SRA gridification guide
From EGEE-see WIki
Seismic Risk Assessment (SRA) is very important for public safety and hazards mitigation. It is also important for the correct determination of earthquake insurance premiums, and for understanding the social and psychological effects of earthquakes. The goal of this study is to develop a Grid-based application for the SEE region, with the acronym SRA, to allow embedding alternative Seismic Hazard Assessment (SHA) models such as deterministic and probabilistic. It aims to produce SHA figures and maps for the region. SRA needs the Grid infrastructure not only to access the earthquake catalogue data, but also to execute the SHA models which may require high amount of resources depending on the model, parameters and area of the site selected by a user. As the output of its execution, SRA will provide assessment figures and map to be used in social, psychological studies, determination of insurance premiums, construction of utility systems and transportation infrastructures.
Basic elements of SRA could be introduced as (I) Web-based GUI, (ii) Method to access a uniform earthquake catalogue, (iii) Data repository keeping the seismology-related data that SHA calculations needs, (iv) Master earthquake seismic source model, (v) Alternative SHA models, and (iv) Engine where the actual SHA calculations are performed.
The GUI of SRA, developed as a portlet for the P-GRADE Portal, runs a Google-powered map, which interacts with users to input the coordinates of the site for which the risk assessment calculations are requested and to output the resulting risk map for that region after the execution of the selected SHA model (Figure 1). Additionally, this GUI enables users to manage all the parameters of SHA, to access the results of the previous runs for the regions around the selected site, to submit the risk assessment calculations onto the Grid, to monitor them, and to get the resulting assessment figures at the end of the executions.
Figure 1. Web-based GUI of SRA
The Administration portlet of the P-GRADE Portal enables the application administrator to handle the user management issues. Meanwhile, SRA users only need to access the Certificates and added SRA portlets to utilize all the functions of this application; hence, the other portlets could be disabled for the SRA users using the Administrator functions of the portal.
Once the site and SHA model to apply is selected together with the other SHA parameters by a user, a job holding the SHA engine is formed and submitted onto the Grid. An input file is prepared and fed into this engine, keeping the site coordinates, SHA model and all the model parameters. The job is formed in the background and not made accessible through the Workflow portlet or its editor of the portal.
The users' certificate proxy management, required for job submissions, is handled through the Certificates portlet. Provided that a certificate proxy is downloaded from a MyProxy server using this portlet, the SHA jobs are submitted and monitored with the help of the gLite job management commands. These CLI commands are called from the scripts connected to GUI elements (mostly buttons) on the SRA portlet. If it is the case that selected site is divided into sub-regions, for each of which a separate SHA engine is formed, then all these jobs are submitted and monitored using the parametric job submission feature of gLite. This feature provides ease of operation in submission and monitoring of these SHA jobs.
The SRA application accesses and uses the earthquake catalogue data through the high-level C++ iterators provided by the SDS application service. However, earthquake catalogue data is not the only seismology-related data needed by the calculations. SRA also needs information about fault planes, area sources, site conditions, site-specific attenuation relations etc. SRA aims to maintain them in a repository with a distributed R-tree-like structure where the actual data is kept on SEs while its meta-data, location pointers and indexes are put into AMGA tables. This repository also keeps the assessment figures produced by the previous runs to be browsed by the users.
As a run-time optimization approach, the CEs on which the jobs are executed and SEs where the data is kept are aimed to be selected in such a way that geographically nearest CE/SE elements to the region to which the data belongs are promoted.
A number of services and tools are currently being considered to be utilized for monitoring, QoS and fault-tolerance purposes. This list includes Event Logger, BbmSAMeX and JTS.
The e-mail address sra@ceng.metu.edu.tr is in use through which users may get support from the developers in case of a problem encountered. Also, one may apply the ASG team assigned for support via the e-mail address sra@grid.ici.ro and helpdesk group SRA_support.
