter corrections and their accuracy
make a significant sea level estimate questionable. By embedding the sea
level in the global system ocean-atmosphere, however, it is possible
to answer open questions and to verify the results.
Two major requirements for future activities must be set up:
- Long-term observations of oceanic and atmospheric
parameters are absolutely necessary for climate studies.
- Well calibrated and overlapping data of different
sensors/missions is needed. Corrections to data must all have the same
level of accuracy.
A number of data was provided by different institutes. We thank the
F-PAF for the operational provision of ERS OPR2 altimeter data. The IRI95
model was kindly provided by D. Bilitza. A thank belongs to O.B. Andersen
and R. Ray for the preparation of the FES95.1 tidal loading modules. The
range correction tables for ERS-1 were obtained from ESRIN/ESA. A special
thank belongs to Richard Reynolds from NMC, Washington, for the preparation
of sea surface temperature models and his helpful comments.
- Allan T.D., 1983
- Satellite Microwave Remote Sensing, Ellis Horwood Series in Marine
- Anzenhofer M., C. Rajasenan, Th. Gruber, F.H. Massmann,
- ERS-2 RA Calibration Using OPR2, Final Technical Report, GFZ,
- Anzenhofer M., Th. Gruber, 1996b,
- Fully Reprocessed ERS-1 Altimeter Data From 1992 to 1995: Detection
of Sea Level Changes, Submitted to J. Geophys. Res.-Ocean Special Section
on Advances in Oceanography and Sea Ice Research using ERS Observations
- Bent R. B., S.K. Llewellyn, Nesterczuk G., 1975,
- The Development of a Highly-Successful Worldwide Empirical Ionospheric
Model and its Use in Certain Aspects of Space Communications and Worldwide
Total Electron Content Investigations, Effect of the Ionosphere on Space
Systems and Communications Goodman J. M.
- Benveniste J., 1996a,
- Minutes of the ERS-2 Radar Altimeter and Microwave Radiometer Commissioning
Working Group (#9), ESA-ESRIN, 30-31 January 1996, rs/pm/jb/96.009,
- Benveniste J., 1996b,
- Minutes of the ERS-2 Radar Altimeter and Microwave Radiometer Commissioning
Working Group (#10), ESA-ESRIN, 25-26 April 1996, rs/pm/jb/96.026,
- Bilitza D., 1996,
- International Reference Ionosphere 1995, personal communication
- Bonavito N.L., R.A. Gordon, J.G. Marsh, 1975,
- On the Nature of the Radial and Cross Track Errors for Artificial Earth
Satellites, Goddard Space Flight Center, Greenbelt, Maryland
- Broecker W.S., Plötzliche Klimawechsel, 1996,
- Spektrum der Wissenschaft, Januar 1996, 86-92
- CERSAT, 1995,
- Altimeter & Microwave Radiometer ERS Products - User manual,
- Cheney R.E., N.S. Doyle, B.C. Douglas, R.W. Agreen,
L. Miller, E.L. Timmerman, D.C. McAdoo, 1991,
- The Complete Geosat Altimeter GDR Handbook, NOAA Manual NOS NGS
7, National Ocean Service, Rockville, MD.
- Church J.A., J.S. Godfrey, D.R. Jackett, T.J.
- A Model of Sea Level Rise Caused by Ocean Thermal Expansion, J.
Clim., 4(4), 438-456
- Fu L.L., R.E. Cheney, 1995,
- Application of Satellite Altimetry to Ocean Circulation Studies: 1987-1994,
U.S. National Report to IUGG, 1991-1994, Rev. Geophys., Vol. 33
- Groetzner A., M. Latif, T.P. Barnett, 1996,
- A Decadal Climate Cycle in the North Atlantic Ocean as Simulated by
the Echo Coupled GCM, Max-Planck-Institut für Meteorologie,
No. 208, Hamburg
- Hancock D.W., F.G. Forsythe, J. Lorell, 1980,
- SEASAT Altimeter Sensor File Algorithms, IEEE J. of Ocean Engineering,
- Hastenrath S., 1984,
- Interannual Variability and Annual Cycle: Mechanisms of Circulation
and Climate in the Tropical Atlantic Sector Monthly Weather Review,
Vol. 112, 1097-1107
- Horai K., 1982,
- Adjustment, Interpolation and Smoothing of GEOS-3 Altimeter Data, J.
Geophys. Res., Vol. 87, 8693-8707
- Houghton R.A., G.M. Woodwell, 1989,
- Global Climatic Change, Scientific American,Vol. 260, 4, 18-26
- Keeling C.D., R.B. Bacastow, A.F. Carter, S.C.
Piper, T.P. Whorf, M. Heimann, W.G. Mook, H. Roeloffzen, 1989,
- A Three-Dimensional Model of Atmospheric CO2 Transport Based
on Observed Winds: 1. Analysis of Observational Data, Geophysical Monograph
- Latif M., R. Kleeman, Ch. Eckert, 1995,
- Greenhouse Warming, Decadal Variability, or El Niño ? An Attempt
to Understand the Anomalous 1990s, Report No. 175, Max-Planck-Institut
für Meteorologie, Hamburg
- Le Provost, C., F. Lyard, J.M. Molines, M.L.
Genco, F. Rabilloud, 1996,
- A Hydrodynamic Ocean tTde Model Improved by Assimilating a Satellite
Altimeter Derived Data Set, submitted to J. Geophys. Res.
- Llewellyn S.K., R.B. Bent, 1973,
- Documentation and Description of the Bent Ionospheric Model, AFCRL-TR-73-0657
- Marsh J.G., R.G. Williamson, 1982,
- SEASAT Altimeter Timing Bias Estimation, J. Geophys. Res., Vol.
87, C5, 3232-3238
- Meier M.F., 1984,
- The Contribution of Small Glaciers to Global Sea Level Rise, Science,
- Nerem R.S., 1995,
- Measuring Global Mean Sea Level Variations Using TOPEX/POSEIDON Altimeter
Data, J. Geophys. Res., Vol. 100, C12, 25135-25151
- Pugh D.T., 1987,
- The Global Sea-Level Observing System, Hydrographic J., No.
- Reynolds R.W., 1987,
- A Real-Time Global Sea Surface Temperature Analysis, J. Climate,
- Reynolds R.W., Th.M. Smith, 1993,
- Improved Global Sea Surface Temperature Analyses Using Optimum Interpolation,
- Schwintzer P., Ch. Reigber, A. Bode, Z. Chen,
F.-H. Massmann, J.C. Raimondo, J.M., Lemoine, G. Balmino, R. Biancale,
B. Moynot, J.C. Marty, F. Barlier, Y. Boudon, 1993,
- Improvement of GRIM4 Earth Gravity Models Using Geosat Altimeter and
Spot-2 and ERS-1 Tracking Data, in: Proceedings of IAG Symposium No.
112 Geodesy and Physics of the Earth, ed. H. Montag, Ch. Reigber, Springer
- Schwintzer P., Ch. Reigber, A. Bode, Z. Kang,
S.Y. Zhu, F.-H. Massmann, J.C. Raimondo, R. Biancale, G. Balmino, J.M.
Lemoine, B. Moynot, J.C. Marty, F. Barlier, Y. Boudon, 1996,
- Long-Wavelength Global Gravity Field models: GRIM4-S4, GRIM4-C4, J.
of Geodesy, in print
- Stammer D., C. Wunsch, 1994,
- Preliminary Assessment of the Accuracy and Precision of TOPEX/POSEIDON
Altimeter Data With Respect to Large-Scale Ocean Circulation, J. Geophys.
Res., Vol. 99, C12, 24584-24604
- Trupin A., J. Wahr, 1990,
- Spectroscopic Analysis of Global Tide Gauge Data, Geophys. J. Internat.,
No. 100, 441-453
- Woodworth Ph., 1996,
- PSMSL Annual Report for 1995, PSMSL
Keywords: ESA European
Space Agency - Agence spatiale europeenne,
observation de la terre, earth observation,
satellite remote sensing,
teledetection, geophysique, altimetrie, radar,
chimique atmospherique, geophysics, altimetry, radar,