The statistics of a pair of Lagrangian particles offer, in principle, a pos
sibility to estimate the structure functions of velocity, then the spatial
autocorrelations, and finally the spatial spectra. On the basis of this str
ategy, the authors have developed an approach to estimate spatial spectra o
f mesoscale horizontal turbulence in the ocean using data of satellite-trac
ked drifters. The approach was applied to the data of 19 drifters deployed
in the California Current System in 1993. It was found that the shape of bo
th those spectra and this spectra calculated using drifterborne longitudina
l and transverse correlations estimated by other authors are qualitatively
in good accordance with theoretical predictions for 2D isotropic nondiverge
nt turbulent flow. To relate obtained spectra to some physical parameters,
kinematic stochastic models were developed that consisted of a population o
f randomly spaced, 2D axisymmetric eddies of a given shape. Numerical exper
iments with different eddy shapes showed that the model spectra obey a self
-similarity; that is, at a given eddy shape they depend on the variance of
stochastic process and a length scale of the eddy only. A model with the ex
ponential eddy shape was Found to fit drifterborne spectra better than othe
r models. The best agreement between the drifterborne and model spectra was
achieved when the radius of an exponentially shaped model eddy was taken e
qual to the internal Rossby radius.