ARTIFICIAL AND NATURAL CHANGES IN THE RATES OF SEISMIC ACTIVITY - A CASE-STUDY OF THE GARM REGION, TAJIKISTAN (CIS)

Seismicity rate variations observed in earthquake catalogues can be natural, i.e. associated with temporal variations in the seismic process, or they can be artificial, i.e. related to changes in the networks of seismic stations used to create these catalogues. Such artificial changes must be identified and corrected before real variations can be reliably examined. This paper demonstrates techniques for modelling of the observed changes in seismicity rates in order to determine their origin. The main technique used here has been previously applied to a number of catalogues to identify times of rate changes. It can be used to demonstrate the magnitude dependence of these changes and the existence of two major types of possible artificial rate variations, detection changes and magnitude shifts. In addition, two new techniques are applied. One of them, the variable window technique, makes use of windows of different sizes to sample the data, thus addressing some of the shortcomings of the original approach. The other new technique, catalogue randomization, helps analyse the cause of some of the rate changes; it is most helpful in the cases of pure magnitude shifts. The use of these techniques is demonstrated examining data from the Garm region, Tadjikistan (former Soviet Union). The Garm earthquake catalogue includes more than 85 000 events for the period 1955 January-1989 April and covers an 80 km x 100 km area in the collisional zone between the Indian and Eurasian plates. Known changes in the Garm network, recently documented in detail, were compared with the rate changes identified with the above techniques. Of the 18 main rate variations identified, at least two-thirds were found to be artificial. Most of them were readily associated with reported changes in the network, such as closure and opening of stations, and/or changes in instrumentation. Of the rate changes determined to be natural, most prominent were the ones associated with a possible quiescence preceding a M6.3 event and its aftershock sequence. Finally, other authors' reports of linear trends and annual periodicity in the seismic activity at Garm are critically examined.