Seismic databases and processing tools currently available are mainly limited to classic three-component seismic recordings and cannot handle collocated multi-component, multi-disciplinary datasets easily. Further, these seismological databases depend on event-related data and are not able to manage state of the art continuous waveform data input as well. None of them allows for automated request of data available at seismic data centers or to share specific data to users outside one institute. Some seismic databases even depend on licensed database engines, which contradicts the open source character of most software packages used in seismology.

This study intends to provide a suitable answer to the deficiencies of existing seismic databases. SeisHub is a novel web-based database approach created for archiving, processing, and sharing geophysical data and meta data (data describing data), particularly adapted for seismic data. The implemented database prototype offers the full functionality of a native XML database combined with the versatility of a RESTful Web service. The XML database itself uses a standard relational database as back-end, which is currently tested with PostgreSQL and SQLite. This sophisticated structure allows for the usage of both worlds: on the one hand the power of the SQL for querying and manipulating data, and one the other hand the freedom to use any standard connected to XML, e.g. document conversion via XSLT (Extensible Stylesheet Language Transformations) or resource validation via XSD (XML Schema). The actual resources and any additional services are available via fixed Uniform Resource Identifiers (URIs), where as the database back-end stores the original XML documents and all related indexed values. Indexes are generated using the XPath language and may be added at any time during runtime. This flexibility of the XML/SQL mixture introduced above enables the user to include parameters or results as well as meta data from additional or yet unknown monitoring techniques at any time. SeisHub also comprises features of a “classical seismic database” providing direct access to continuous seismic waveform data and associated meta data. Additionally, SeisHub offers various access protocols (HTTP/HTTPS, SFTP, SSH), an extensible plug-in system, user management, and a sophisticated web-based administration front-end. The SeisHub database is an open source project and can be freely downloaded via the project home page http://www.seishub.org.

The SeisHub database has already been deployed as central database component within two scientific projects: Exupéry, a mobile Volcano Fast Response System (VFRS), and BayernNetz, the seismological network of the Bavarian Seismological Service Erdbebendienst Bayern. Acknowledgements

This project was funded by the German Science Foundation (DFG) via grant DFG IG 16/9-1.

• Barsch, Robert (2009): Web-based technology for storage and processing of multi-component data in seismology: First steps towards a new design. Dissertation, LMU München: Fakultät für Geowissenschaften.
• A tutorial for writing SeisHub Plugins: seishub.plugins.how_to_extend_seishub

• Bernsdorf, S., Barsch, R., Beyreuther, M., Zakšek, K., Hort, M., Wassermann, J. (2010), Decision support system for the mobile volcano fast response system. International Journal of Digital Earth. 3 (3), 280-291.
• Beyreuther, M., Barsch, R., Krischer, L., Megies, T., Behr, Y., and Wassermann, J. (2010), ObsPy: A Python Toolbox for Seismology, Seismological Research Letters, 81 (3), 530-533.
• Megies, T., Beyreuther, M., Barsch, R., Krischer, L., and Wassermann, J. (2011), ObsPy - What can it do for data centers and observatories?, Annals of Geophysics, 54 (1).

The following section shows how to install SeisHub and associated components. It will not cover the installation of a relational database back-end, like PostgreSQL. Please refer to the manual of the preferred database.

For Linux and UNIX systems the author suggests to install !SeisHub as a non-administrative user applying a new, separate, local Python >= 2.6.x instance.

Installing Python on a Windows operating system is more complicated because development tools like a C compiler are not part of a standard Windows distribution. Therefore many Python modules using C extensions have to be delivered as binary package with an executable installer. The fastest, most unproblematic way is to install Python and all extensions as the administrative system user.

1. Download and uncompress the latest stable Python 2.6.x package for the used operating system from http://www.python.org/download/. Windows user may just use the executable installer and skip to the next subsection.

2. Run

    ./configure --prefix=$HOME
    make
    make install
   

3. Add $HOME/bin to the PATH environmental variable, e.g. in bash:

    export PATH="$HOME/bin:$PATH"
   

4. Call python in command line. It should show the correct version number.

Easy Install is a powerful command-line based package management tool for Python. Like CPAN for Perl, it automates the download, build, installation and update process of Python packages.

1. Download http://python-distribute.org/distribute_setup.py.

2. Run

    python distribute_setup.py
   

easy_install SQLAlchemy
easy_install Cheetah
easy_install pycrypto
easy_install Twisted
easy_install pyparsing
easy_install pyasn1
easy_install lxml               # Linux requires libxml2-dev and libxslt-dev
easy_install pyOpenSSL          # Linux requires libssl-dev
easy_install numpy              # see link in Notes
easy_install obspy

The seismology plug-in requires the following additional modules:

easy_install matplotlib

• More details (especially for matplotlib and numpy installation on linux) can be found at https://github.com/obspy/obspy/wiki/Installation-on-Linux:-Dependencies
• Windows users need to install pywin32 (Python for Windows extension). Download and install from http://sourceforge.net/projects/pywin32/.
• Package lxml requires the libxml2-dev and libxslt-dev packages. Compiling takes a while. This does not apply to an installation on Windows - here are binaries delivered.
• Binary packages for pyOpenSSL can be found at http://www.egenix.com/products/python/pyOpenSSL/ or http://sourceforge.net/projects/pyopenssl/

SeisHub uses as default data back-end SQLite, which comes with Python 2.6.x.

For PostgreSQL additional database bindings are required. Those bindings can be installed via:

easy_install psycopg2           # Linux requires libpq-dev

• On Debian/Ubuntu install python-psycopg2 via package management
• Windows binary packages for psycopg2 can be found at http://www.stickpeople.com/projects/python/win-psycopg/

1. Get the latest SeisHub version via PyPI

    easy_install seishub.core==dev
   

2. Optionally get any plug-in you need, e.g.:

    easy_install seishub.plugins.seismology==dev
   

3. Go into your Python script directory and initialize a new local instance:

    seishub-admin initenv /path/to/instance
   

4. Change to the bin directory within the instance path and start the server

    cd /path/to/instance/bin
    ./start.sh
   

5. Open http://localhost:8080/manage in your browser. Default user name and password are both set to "admin".

You probably want to stop the server after the first run and adjust the settings within the configuration file seishub.ini within the conf directory or use the web interface for that.

Using PostgreSQL as default database backend requires a few more additional steps:

1. Login as postgres super user

    su - postgres
   

2. Create a new database user "username" using createuser. It will prompt for a password.

    createuser -D -P -S -R username
   

3. Create a new database "databasename" for the user "username" using createdb.

    createdb -O username databasename
   

4. Logout

After creating the user and database you may use the connection string postgresql://username:password@host:port/databasename (postgres default port is 5432).