Аннотация:At present, the deep-sea measurement method is one of the main ones for earthquake and tsunami detection. The main advantage of deep-water tsunami detection stations in comparison with other methods is the possibility of earlier registration of the wave and the almost complete immunity of the whole system in relation to the impact of the tsunami. Several hundred seafloor observatories equipped with ocean bottom seismometers (OBS) and pressure gauges (PG) were installed in the oceans all over the World. DONET (Dense Ocean-floor Network system for Earthquakes and Tsunamis) (Kaneda et al., 2015; Kawaguchi et al., 2015), S-net (Seafloor Observation Network for Earthquakes and Tsunamis) (Kanazawa, 2013), NEMO-SN1 (NEutrino Mediterranean Observatory Submarine Network 1) (Favali et al., 2013), NEPTUNE (North East Pacific Time-series Underwater Networked Experiments) (Barnes and Team, 2007), MACHO (MArine Cable Hosted Observatory) (Hsiao et al., 2014), this is not a complete list of such systems. Sensors (PG, OBS) are normally exploited for many years. In order to make sure a sensor provide accurate measurements of a physical value it is useful to periodically calibrate sensor. Since the sensors are located at depth of a few thousand meters, their replacement is not only a rather complex technical problem but also a very expensive procedure. A method for examining the performance of seafloor observatory sensors was proposed and substantiated in our works (Nosov et al., 2018, Karpov et al., 2020). The only we need for the calibration is a rather strong seismic event recorded by the couple of sensors. The method is based on variations of the bottom pressure 𝑝 and the vertical acceleration component 𝑎𝑧 of the ocean bottom motion being related linearly: 𝑝 = 𝑚𝑎𝑧 where 𝑚 is the mass of a water column of unit cross section. This relationship is valid within the frequency range 𝑓𝑔 < 𝑓 < 𝑓𝑎𝑐 , where 𝑓𝑔 = 0.366√𝑔/𝐻, and 𝑓𝑎𝑐 = 𝑐/4𝐻, 𝑔 is the gravity acceleration and 𝑐 is the speed of sound in water, 𝐻 is the ocean depth. The main point of the method for examining the performance of seafloor observatory sensors consists in finding the ratio of the power spectra of pressure and z-acceleration variations registered during an earthquake (Karpov et al., 2020). In the case of correct calibration of sensors in the frequency range 𝑓𝑔 < 𝑓 < 𝑓𝑎𝑐 the ratio of the spectra should be a constant value equal to 𝑚2 . In refs. (Nosov et al., 2018, Karpov et al., 2020, Nosov et al., 2021) we have shown it to be advisable to base accurate calculations on the value measured by PG, i.e. 𝑃 the total pressure averaged over time, and on the relationship of hydrostatics, 𝑚 = 𝑃/𝑔. In our presentation we demonstrate an algorithm for estimating the sensor calibration accuracy, which would permit to automatically provide a quantitative estimation of the calibration accuracy of sensors of seafloor observatories, or a conclusion asserting it to be impossible to perform a test for objective reasons. We also present the results of applying the algorithm on the DONET data recorded during a number of seismic events.