Purpose. The purpose of the work is to develop a regional statistical methodology for deriving the Black Sea thermohaline structure based on satellite data on sea surface temperature and sea level.

Methods and Results. The study included in-situ temperature and salinity data, daily anomalies of altimetry sea level (0.125° grid) and sea surface temperature (0.05° grid) at the L4 processing level for 1993‒2023, as well as a climatic array of the Black Sea thermohaline fields for 1950‒2023. To reproduce the hydrological fields, the method of multiple linear regression was used. It made it possible to determine statistical relationships between the physical parameters of the sea surface layer and the values of temperature and salinity at depths up to 300 m. In order to take into account long-term changes, corrections to modern climatic values and linear trends were applied. Validation of the methodology through comparison of the calculated 30-year reanalysis and the in-situ measurement data showed that the total root-mean-square errors of temperature and salinity reconstruction were 0.4 °C and 0.2 psu, respectively, which corresponded to an average normalized error of 0.4, indicating a good result for a statistical method. The errors were highest in the seasonal thermocline and permanent halocline layers, where the amplitude of mesoscale noise increases sharply. The best values of calculation accuracy were observed in the cyclonic gyres in the central sea, while the worst ones were found in the sea periphery (Batumi anticyclonic eddy and Rim Current zone), and, as for salinity, on the north-western shelf. A comparison of the error values of several arrays of the Black Sea thermohaline field reanalyses showed that the accuracy of this method is close to that of the CMEMS arrays calculated using modern hydrodynamic models and complex data assimilation methods. Some indicators relating to salinity show that, despite the absence of an in-situ data assimilation algorithm in this version, the regression model even exceeds the accuracy values of the specified reanalyses.

Conclusions. A method for operational diagnosis and reanalysis of the Black Sea 3-dimensional thermohaline structure based on satellite data and using regression relationships has been developed. It provides a sufficiently good level of errors in temperature and salinity reconstruction while minimizing computational resources. The methodology implies the application of regular open-source data, requires no high-performance equipment (which is of practical importance), and can be regarded as a quick and effective instrument for assessing the state of the marine environment. The regression algorithm can also be used in operational data assimilation procedures in forecast models when in-situ oceanographic data are absent.