From scientific requirements to data quality control: a technological support for research

16 October 2008, 14 pm | Giuditta Marinaro | Conference Room Rome | Headquarters | Seminars Section Rome 30

The scientific interpretation of data acquired in a terrestrial environment, especially during temporal monitoring, strictly depends on the reliability of the measurement system used and on the quality of the data itself. Quality control of a data should always precede scientific analysis. Upstream of the "clean" and validated data available to the researcher for his scientific analysis, there is therefore a complex instrumentation and final data control activity typically performed by a technologist.

The experience acquired in research based on complex monitoring in the marine environment has shown the need to operate with a set of rigorous procedures both for the choice and methods of use of the measurement instruments, and for the management and control of the data from the of its acquisition, up to its archiving and distribution to the end user. All these steps are carried out with close collaboration between the technologist and the researcher.

In the long-term submarine monitoring experiments, INGV has used "intelligent" units for the automatic acquisition and control of data measured by various instruments. This Local Data Management (LDM) it is carried out by units called DACS (Data and Acquisition and Control System) able to check the hardware and software status of the monitoring system, and perform an initial data quality check (e.g., outliers outside a "natural" range). At a later stage, after the recovery (physical download) of the data, procedures of Offline Data Management (ODM), performed by an operator, make it possible to verify that the data acquisition has been continuous for the entire duration of the experiment and therefore the stream of data for each parameter useful for the study of the monitored phenomena. In this phase, a further quality control is necessary to discriminate any "false" signals due to malfunctions of the monitoring system (electronic interference, disturbances of other sensors) from the "real" ones which will have to be studied.

Experience has shown that "false" signals can therefore occur in complex environmental monitoring systems which could lead to completely incorrect scientific interpretations. In many cases, these checks can only be performed if multi-parameter monitoring has been carried out, i.e. if data from different parameters are simultaneously available.