WP 2.1 Pattern search in multiparameter seismic data

Start date or starting event:


Lead contractor:





Objectives: The occurrence of earthquakes is a complex and highly variable  process coupled in space and time. The resulting difficulty to  separate superimposed seismicity patterns stemming from different  causes impedes the search for anomalies possibly  preceding large earthquakes [39]. It has been demonstrated that by a novel approach of Spatio-Temporal Principal Components  Analysis (PCA), developed byn the partner, that it is possible to address this problem: in a first  application to real data it was possible to unambiguously separate  background activity and different individual event sequences for  the first time. Furthermore, some constitutents seemingly showed  precursory qualities.  Within the ideal setting of PREPARED project the objectives are twofold:  confirmation and enhancement of the method on the one hand,  application to seismicity in the SISZ on the other to detect long- term premonitory changes before the large events in 2000 and to  search for precursors of impending events.


Methodology / work description: The SIL-data since 1991 are of extraordinary quality for this work, with a  magnitude completeness seemingly almost down at 0. Within the framework of  PREPARED it will be easy to rule out catalogue inconsistencies  which might otherwise lead to spurious results. Close  collaboration with WP 2.2 will further aid in this and  possibly allow utilization of seismicity data from before 1991.  After an initial analysis of the complete dataset using existing  software, initial results, geological knowledge and data from  other workpackages will be incorporated to vary analysis  parameters, enhancing the software in the process. Correlation  between observed anomalies and the occurrence of large events will  finally lead to the establishment of a relation between anomaly  characteristics and earthquake parameters.

Work description: Work in this package will basically comprise familiarization with  the existing earthquake catalogue data and subsequent PCA. The  first part must also take geological and geophysical constraints  such as areas of maximum strain into account, the latter part  involves variations in the preparation of the input data as well  as enhancement of the existing software. The final step involves  the quantification of the PCA results and interpretation of the  results with respect to their precursory qualities. PCA of the existing data will commence immediately, if necessary  by setting an artificial threshold for the minimum magnitude. With  the  software existing so far, we will be able to directly analyze  epicenter distributions of the SISZ and appropriate subsets  thereof, identified by geological knowledge and initial analysis  results of the whole area. Depending on results, variations mainly  in the preparation of the input data will be carried out: spatial  resolution of cells for data gridding, duration of temporal  windows for temporal discretization, time shift of windows etc.  PCA itself possesses no free parameters which is one of the major  advantages of the approach. An extension of the method to include  hypocenter depths information will be carried out. The objective  delineation and quantification of decomposed constituents  (coherent patterns) is a necessary final step which has not been  addressed in the qualitative interpretation up to now. The latter  step will utilize the functionality of a geographical information  system (GIS). We plan the analysis of parameters such as Benioff  strain in addition to earthquake rates, also in view of testing  the hypotheses of growing correlation length and accelerated seismic moment release  which  are regarded as promising precursory phenomena.


Deliverables including cost of deliverable as percentage of total cost of the proposed project:


Application of PCA to SIL-data, emphasizing computational statistics.

M10   Re  PU  0,5%


Application of PCA to SIL-data, emphasizing computational statistics.

M12   Re  PU  0,5%


Application of PCA to SIL-data, emphasizing seismology.

M22   Re  PU  0,5%


Application of PCA to SIL-data, emphasizing seismology.

M24   Re  PU  0,5%


Release a software package for PCA analysis of seismicity.

M24   O   PU  1,0%



Milestones: Delivery of the above items at the date indicated.