Monitoring El Nio With Satellite Altimetry

1
Monitoring El Niño WithSatellite Altimetry

• Dr. Don P. Chambers
• Center for Space Research
• The University of Texas at Austin
• El Niño Workshop- Online for Education
• March 16, 1998

Center for Space Research, The University of
Texas at Austin
2
Introduction

• El Niño has received a lot of attention this
  year, mostly due to the fact that several
  different instruments, both in the ocean and in
  space, detected signals of an impending El Niño
  nearly a year before the warming peaked. One of
  these instruments was a radar altimeter on a
  spacecraft flying nearly 1300 km (780 mi.) over
  the ocean called TOPEX/Poseidon (T/P). T/P was
  launched in September 1992 and is a joint mission
  between the United States and France. It is
  managed by NASAs Jet Propulsion Laboratory and
  the Centre National dEtudes Spatiales (CNES).
• In this presentation, I will describe in general
  the altimeter measurement and discuss how it can
  be used to monitor El Niño. In particular, I
  will discuss how the altimetry signal is related
  to sea surface temperature, and how it is
  different, and how altimetry can detect El Niño
  signals months before peak warming occurs in the
  eastern Pacific.
• To begin with, I would like to show two recent
  images of the sea-level measured by
  TOPEX/Poseidon and point out the El Niño
  signatures. Then, I will discuss the
  measurements that went into the image and discuss
  how we know they are really measuring what we
  think they are.

3

• This is the sea-level anomaly map from the
  TOPEX/Poseidon altimeter, for January 28 to
  February 7, 1998.

4

• This is the sea-level anomaly map from the
  TOPEX/Poseidon altimeter, for February 7 to
  February 17, 1998. Notice the changes occuring
  in the eastern Pacific.

5

• A sea-level anomaly is the difference between the
  total sea-level and an average sea-level for this
  time of year. We look at anomalies because the
  total sea-level measurement made by the altimeter
  varies from 100 meters. Most of this is
  constant, though, and is due to the Earths
  gravity and the ocean circulation. Sea-level
  variations caused by El Niño account for less
  than 1 of the total signal. If the constant
  part were not removed, the El Niño signal would
  not be observable.
• El Niño is evident in the figures on the previous
  page as higher than normal sea-levels in the
  east, and lower than normal sea-levels in the
  west. Sea-level is beginning to drop in the
  east, though, and indications are that El Niño is
  beginning to dissipate. The sea-level is
  actually highest north and south of the equator.
  This is a sign that El Niño is dissipating, and
  will be discussed later. Also, note the high
  sea-levels in the Indian Ocean, which are nearly
  a mirror-image to El Niño in the Pacific. I will
  discuss this later in the presentation as well.
• The altimeter does not directly produce the image
  shown on the previous slide. Instead, it
  measures the height of the spacecraft above the
  ocean surface, which can be converted into
  sea-level. These sea-level measurements can then
  be mapped to a uniform grid, color coded, and
  displayed. However, the conversion to the
  sea-level measurement is not a simple procedure.

6

• Recall that TOPEX/Poseidon orbits the Earth at an
  altitude of 1300 km. The altimeter sends several
  thousand radar pulses towards the ocean each
  second then measures the time for the pulse to
  return to the spacecraft. Not only does the
  computer have to match the right transmitted
  pulse with the right received pulse, but it has
  to compute the time of transit properly. An
  error in time of only a microsecond would mean an
  error of tens of centimeters.
• To get the height of the satellite above the
  ocean, the time for the pulse to return has to be
  converted to range using the speed of light.
  However, the speed of light is constant only in a
  vacuum. The radar pulse travels through the
  atmosphere twice, where it is refracted by air
  molecules, water vapor, free electrons, and is
  partially scattered by surface waves. The size
  of these errors add up to more than 3 meters.
  However, all of them can be measured or modeled.
• Finally, the precise location of the satellite
  needs to be known, since the sea-level is the
  difference between the satellite location above
  the center of the Earth and the height of the
  satellite above the ocean. If the satellite
  location were only good to 1 meter, then the
  sea-level measurement would only be good to 1
  meter.
• But, how accurate is the image on the previous
  slide? The accuracy is amazing, considering all
  of these problems listed above about 2 to 3 cm
  (1 in). Thats about the diameter of a quarter.
  We know this because we can compare sea-level
  measured by T/P with sea-level measured at the
  ocean surface with tide gauges.

7

• Here we have compared altimeter measurements with
  nearby tide gauges at two sites Pohnpei in the
  western Pacific and Baltra in the eastern
  Pacific. Sea-level in these regions are very
  large during El Niño events (indicated by
  arrows). Sea-level is much lower than normal in
  the west and higher than normal in the east as
  will be explained later.
• If the tide gauges are considered to be perfect
  (which they are not), the accuracy of the
  altimeter is about 3 cm. However, if a more
  realistic error is assumed for the tide gauges
  then the accuracy of the altimeter measurement is
  around 2 cm. Note that we are using the mean
  standard deviation as a measure of error
  differences at some times can be up to 3 times
  this number. Also, comparisons at some tide
  gauges are better or worse. The average is about
  2.5 cm.

Pohnpei in Western Pacific
Baltra in Eastern Pacific
(7N, 158E)
(0, 90W)
Standard Deviation of Differences 3 cm
Standard Deviation of Differences 3.5 cm
Sea-Level Anomaly (cm)
Sea-Level Anomaly (cm)
Year
Year
8

• T/P flies in a groundtrack that repeats every
  10-days and goes as far north and south as 66
  latitude. This means that it samples
  approximately 400000 points over the ocean every
  10-days. Compare that to ship measurements,
  moored buoys, or tide gauges. However, there are
  still gaps in the data, which means that the
  image in the earlier slide is based on
  interpolating real measurements to a constant
  grid, then smoothing them and plotting them
  graphically and assigning colors based on the
  heights.

Latitude (N)
Longitude (E)
Lines are T/P groundtrack and dots are locations
with more than 24 ship measurements in 3 years.
9

• Before describing how altimetry can monitor El
  Niño, I would like to discuss how El Niño begins,
  and why El Niño warming causes a sea-level
  variation.
• During normal conditions, winds blow from east
  to west, due to differences in the atmospheric
  pressure. Normally, a high pressure system sits
  over the eastern Pacific, while a low pressure
  system sits over the western Pacific. Because of
  the low pressure system in the west, there is
  increased upward convection which puts water
  vapor into the atmosphere and results in more
  rainfall in the west than in the east.
• At the same time, the surface currents along the
  equator generally move east to west. This
  transports water warmed year round by the sun to
  the western Pacific, where it tends to pile up
  before flowing north and south as other currents.
  This pushes down the thermocline, or the region
  where the temperature change with depth is the
  greatest. In the east, cold water upwells, or
  rises up from great depths to
• replace the warm water which flowed west, and the
  thermocline
• is shallow.
• All of this will cause a certain sea-level
  signature, since
• sea-level is a measure of the integrated water
  density. Warmer
• water has a lower density than colder water, and
  takes up a
• greater volume. Thus, sea-level is higher where
  the
• thermocline is depressed and the upper waters
• are warm, and sea-level is lower where the
• thermocline is raised and upper water is cool.
• Changes in the atmosphere over the
• western Pacific cause all of this to change.

Figure courtesy of NOAA PMEL
10

• Several weeks to several months before El Niño
  begins to manifest in the eastern Pacific, a
  dramatic change occurs in the atmosphere over the
  Pacific. The pressure over the western Pacific
  increases while the pressure in the east
  decreases. This causes a change in the wind
  pattern and the convection, as shown in the
  figure below. This atmospheric change is called
  the Southern Oscillation. The Trade Winds
  decrease or even reverse and blow from west to
  east. The equatorial surface current slows, and
  a subsurface current that always exists
  strengthens. This undercurrent has a core at
  about 100 meters depth and always flows west to
  east. Most of the time it dies out in the
  central Pacific. During El Niño, though, the
  current is strong all the way to the Galapagos
  Islands off the coast of South America it
  sometimes surfaces.
• Thus, there is a change in the ocean associated
  with the
• Southern Oscillation. In the west, the
  thermocline rises
• and warm water flows east. In the east, months
  after the
• initial wind changes, the thermocline gets
  deeper.
• The upper water column warms, and the sea-level
  rises.
• This is a simplified model. The ocean takes
• some time to adjust to the wind changes. One of
• the primary ways it adjusts during El Niño is
• through the creation of Kelvin waves. These
• waves are very different from the surface waves
• you are probably familiar with and are discussed
• more in the following slides.

Figure courtesy of NOAA PMEL
11
Cross-Section Looking West Winds east to west
Cross-Section Along Equator, Looking North

• Although surface currents along the equator
  typically flow east to west, the net transport of
  water in the equator occurs deeper and is away
  from the equator as shown above. This is called
  the Ekman transport, and is named after the
  oceanographer who discovered it. The transport
  is caused by the winds and the Coriolis force,
  which is caused by the rotation of the Earth.
  Ekman found that the net water transport was
  below the surface and that it was perpendicular
  to the wind direction to the right of the wind
  in the north, and to the left of the wind in the
  south. Thus, since the Trade Winds typically
  blow east to west in the equatorial region, the
  net transport is away from the equator. Notice
  the much deeper thermocline, or boundary between
  the warm and cold water, in the west than in the
  east. Most of the sea-level variations measured
  by altimetry are caused by changes in the
  thickness of this layer of warm water.

12
Cross-Section Looking West Anomalous Winds west
to east
Cross-Section Along Equator, Looking North

• During the Southern Oscillation, winds are weaker
  in the west thus the wind anomaly, or deviation
  from an average wind for that time of year is
  from west to east. This causes a change in the
  Ekman transport in this region. Instead of
  flowing away from the equator, it flows toward
  the equator. To balance this anomalous influx of
  water, the warm water in the upper layer
  downwells to the lower layer. However, the warm
  water is lighter than the cooler water, and is
  naturally more buoyant. The forcing of the
  warmer water into the cooler water will cause an
  oscillation, normally at the steepest density
  gradient, the thermocline. These oscillations
  will propagate away from the source of the wind
  anomaly as very long waves. Along the equator,
  these are called Kelvin waves and they change the
  thickness of warm water in both the west and the
  east, which causes large sea-level changes and
  leads to El Niño.

13

• Kelvin waves can cross the Pacific in two months.
  They can only exist near the equator due to the
  Earths rotation. The amplitude of the Kelvin
  wave is several tens of meters along the
  thermocline, and the length of the wave is
  thousands of kilometers (1 longitude 111 km)
• The figure at the right shows Kelvin waves inside
  the ocean, computed with temperature data from
  moored buoys operated by NOAA. It shows the
  depth of the 20C temperature level as a series
  of standard latitude-longitude plots stacked in
  time. The latitude width of each time step is 4
  (2S to 2N) and time is from March 1996 at the
  top to March 1998 at the bottom.
• Notice the yellow/orange lines which slope across
  the Pacific beginning in January 1997 (marked by
  thick lines). These are Kelvin waves. Eastward
  movement is indicated by the slope in time from
  west to east. These waves set up a change in the
  warm water thickness in the eastern Pacific
  beginning in March. Other Kelvin waves are
  visible after El Niño developed the first two,
  however, were early indicators that an El Niño
  would probably occur this year.

1996
1997
1998
Figure courtesy of NOAA PMEL
14

• Because a Kelvin wave is associated with density
  fluctuations inside the ocean, it can be seen in
  the sea-level measurements made by altimeters,
  although with a reduced magnitude. The figure to
  the right shows sea-level anomalies determined
  from TOPEX/Poseidon. Notice the two Kelvin waves,
  with amplitudes of 10 and 15 cm in sea-level
  compared to 30 and 40 meters in thermocline
  change. Note that a depression in the
  thermocline (from the previous picture) is
  associated with an increase in sea-level.
• The Kelvin waves travel east and set up changes
  in the eastern Pacific that lead to El Niño by
  depressing the thermocline there. There are
  smaller Kelvin waves in 1995 and 1996. These are
  seasonal, and do not cause El Niño events.
  However, large Kelvin waves, such as those
  observed in early 1997 almost always lead to El
  Niño events.

A larger version of this image is available
from http//www.csr.utexas.edu/eqpac
15

• An important thing to realize is that the figures
  on the previous two slide are updated in
  near-realtime they are not figures that are
  produced months to years after the data have been
  processed. This means that the first Kelvin
  wave was noticed by scientists connected with the
  TOPEX/Poseidon project and the TOGA-TAO project
  in early February 1997. Because the Kelvin wave
  was large, it was expected that there would be at
  least a moderate El Niño warming either in the
  late spring or early summer, although at that
  time, no one expected that it would surpass the
  event of 1982-1983. Thus, these scientists were
  not surprised when an official NOAA El Niño
  advisory was issued in April.
• The plots from the previous slides are still
  updated on a regular basis and are available over
  the World Wide Web. The TOGA-TAO data can be
  accessed from the NOAA Pacific Marine
  Environmental Laboratory
• http//www.pmel.noaa.gov/toga-tao/realtime.html
• The TOPEX/Poseidon data can be accessed from the
  University of Texas, Center for Space Research
• http//www.csr.utexas.edu/eqpac

16
Sea-Level, Ocean Heating, and El Niño

• I have mentioned several times the correlation
  between sea-level and changes in temperature and
  thickness of the upper layer. Basically, as the
  temperature and thickness of the upper layer
  increases, the sea-level will increase. How well
  does the sea-level follow temperature changes in
  ocean? In the tropical Pacific, the answer is
  very well. One can see this by comparing T/P data
  to data obtained from the TAO moored buoys in the
  Pacific. The buoys measure only temperature at
  various depths, and these can be converted to
  density and then sea-level due to heating
  variations only. The agreement is very good, as
  seen in the figures below. There is more
  disagreement in the western Pacific than in the
  east, due to factors other than heating that
  effect the T/P sea-level measurement in this
  region.

Averaged at TAO Buoys in Eastern Pacific
Averaged at TAO Buoys in Western Pacific
(5S to 5N, 260E - 290E)
(5S to 5N, 150E - 175E)
Standard Deviation of Differences 2.8 cm
Standard Deviation of Differences 3.8 cm
Sea-Level Anomaly (cm)
Sea-Level Anomaly (cm)
Year
Year
17
Sea-Level vs. Sea Surface Temperature

• Another important type of instrument used to
  measure El Niño is the Advanced Very-High
  Resolution Radiometer (AVHRR). These instruments
  are also based in space, and convert the observed
  infrared radiation into Sea Surface Temperature
  (SST). The SST measured by these satellites is
  really the temperature on the very thin layer at
  the top of the water column. Since sea-level
  change during El Niño is caused mainly by heating
  variations, one would expect that SST and
  sea-level would be very similar.
• However, there are significant differences.
  Ignoring for a moment the sea-level effects of
  currents, winds, and other forces, recall the
  previous discussion that much of the early El
  Niño signal occurs below the ocean surface. It
  takes some time for heating at 100 m to reach the
  surface. However, the changes will be seen in
  sea-level measured by altimetry or the moored
  buoys which monitor temperatures to a deep level
  as they are happening. This means that altimetry
  can actually see El Niño signals slightly before
  the SST measurements can. The difference in time
  is only slight, about 2 weeks to a month.
  However, by combining the two measurements, one
  can get a more complete picture of how El Niño
  evolves.
• The next four slides will demonstrate this by
  showing complementary sea-level and SST anomalies
  for each month during 1997. The fifth slide
  following will summarize important events during
  the evolution of the 1997 El Niño.

18
cm
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22

• The T/P data show a Kelvin wave in the central
  Pacific in January, while the SST data indicate
  cooler than normal surface waters in the region.
  By February sea-level and SST were both higher
  than normal in the eastern Pacific after the
  Kelvin wave reached the coast. In March, a
  larger Kelvin wave began forming and can be seen
  in the T/P data again, it is not observable in
  the SSTs.
• The second Kelvin wave moved across the Pacific
  in April and May, as observed by T/P. By May,
  there was a tongue of warm water extending
  westward past 250E. The T/P data, sensitive to
  heat changes at depth, shows that the warming
  approached below the surface from the west. The
  SST data on the other hand shows the surface
  warming moved west from the coast after the
  Kelvin waves reached the eastern Pacific and
  depressed the thermocline.
• By July, El Niño had fully developed. Note the
  similarities in sea-level and SST at this point,
  since the heating has mixed throughout the water.
  There are significant differences, though. Peak
  sea-level was to the west of peak SST, and there
  were much larger sea-level changes in the western
  Pacific than SST changes. All of these
  differences were due mainly to sub-surface heat
  changes which the SST data cannot detect.
• Both sea-level and SST continued to rise through
  the Fall. Both peaked in early December. Note
  the lobes of high sea-level forming north and
  south of the equator in December (indicated by
  arrows). These are indicative of another type of
  internal wave, a Rossby wave. Just like Kelvin
  waves were created when the ocean was adjusting
  to the formation of El Niño, Rossby waves are
  created as it dissipates. These waves move
  westward, and help re-adjust the thermocline in
  the west and east to normal.

23

• I pointed out the large sea-level anomalies in
  the Indian Ocean at the beginning of this
  presentation. Ill discuss this in more detail
  now. A Quicktime movie that is available
  separately shows the Indian Ocean fluctuations
  more clearly, but the figures below summarize it.
  Sea-level anomalies in the eastern Pacific began
  to move westward from the coast of South America
  in September. Shortly after this, sea-level
  increased suddenly and moved eastward again,
  indicating another Kelvin wave. At the same
  time, large sea-level anomalies developed in the
  Indian Ocean and began to move westward. By the
  end of 1997, sea-level in the southwestern Indian
  Ocean was as anomalously high as it was in the
  eastern Pacific. What was happening?

24
El Niño in the Indian Ocean
/s

• In October, there was a large eastwardly wind
  burst in the central Pacific. At the same time,
  there was a westwardly wind burst in the central
  Indian Ocean. Wind bursts in these two oceans
  are well correlated during El Niño events as
  shown in the bottom left figure. The wind bursts
  are associated with the Southern Oscillation, as
  shown in the bottom right figure. El Niño events
  are marked with arrows.

)
2
2
SOI
Zonal Wind Stress (m
Zonal Wind-Stress (m
Year
Year
Monthly wind stress anomalies, averaged over the
central Pacific (5N, 180E to 200E) and the
central Indian Ocean (5N, 80E - 100E).
Positive values are eastwardly (Pacific) or
westwardly (Indian). Data are from Florida State
Univ., Center for Ocean-Atmospheric Prediction
Studies.
Difference in Indian and Pacific eastwardly wind
stress, compared against the Southern Oscillation
Index (SOI). The SOI is determined from the
difference in pressure at Darwin, Australia and
Tahiti.
)
2
2
/s
25

• Recently, several studies have looked at SST
  warming in the southwestern Indian Ocean and the
  eastern Pacific, and have found that there is a
  significant correlation between anomalous warming
  in the Indian Ocean and El Niño (see figure
  below), although SST signals in the eastern
  Pacific are much larger. These studies (Tourre
  and White, 1995 and Nicholson, 1997) have found
  correlations between the Indian Ocean and the
  Pacific for almost every El Niño back to 1946. I
  have looked at the T/P altimeter data and see a
  similar correlation, but the extreme sea-level
  values are closer to the same magnitude (see
  below).

Sea Level Anomaly (cm)
Pacific SST anomaly (C)
Indian SST Anomaly (C)
Year
Year
T/P sea-level anomalies averaged over the same
region.
SST data averaged over southwestern Indian Ocean
and eastern equatorial Pacific. The scale for
Pacific SST anomalies is on the left hand side
the scale for Indian Ocean SST anomalies is on
the right.
26

• By combining all of these observations with the
  high resolution maps made from the TOPEX/Poseidon
  data, one can begin to see how the Indian Ocean
  warming is related to El Niño.
• The October wind burst in the Pacific forced a
  Kelvin wave in the eastern Pacific, which caused
  sea-level to peak for a second time off the coast
  of South America. The wind burst in the Indian
  Ocean created Rossby waves which moved westward
  in the Indian Ocean. By December, sea-level in
  the southwestern Indian Ocean was as high as
  sea-level in the eastern Pacific, and was a near
  mirror-image of El Niño in the Pacific. Similar
  variations have been observed during the 1994 El
  Niño. Based on the altimeter observations, it is
  beginning to look like El Niño has a very similar
  mode in the Indian Ocean.
• Although these results are preliminary, they show
  that there is still a lot about El Niño that we
  do not understand. Continued data from satellite
  altimeters, space-borne AVHRR, and moored buoys
  will all contribute to improving our knowledge.
  Numerical models will also play an important part
  in testing our theories, as I am sure Professor
  OBrien will discuss next week. Understanding
  this apparent El Niño mode in the Indian Ocean
  may help us explain teleconnections between El
  Niño and weather patterns in Asia, Africa,
  Europe, and other sites that are far removed from
  the Pacific.