ANNUAL CYCLES OF ERS-1 ALTIMETRIC SEA SURFACE HEIGHT DATA AND ATSR SEA SURFACE TEMPERATURE DATA

Abstract

Along Track Scanning Radiometer (ATSR) data from the ERS 1 satellite mission are used in a global analysis of the surface temperature (SST) of the oceans. Detailed analysis of the annual cycle of the SST is performed and the result is compared to the annual cycle in the sea level anomalies (SLA) for the same period, as seen from the first 35-day repeat mission (phase C) of the ERS-1 altimetry.

The amplitude of the annual signal of the SST has the largest values in the latitude belt from 25 N to 50 N with maximum amplitudes of 5­6 C off the east coasts of the continents. In this band the amplitudes of the annual cycle of the SLA and SST varies similarly. Maximum amplitudes of the seasonal cycle of 9­11 cm is found in the regions of largest amplitude in the SST on the northern hemisphere. On each of the hemispheres the surface temperatures reach their maximum after summer heating. The seasonal sea level variability, as observed from ERS-1, reaches its maximum one months later. On the southern hemisphere the largest amplitudes in the annual signal of SLA is found further to the south than the corresponding maximum in the SST

Keywords: ATSR, Altimetry, Annual Cycles

Introduction

The ATSR instrument onboard ERS-1 was designed to provide:

1) Sea surface temperature (SST) images with a 1 km resolution, 500 km swath and a relative accuracy of 0.1K and

2) SST with an absolute accuracy of better than 0.5 K and a spatial resolution of 50 km, in conditions of up to 80% cloud cover [ESA, 1992].

The instrument uses spectral channels very similar to those on the recent NOAA satellites, with many improvements in accuracy. Early validations of the ATSR SST data have shown that the ATSR data are consistent with the 0.5 K accuracy objective [Forrester et al., 1993]. Furthermore, Barton et al. [1994] demonstrated the increased accuracy of the ATSR SST products over the AVHRR data in a comparison to in situ measurements.

In this analysis the low resolution ATSR data from the ERS 1 satellite mission are evaluated with respect to a recovery the spatial and the temporal characteristics of the annual cycle of the SST. Finally, the seasonal variability of the SST are compared with the seasonal sea level height variability obtained FROM the same satellite.

The 50 km resolution, or more precisely the 0.5 by 0.5 average temperatures from April 1992 through December 1993 are used. The 500 km swath scanning across the satellite ground track results in an almost global coverage every three days. Unfortunately, the 3.7 µm sensor failed in May 1992 [Llewellyn-Jones, 1994], which means that some erroneous night time data are left undetected. These erroneous temperatures are typically caused by fog, and are about 5-12 K below the normally observed SST [Jones et al., 1995]. Therefore a procedure for the detection of the erroneous data, is needed. This procedure removes an observation if the temperature, relative to a local mean and annual variation, is outside a range of three times the local standard deviation. Note, that the editing criteria are computed using day data only [Knudsen et al., 1996].

The ATSR SST data were averaged in cells of one degree latitude by five degrees longitude respectively, in time intervals of 10 days. A time interval of 10 days was choses in order to enable a more precise phase estimation than was allowed by the 35 days repeat period. Having chosen the time interval to 10 days, the spatial dimension of each cell were chosen such that at least two measurement are made in each cell within the time interval of 10 days. One on a descending track and one on an ascending track. Subsequently, SST anomalies were formed through a removal of the mean temperature in the respective cells.

The ATSR sea surface temperature data were chosen in the same time interval as the sea level height anomalies covering the first 35-day repeat mission (phase C) of the ERS-1 satellite. This corresponds to almost 20 months observation lasting from April, 10, 1992 until December, 31, 1993. The altimetric sea level data was obtained as one second normal point data from the NASA ocean altimeter pathfinder altimetry, which was recently released on CD-rom. In addition to the standard corrections (altimeter sensor effects, electromagnetic bias, ionosphere, wet troposphere, and dry troposphere) also a timing tag bias correction of 1.7 msec as well as a 100 % inverse barometer correction was applied [Callahan, 1993] in order to avoid effects of seasonal changes in the atmospheric pressure. Geophysical corrections for ocean tides, earth tides, and pole tides were applied. The precise orbits were computed with the DGM-E04 gravity field model [Scharroo et al., 1997], and a new ocean tide correction based on the UT/CSR 3.0 model was applied. Subsequently, the altimeter data were processed using a similar averaging as for the ATSR data forming the sea level anomalies (SLA).

Figure 1 Amplitude (upper) and phases (lower) of the annual cycle of sea surface temperature from ERS-1 ATSR. Data from first 35-day cycle (phase C) has been used.

The Annual cycles

Global maps of the amplitudes and phases of the annual cycle of the SST are shown in Figure 1.The amplitudes of the annual signal (upper panel) have the largest values in the latitude belt from 25 N to 50 N. Off the east coasts of the continents where continental air is moved eastward by the predominantly westerly winds, maximum amplitudes of 6-7 C are found. Similar values are seen in the Mediterranean Sea. The largest amplitudes on the southern hemisphere are found between latitudes 30 S and 40 S where values between 2-4 C are found.

Spatially, the phases of the annual cycle of the SST (Figure 1 lower panel) are generally shifted by 180 between the two hemispheres. On the northern hemisphere maximum temperatures occur in late August and early September, and on the southern hemisphere maximum temperatures occur in late February to early March. Off the east coasts of the continents the phases are slightly smaller than further east. A detailed study of the phases at mid latitudes in the North Pacific ocean show that the phase changes by about 20 days from late August to mid September going east from the coast of Japan towards the centre of the ocean. A similar phase change, though smaller, is found in the North Atlantic ocean.

The annual cycle of the SLA is shown in Figure 2 and display features quite similar as earlier analysis of the annual cycle from satellite altimetry [e.g., Knudsen, 1994; Zlotnicki et al., 1989]. In both mid-latitude belts the amplitudes (Figure 2, upper panel) vary similarly to the amplitudes of the annual cycle of the SST. Off the east coasts of the northern continents the highest amplitudes are found (9-11 cm). On the southern hemisphere the largest amplitudes (4-5 cm) are found between latitudes 30 S and 40 S. Note, that this is slightly further to the south than the corresponding maximum in the SST.

Figure 2 Amplitude (upper) and phases (lower) of the annual cycle of sea level anomalies from ERS-1 altimetry.

Data from first 35-day cycle phase C) has been used

The phases of the annual cycle of the SLA (Figure 2 lower panel) in the two mid-latitude bands, are shifted by a little more than one month relative to the phases of the annual cycle of the SST. Also the high SLA values off the east coasts appear to reach their maximum slightly earlier than further east in the middle of the oceans. At mid latitudes in both the North Pacific ocean and the North Atlantic ocean the phases change from late September to early October.

In the equatorial regions the annual signal from the SLA is much more detailed than the annual cycle of the SST. Here the change in the sea level height depends more on seasonal changes in the winds [e.g., Gill and Niiler, 1973; Knudsen, 1994].

A detailed investigate of the amplitudes and phases of the sea level height and the sea surface temperature were performed in two mid latitude bands. The results of these are found in Figure 3 for the northern mid latitude band and in Figure 4 for the southern mid latitude band.

On average in the northern mid latitude band (25 N-50 N), the amplitudes of the annual cycle are 4.6 C and 6.4 cm for the SST and the SLA, respectively. The corresponding cosine phases are 7.8 and 8.8 months, respectively. Hence, the amplitude ratio is about 1.4 cm/ C and the phases differ by 1.0 months. In the southern mid latitude band (25 S-50 S) the amplitudes of the annual cycle are 2.5 C and 2.5 cm for the SST and the SLA, respectively. The cosine phases are 1.9 and 4.1 months, respectively. Here the amplitude ratio is about 1.0 cm/ C as in the northern latitude band. The phases differ by 2.2 month.

Figure 3 Time series of SST (circles) and SLA (crosses) averaged over the northern latitude band between 25 N and 50 N. The estimated seasonal cycle is shown with solid line for the SST and with dashed line for the SLA.

Figure 4 Time series of SST (circles) and SLA (crosses) averaged over the southern latitude band between 25 S and 50 S. The estimated seasonal cycle is shown with solid line for the SST and with dashed line for the SLA.

The SST values agree remarkably well with Shea et al. [1991] who found maximum SST on the northern hemisphere to occur in late August to early September which corresponds to a phase of 8 months. On the southern hemisphere Shea et al. found maximum SST in late February corresponding to a phase close to 2 month

Conclusions

A global analysis of the seasonal cycle of the SST has been investigated and compared with the seasonal cycle of the sea level height variability obtained from almost 20 month of the 35-day mission (phase C) of the ERS-1 satellite.

At mid-latitudes, considering a phase difference of about one months, the correlation is very high. The correlation is poor in the equatorial regions where the sea level variability depends more on the wind. Also on shorter length and time scales the ATSR data are expected to provide valuable information about the ocean dynamics [Jones et al., 1995]. Such analysis has previous been presented using Geosat altimetry and AVHRR data by Vazquez et al. [1990].

Acknowledgments. This analysis is a contribution to a project supported by the Danish Space Board. The authors wish to thank ESA/ESRIN for the ATSR data. B. Beckley Hughes STX Corporation provided the ERS-1 sea level anomalies from the NASA ocean altimeter pathfinder CD-rom.

References

Barton, I.J., C.T. Mutlow, A.M. Zavody, and D.T. Llewellyn-Jones 1994., Intercomparison of AVHRR and ATSR data and data products, Proc. Second ERS­1 symposium, Space at the Service of our Environment, Germany, ESA Spec. Publ. 361, 1099-1102.

Callahan, P, 1993, TOPEX/POSEIDON project GDR users handbook, Rep. JPL D­8944, rev A, 84 pp., Jet Propuls. Lab., Pasadena Calif.

European Space Agency (ESA) 1993, ERS user handbook, ESA Spec. Publ. 1148, 128pp.

Forrester, T.N., T.H. Guymer, and P.G. Challenor 1993, Preliminary validation of ATSR sea surface temperatures near the Faeroes, Proc. First ERS­1 symp., Space at the Service of our Environment, ESA Spec. Publ. 359, 807-814.

Gill A.E. and P.P. Niiler 1973, A theory of the seasonal variability in the ocean, Deep Sea Res., 20, 141­177.

Jones, M.S., M.A. Saunders, and T.H. Guymer 1995, Reducing cloud contamination in ATSR averaged sea surface temperature data, J. Atmos. Oceanic Technol., in press.

Knudsen, P. 1994, Global low harmonic degree models of the seasonal variability and residual ocean tides from TOPEX/POSEIDON altimeter data, J. Geophys. Res., 99, 24643­24655.

Knudsen, P., O.B. Andersen, and T. Knudsen 1996, ATSR sea surface temperature data in a global analysis with TOPEX/POSEIDON altimetry, Geophys. Res. Lett., Vol. 23, No. 8, 821-824.

Llewellyn-Jones, D.T. 1994, Water surface temperature, Proc. Second ERS­1 symposium, Space at the Service of our Environment, Germany, ESA Spec. Publ. 361, 1079-1080.

Scharroo, R., P Visser, and G. Metz, TOPEX class orbits for the ERS satell