A procedure to ensure a good quality of signals recorded by multidisciplinary seafloor observatories

Important processes that affect the Earth, for example climate change and geohazards, are driven by phenomena that take place in the oceans. To better understand and mitigate the effects of these processes, an important international effort is taking place to deploy permanent and remotely controlled seafloor and water-column observatory systems. In Europe, large cabled systems with marine sensors are being developed for near real-time and real-time long-term monitoring of ocean processes. Many of these system are part of the EMSO (European Multi-disciplinary Seafloor and water column Observatory www.emso-eu.org) Research Infrastructure. A multidisciplinary approach based on different types of sensors, is necessary to understand complex natural phenomena. In fact, given a signal of interest, by using several sensors recording simultaneously it is possible to identify the contribution of different sources to this signal. On the other hand, obtaining good quality long-term continuous data from a variety of sensors installed on a seafloor observatory can be a challenging technical task. We consider long-term time series acquired by GEOSTAR class seafloor observatories deployed at two European offshore sites (Western Ionian sea in the Mediterranean and Iberian Margin-Gulf of Cadiz in the Atlantic Ocean), where there are important sources of geohazard. The observatories are either standalone or cabled, each with its particular technological difficulties and features. The quality and reliability of a signal depends on a chain of events defined by the following aspects: the sensor itself, the installation procedure of the sensor, the possible interaction of the sensor with nearby electronics or with other sensors (e.g. active acoustic current meters signal is recorded from hydrophones), the time stamping procedure, the architecture of data acquisition and transmission (or storage in the standalone case) system, and finally data processing to improve signal to noise ratio. The definition of what is signal and what is noise depends on the study of interest. For example, a seismologist interested in the signals that are caused by earthquakes considers environmental and anthropic effects as noise which would be gladly removed. On the other hand, if one is trying to study the seasonal variations of the seismic signal, then this contribution becomes the signal of interest. To correlate time series coming from different sites, campaigns or instruments, absolute calibration of sensors is obviously a critical issue. A rigorous estimate of timestamp precision is needed to achieve reliable results from multidisciplinary studies. In the present work we focus on the quality of the signal recorded by a broadband seafloor seismometer (from 0.003 Hz to 50 Hz), by a prototype gravimeter and on signals recorded by auxiliary sensors that are helpful in discriminating different noise sources. We first illustrate the steps that were taken to obtain a good quality reliable signal and finally we focus on how, thanks to a multisensor approach, we can identify the main noise sources, such as sea currents and temperature variation in the long period part of the spectrum. It is possible, in principle, to reduce this noise through signal processing techniques. The improvement of the S/N with such procedures allows for a more efficient detection of interesting seismic signals. We show some examples of this signal improvement procedure.