El Niño observed by ERS/Envisat altimetry (details)
Last update: Wed May 9 23:23:44 CEST 2012
Observations from ERS/Envisat satellite altimetry
Every 35 days Envisat (like is predecessors ERS-1 and ERS-2)
samples the oceans along a regular dense pattern of ground
tracks. The along-track distance between the 1-Hz measurements used for
this study is about 7 km; the spacing between parallel tracks is about 80
km at the equator. After 16 days into this repeat cycle the pattern is
nearly half filled, except that the begin point and end point do not exactly
meet, but are offset by 80 km in longitudinal direction (Figure 1). This 16 days is called a
sub-cycle of the repeat period and is the time span over which we
have combined the data for this study. Given the slow temporal fluctuations of
the El Niño phenomenon, we can assume that the sea level is constant
during this period and that this assumption will not deteriorate our results.
Apart of a 35-day repeat cycles, ERS-1 also flew two 168-day repeat cycles
and several 3-day repeat cycles. The 168-day repeat is processed similar as
the 35-day repeat because it also has a 16-day sub-cycle. The 3-day repeat
(Jan-Mar 1994), however, is too coarse to give a detailed picture of sea
level variations and is excluded. This you will find back as a white band
covering three months.
Figure 1.
Ground track of Envisat over the Equatorial Pacific during a single 35-day repeat
period. The tracks for the first 16 days are colour coded in blue, the
second 16 days in red, and the remaining 3 days in black. Note that the
16-day periods closely resemble a perfect repeat cycle.
Details of the data processing
The ERS-1 and ERS-2 altimeter data are extracted from the
ESA/F-PAF Ocean Products:
OPR02 data available on CD-ROM.
The Envisat data are extracted from the
RA2 IGDR products provided by ESA.
To create sea level anomalies,
the 1-Hz altimeter measurements are corrected for:
- DUT/DEOS For ERS-1/2: precise,
preliminary, or near-realtime orbits based on SLR
and crossover tracking data and using the DUT/DEOS DGM-E04 gravity field
model. For Envisat: CNES preliminary orbits.
- Bias jumps and oscillator drift.
- Dry tropospheric (ECMWF), wet tropospheric (Microwave
Radiometer), and ionospheric (IRI95, JPL GIM, or dual-frequency) path delays.
- Solid Earth tides and pole tide.
- Ocean tides and loading (GOT00.2).
- Sea State Bias correction (BM3).
- Inverse barometer correction.
- Mean sea surface (CLS01).
The relative sea heights are then gridded to regular grids with a resolution of
1°×1°. Although the epoch of the grids are spaced by 7 days,
the data span for each weekly solution is 16 days (one sub-cycle) centred
around the epoch. The gridding uses a Gaussian function with a length scale
of 1.2° to weight the along-track measurements to the grid points.
Data refinement
This image above was produced with the preliminary orbits, and no
orbit error correction was applied. But since some refinement can be
achieved by removing the 5-7 cm of orbit error from the sea level
anomalies, we have also performed a minimisation of sea level
differences at crossovers of ascending and descending tracks with a maximum
time interval of 17.5 days. Using a two-parameter (tilt and bias) orbit
error model for each of the 5228 tracks, the root-mean-square of the
152941 crossover height differences was reduced from 59 to 50 mm.
The crossover minimised data have also been gridded as described above and
then folded into Hovmuller or time-longitude diagrams
of sea level anomalies.
Such diagrams show the evolution in time of the relative sea level along a
single latitude band. The construction of such a diagram is illustrated in
Figure 2.
Figure 2.
Construction of a time-longitude diagram along the equator. The grid values
of each weekly grid along the equator are copied into the time-longitude
diagram at the correct location.
The time-longitude diagram
of sea level anomalies along the equator is shown in Figure 3. After two years of relatively low sea level in the
East Pacific (which, however, can not be characterised as La Niña)
there are two waves visible: one hits the American coast in July 1997, the
second in October 1997. But also more minor
waves are seen to propagate rapidly from the West to the East Pacific. The
propagation speed can be estimated at about 4 m/s !
Figure 3.
Time-longitude diagram of sea level anomalies along the equator.
(Click on the image to get the snapshot for the time given by the
vertical position of your cursor.)
Time-longitude diagrams for higher latitudes
are also available on this site:
2°N,
2°S,
5°N,
5°S,
8°N,
8°S,
10°N,
10°S,
15°N,
15°S,
20°N,
20°S,
25°N,
25°S,
30°N and
30°S,
It can be shown that the crossover minimisation has a slight positive effect
on the results, but does not remove any of the ocean signal. This
is done by differencing time-longitude
diagrams.
