Monitoring volcanic and geothermal areas by full seismic moment tensor inversion: are non-double-couple components always artefacts of modelling?

Non-double-couple mechanisms in volcanic environments are indicators of loc al modifications of the stress field induced by dyke injection, high fluid pressure or by thermal cooling. However, the possible biasing effects of wa ve propagation in structurally complicated regions, the presence of noise, or inappropriate station coverage can make the identification of non-double -couples as true source phenomena uncertain. In this paper we review possible sources of false non-double-couples, and w e estimate the reliability of the non-double-couple solutions on the basis of an error analysis that includes the variance of the modelling and of the noise in the data. Our analysis of synthetic and real data shows that we c an identify, within confidence levels of 99 and 95 per cent, respectively, spurious non-double-couples due to inadequate station coverage, to noise in the data and to inadequacies of the structural model. Real data analyses m ust be preceded by synthetic tests in order to define the lower limits from which the nondouble-couple components can be considered statistically sign ificant at a given confidence level.