Fig. 1PROCESSING AND ASSIMILATION
FOR EL NINO FORECASTS
Bob CHENEY, Laury MILLER, C.K. TAI, John LILLIBRIDGE, John
KUHN
NOAA Laboratory for Satellite Altimetry
Silver Spring, Maryland 20910
Ming JI, Dave BEHRINGER
NOAA National Centers for Environmental Prediction
Washington DC
ABSTRACT - The Topex/Poseidon (T/P) altimeter satellite, launched in 1992 as
a research mission of the American and French space agencies, has recently
become an integral part of NOAA's operational satellite system for monitoring the
oceans. The transformation was achieved through the joint efforts of NOAA, JPL,
and NAVOCEANO. Since late 1996, this team has produced accurate T/P sea level
observations with a delay of only two days--fast enough to be included in NOAA's
weekly ocean model run. Operational assimilation began in March 1997. The
T/P data improve both the ocean initial conditions and the sea surface temperature
forecasts with lead times of up to 6 months.
1 - INTRODUCTION
El Nino-related climate variations often have widespread and
devastating impacts. These include the frequency and severity
of storms and the occurrence of droughts and floods. In the
U.S. alone, business losses associated with 1986-87 El Nino
amounted to $10-15 billion. Although many of the
consequences of El Nino cannot be prevented, skillful
forecasts enable resource managers in climate-sensitive
sectors to alter strategies and reduce economic vulnerability.
Coupled ocean-atmosphere models have attained significant
forecast skill during the past decade, but they continue to be
limited by an ocean observing system which is extraordinarily
modest. Satellites represent part of the solution. In particular,
an operational flow of altimeter data has long been desired by
the modeling community as a means of estimating changes in
upper ocean heat (to first order, sea level is determined by the
integrated temperature of the water column). But even though
altimeters have flown nearly continuously since 1985, two
challenges have stood in the way of progress: (1) The
altimeter data must be made available fast enough (1-2 days)
and with sufficient accuracy (a few cm) to track changes in the
ocean within a tolerance that is useful for the ocean model;
(2) The assimilation method must be capable of using a single
parameter, sea level, to correct the model temperature as a
function of depth.
Using T/P altimeter data, both of these problems have been
solved through a collaboration among NOAA, the Naval
Oceanographic Office (NAVOCEANO), and the Jet
Propulsion Laboratory (JPL). The program became
operational in March 1997, just in time to assist in the long-range
forecasts associated with the ongoing 1997 El Nino.
2 - TOPEX/POSEIDON DATA PROCESSING
Data downlinked from the T/P satellite to JPL are transmitted continuously to NAVOCEANO (Fig. 1) where interim geophysical data records (IGDRs) are prepared. These data are similar to the final T/P GDRs except that the satellite orbit is a prediction and contains radial errors of several decimeters. However, the problem of quickly determining accurate satellite orbits has been solved by the Global Positioning System (GPS). Lichten et al. (1996) have developed an automated orbit determination program at JPL which computes accurate (5 cm rms) T/P orbits with a delay of only 1-2 days. These orbits are provided to NAVOCEANO and updated IGDRs are forwarded to NOAA, where an adjustment is performed to further reduce residual orbit error.
Fig. 1Finally, sea level deviations are computed at 1-degree latitude intervals along the T/P track. Based on an overlap in time with the more accurate GDRs, the IGDR values are referenced to the 1993-95 mean. The sea level time series shown in Fig. 2 is for a 1-degree segment near the Galapagos Islands. Agreement of the 2-day IGDR and the final GDR is at the level of a few cm. The rising sea level here is associated with the 1997 El Nino.)
Fig. 2Based on a 15-cycle (5-month) overlap, Fig. 3 shows the rms difference between the GDR and IGDR at each 1-degree segment of the ascending ground tracks. The total delay in the operational T/P system is only 2 days, yet agreement is usually 2-4 cm. Accounting for uncertainties in the GDR itself, total error of the 2-day IGDR analysis is estimated to be 4-5 cm.
Fig. 33 - IMPACT OF TOPEX ON THE MODEL
At NCEP, the Topex data are assimilated in an ocean model together with subsurface temperature profiles from XBTs and TAO moorings. The altimeter assimilation scheme preferentially corrects the model temperature field where the probability of its being in error is greatest, making those corrections in such a way as to bring the model sea surface height into closer agreement with the Topex data.
Figs. 4 and 5 below show show the impact of the altimeter data on the model surface dynamic height during 1993-96 compared to a parallel run without Topex. Only small height differences were seen during the initial years, but in 1996 Topex produced substantial changes (which were verified by tide gauge data). It was found that salinity anomalies developed during 1996 which were not accurately represented in the model. The Topex assimilation resulted in a more accurate picture of the tropical ocean surface topography and circulation.
Fig. 4
Fig. 54 - IMPACT OF TOPEX ON PREDICTION
The value of the altimeter data assimilation in the coupled
model system can be tested by examining the skill with which
Pacific sea surface temperature (SST) can be predicted.
For the 3-month period Dec 96 to Feb 97, Fig. 6 below shows
observed SST (top) and two forecasts made 6 months in
advance: the first with Topex and the second without.
Only the Topex version of the model was able to accurately
predict the cold tongue in the eastern Pacific.
Fig. 6
5 - CONCLUSIONS
These results suggest that predictions of Pacific SST made with Topex ocean initial conditions have higher skill than those made with XBT and TAO data alone. Topex data analyzed within 2 days of real time have therefore been added to the operational assimilation scheme implemented at NOAA NCEP.
Shown in Fig. 7 are SST predictions for 3-month periods through June 1988. The model upon which these forecasts are based includes Topex data through September 1997. The ongoing El Nino, the strongest since the dramatic 1982-83 warm event, should reach a maximum in late 1997 and then weaken by mid-1988.
Fig. 7The Laboratory for Satellite Altimetry will participate in three upcoming satellite altimeter missions, all of which will have operational aspects: the Navy's Geosat Follow-On, the CNES/NASA Jason-1, and ESA's Envisat. Based on the success of the T/P assimilation project at NCEP, near-real time processing of altimeter data will continue to be a priority.
REFERENCES
Lichten, S.M., B.J. Haines, R.J. Muellerschoen, Y. Vigue, and T. Munson, Rapid service precise orbit determination capability from GPS for altimetry missions, EOS, 77(46), F129, 1996.