FY 1986

An assessment of the XBT sampling network in the central Pacific

Kessler, W.S., B.A. Taft, and M.J. McPhaden

Univ. Corp. Atmos. Res., USTOGA 4, 62 pp (1985)

Since 1979, a joint France-U.S.A. program has deployed expendable bathythermographs (XBTs) from merchant ships on a nearly meridional track crossing the equator at about 160°W in the central Pacific. The average sampling rate is 1.3 probes per degree latitude per month, and the probes measure to 450 m. In addition, surface temperature and salinity have been measured on these lines since 1975. In order to assess whether there is over-sampling, the data sets were subsampled and isotherm depth, dynamic height, zonal geostrophic velocity, geostrophic volume transports and surface temperature and salinity were computed from both the entire data set and the subsamples. The subsamples investigated are halves of the data chosen to approximate the situations in which: 1) half the cruises were made; 2) each cruise deployed an XBT half as frequently; and 3) each probe sampled only the upper 200 m. Either of the first two cases produces estimates of the variables studied which give an adequate picture of the massive El Niño event of 1982-83, but poorly resolve the weaker seasonal signals of ordinary years. In the third case, 450 m profiles were reconstructed by removing the bottom 250 m of each cast and replacing it with a mean temperature profile. The variables computed from the resulting reconstructed casts differ only slightly from those computed using the entire data set. In addition, analysis of data from the FGGE/Norpax Hawaii-Tahiti Shuttle shows that more than 80% of the 0/1000 m dynamic height variance is contained in the upper 450 m. The surface data resolve both seasonal cycles and El Niño signal with only one fourth of the observations. We make the following recommendations. First, reduce the surface sampling by one half; second, continue using the 450 m XBT; third, do not reduce central Pacific XBT sampling. We also recommend that a study of this type be carried out for the western Pacific XBT line and that studies of the signals of seasonal and interannual variation below the present XBT sampling depth of 450 m be done. The present work has only considered the effect of subsampling on the measurement of certain oceanographic variables and has not dealt with the impact of sampling density on modeling and diagnostic studies of the thermal field, which may place more stringent requirements on the data set.