Physical Oceanography, Vol. 18, No. 1, 2008
STATISTICAL STRUCTURE OF THE LARGE-SCALE FIELDS OF
TEMPERATURE AND SALINITY IN THE BLACK SEA
A. B. Polonskii and I. G. Shokurova
In order to reconstruct the large-scale temperature and salinity fields by the method of optimal
interpolation of the archival data, we compute the correlation functions and analyze the space
and time variations of the statistical structure of the fields. On the sea surface, the thermohaline
fields are spatially inhomogeneous. Thus, the correlation functions are anisotropic in the region
of the northwest shelf and close to isotropic in the inner parts of the sea. The values of correla-
tion length vary from season to season. In the layer of pycnocline, the temperature and salinity
fields are anisotropic. In the zonal direction, the correlation length is
times greater than in
the meridional direction. The indicated anisotropy becomes stronger in the winter season and
weaker in the summer season as a consequence of the seasonal variability of large-scale circula-
tion. We study the dependence of the error of reconstruction of the fields by the method of opti-
mal interpolation on the form of approximation of the correlation functions with regard for an-
In reconstructing climatic fields for which the available data of measurements are insufficient, an important
role is played by the proper choice of the method of interpolation. As one of the most efficient methods used for
the numerical analysis of fields on a regular grid, one can mention the method of optimal interpolation [1–3] de-
veloped and extensively used for the investigation of the meteorological fields. In view of the specific features
of the mutual location of stations, this method guarantees the maximum accuracy of reconstruction of the fields
and gives much better results for a sparse network of stations than the other methods of interpolation [1, 2].
The application of this method requires the preliminary analysis of the statistical structure of fields, i.e., the
evaluation of the correlation functions. The complexity of investigation of the statistical structure of oceanologi-
cal fields for a small amount of data and their asynchronous character is the main cause of the insufficiently ex-
tensive application of the method of optimal interpolation in oceanography. At the same time, in numerous
works, this method is successfully used not only for the reconstruction of hydrological fields according to vari-
ous collections of the data of observations [3–6] but also for the purpose of optimal arrangement of the network
of stations in planning hydrological surveys and for the assimilation of data in the process simulation [7, 8].
The correlation functions of the temperature and salinity fields in the Black Sea are computed in [3, 9, 10]
according to the data of several surveys or for certain seasons. In all these works, it is indicated that the thermo-
haline fields are are anisotropic. The problem of approximation of the correlation functions is studied in [3, 10].
At present, the amount of data accumulated for the Black Sea is quite large. These data enable us to systematize
and generalize all available information on the statistical structure of the temperature and salinity fields for all
In the present work, the space one- and two-dimensional autocorrelation functions of temperature and sali-
nity fields are computed according to the many-year data accumulated for February, July, August, and October
Marine Hydrophysical Institute, Ukrainian National Academy of Sciences, Sevastopol, Ukraine.
Translated from Morskoi Gidrofizicheskii Zhurnal, No.
51–65, January–February, 2008. Original article submitted September 1,
2006; revision submitted October 25, 2006.
38 0928-5105/08/1801–0038 © 2008 Springer Science+Business Media, Inc.