EARTHWORM Status Report
Feb. 12, 1998
This is an update of Earthworm status since the notes of
9/25/97.
The Earthworm community is alive and growing. The number of
participants is growing and, more significantly, so is the number of
sites participating in development. Currently, eight individuals at
UofW Seattle, PNL Richland, USGS Menlo Park, ATWC Palmer, and USGS
Golden are actively participating in design, implementation,
modifications, and testing. This group has been keeping in touch via
email, individual phone calls, and a weekly conference call meeting.
This has resulted in a very effective development team, especially when
one considers that we're spread over three time zones, several thousand
miles, and report to six different supervisors. The benefits have been
to (1) accelerate progress by sharing specialized knowledge, (2)
provide a very rapid mechanism for isolating and fixing bugs, (3)
identify and implement mutually desired enhancements, and (4) assure
that new designs are as generally useful as possible.
CURRENT DIRECTIONS:
- 1. NOAA-USGS Tsunami project:
This has been one of the main objectives. Work to date
consisted of installation of systems at the Alaska Tsunami
Warning Center (ATWC) in Palmer Alaska, UofW in Seattle, USGS
Menlo Park, and an installation is in progress at PGC British Columbia.
The result is that ATWC is receiving trace data from IDA, USNSN, Menlo
Park, and Seattle. All data are interfaced to ATWC's existing
processing system. In addition, hypocentral parameters are
being received from Menlo Park and Seattle. Our current
installation there consists of two computers: a Sparc Ultra,
which acts as their 'loading dock', and an NT Intel machine
which digitizes local analog data and runs the core ATWC
functions. The digital data are currently being shipped via
a combination of public Internet and DOI net, which, as expected,
has proven to be a rich source of rare link failure-modes,
and has thus provided a good test-bed for debugging and ruggedizing
our communications software. Work is continuing, as described below.
- 2. Support of regional networks:
This has focused on support of individual installations, and
'completion' of Earthworm, in terms of interactive analysis and
archiving. The work done at the University of Fairbanks (IceWorm)
has resulted in an excellent integration of the DataScope interactive
analysis package. The approach here has been to create 'connector'
software, which facilitates interfacing Earthworm to various
interactive analysis schemes already in use at various sites. An
interface to the UofW 'uw2' format
has been completed, and a SAC interface is in progress. The
object is to be in position to accommodate pre-existing
interactive analysis packages, as well as being able to offer a
variety of pre-integrated options. A number of site visits have
been scheduled with the objective of discussing areas in which
we can be of service, and to produce task lists and schedules.
- 3. EWAB:
We have responded to the
EWAB request for documentation (see below); in addition,
a 'request for comment' (RFC) mechanism is being created to
coordinate development efforts at various sites. The objective
is to assure that the whole community be aware of various planned
and ongoing developments, and that there be an opportunity to
submit comments which will be considered by the developing
institution. The hope is that this will permit sharing of
programming talent and software components, and will
result in modules, which will be of more general utility.
RECENT DEVELOPMENTS:
- Socket utility routines: Our experiences with long-distance, long-term
socket connections, multi-threaded servers, automated client programs
as well as the seamy underside of NT and Solaris 2.6, have led us
re-write our socket utility routines. This new suite features
user-specified timeout values and proper operation on Solaris and NT.
- WaveServerV: This is an enhancement of WaveServerIV. The additional
features are a robust crash-recovery scheme, incorporation of the new
socket routines (see above), performance enhancements, and bug fixes.
- WaveClientII: This is a set of utility routines to ease creation of
WaveServer clients. This set replaces the earlier, simpler such set,
and features a wider variety of services for both 'simple' and
'sophisticated' client programs, is multi-thread safe, and utilizes the
new socket routines (see above).
- 'UW2' interface: An interface between Earthworm and the University of
Washington 'uw2' interactive analysis package has been completed. The
routine is executed either automatically or manually, takes either
Earthworm 'trigger' messages or UW2 format pick files, and constructs
the UW2 event files by accessing a specified list of WaveServers. This
routine is the first use of a series of utility routines, which will
permit quick integration with other interactive analysis formats.
- SAC interface: Work is about to start on an interface to SAC, modeled
on the UW2 interface code, above. The next task is to submit to the
EWAB a request for comment (RFC). The plan is to collect the responses,
produce an implementation plan, which addresses the comments, and to
start implementation of a first version module.
- CarlTrig: Named in honor of its perpetrator, this is a port of the
traditional STA/LTA trigger algorithm as popularized by Carl Johnson.
The initial version is nearing completion. The motivation was to
provide a more conservative method of saving event data than via
hypo-inverse solutions. This module generates 'trigger' messages, the
same as those generated by hypoinverse solutions (arc2trig), or the
John Evans long period coincidence detector (lptrig). This initial
version features subnet logic, and may include individual station
trigger parameters and inclusion of delayed data. An RFC is being
prepared to provide the specs for future versions.
- Documentation: An overview document, module descriptions, and sample
parameter files have been submitted to the EWAB, and have been posted
on the CNSS site. This is an ongoing process; additional descriptions
are being prepared. It has been noted that there is currently a
discontinuity between the overview document and the descriptions of
individual modules. It is agreed that there should be a section
explaining more about the various components of Earthworm, and an
explanation of how they interact. It has been suggested that this could
be done via several examples, stating the problem to be solved, a
diagram of an Earthworm construction to solve the problem, and an
explanation of how the various components of the construction would
interact. Further comments and contributions are welcome.
- RCV: RCV, the USNSN computer code which receives USNSN data has been
encapsulated into an earthworm module. The approach here has been to
encapsulate the existing RCV code with no modifications. Thus, as the
USNSN produces internal enhancements to RCV, the new code can be moved directly
into this module. Modifications affecting the external behaviour, will, of
course, require a change to the encapsulating wrapper. This connects to the USNSN
system in Golden, CO and receives triggered, compressed trace data,
possibly out of sequence. The USNSN system in Golden controls which
stations are sent. The encapsulating wrapper decompresses the data,
re-sequences the data (as permitted within a user-specified time
limit), and produces a stream of standard Earthworm trace data
messages. This is currently installed and undergoing testing between the
USNSN Golden and ATWC in Palmer. It appears that RCV may require some
minor enhancements to make it more robust, and USNSN has agreed to
work on this. It is expected that this module will be documented
and released as part of Earthworm after it has demonstrated acceptable
performance.
- import_IDA: A similar encapsulation has been done for the receiver of
IRIS/IDA data from UCSD. A module description document is being produced.
FUTURE DIRECTIONS:
- Tsunami: This will continue to be a large effort. Upcoming tasks
include implementing a list of enhancements for ATWC, generating a
second-generation bug- and enhancement list based on experience with
the current installation at ATWC, installation of dedicated
communication links, integration of data acquisition from several types
of data loggers, and installation and testing of a redundant system.
- DBMS: Phase II work is about to begin. This includes evaluation and
selection of suitable data structures, the production of a working
system capable of acquiring and storing real-time data in a commercial
data base, providing operator displays and controls, and to provide
interactive access to the data which permits modifications, additions,
and archiving of the stored data.
- Complete Earthworm: Work toward making Earthworm a complete network
processing system capable of rapid exchange of data with other sites,
and offering a suite of interactive analysis tools suitable for
low-level routine processing as well as research support.
- Regional support: Assist various regional networks as required to help
them meet their goals.
Return to EARTHWORM main page, or
CNSS main page.