Release Notes:  Earthworm Version V2.6
10/10/96


NEW OPERATING SYSTEMS:
***********************

Version 2.6 runs on OS2 Warp 3.0 and Solaris.  We have abandoned OS2
v2 because it is no longer available from IBM.


MODIFICATIONS/BUG FIXES TO EXISTING MODULES:
*******************************************

Import_generic:
Does not close socket upon loss of framing on incoming messages. As per request by Ray Buland. alex, 7/23/96

Export_generic:
Performs circular buffering instead of fifo. Buffer is not flushed on re-try attempt. As per Ray Buland. Alex 7/23/96

Hypoinverse:

There are two versions of hypoinverse, one for Solaris and one for
OS2.  They live in ~src/hypo_sol and ~src/hypo_os2 respectively.  The
make_ew_solaris and make_ew_os2.cmd insure that the right version is
compiled.

Binder:
Binder's new pick-assessment feature;  9/96 LDD

A new feature was added to binder to detect and reject outliers from a 
group of picks which supposedly belong to a single earthquake.  Such outliers 
have been notorious for pulling some of binder's event locations far from 
the true locations, thus allowing the event to be corrupted by the association 
of junk phases.  This pick-assessment is based on the resampling technique 
of Bill Ellsworth, which in turn is based on Rex Allen's phase associator.

Pick-assessment is a multiple pass process.  First, any glitches are removed
from the group of picks.  Then the first resampling pass is performed on the
earliest (up to 25) remaining P-phase picks which meet the testing criteria.
In a resampling pass, many possible hypocenters are calculated from different
combinations of four picks at a time using the exact, four-station hypocenter 
solution method of Ingelada.  Traveltime residuals are calculated for each pick 
with respect to every possible hypocenter.  The first resampling pass is used 
to winnow the data to arrivals with median traveltime residuals below a cutoff 
value.  A second pass resamples the winnowed data to determine a trial hypocenter 
to be passed to binder's L1 locator.  If any picks are rejected as glitches or 
outliers, they are freed to restack or reassociate.  
    
       
The binder configuration file commands that control the pick-assessment 
feature are:

  assess_pk        Define when the pick-assessment feature will be invoked.
  define_glitch    Define a glitch as N picks within X seconds. 
  log_accepted     Log info about good picks in addition to rejected picks.
  maxtrial         Maximum number of four-pick hypocenters to calculate.
  maxwt	           Maximum pick-quality to use in pick-assessment.
  residual_cut     Cutoff value for absolute median traveltime residual.
  v_halfspace      Halfspace P-wave velocity used in Ingelada locator.  

A more detailed description of each command can be found in the file
.../earthworm/v2.6/user_doc/binder.commands


CHANGES TO CONFIGURATION FILES:
*******************************

export_generic.d:
New parameter RingSize specifies number of elements in circular buffer.

binder.d
New parameters to control the pick-assessment.

tankplayer.d
New parameter, StartUpDelay,  which defines the number of seconds to
wait on startup before beginning the playback of the first waveform
file.


KNOWN BUGS or DEFICIENCIES:
**************************

Will has moved us from os/2 v2 to os/2 v3. v2 is no longer available,
and v3 offers better stability; v2 seems to have some GUI bugs. Never
mind that NT may come: we play the ball where it lies.  Will had
to create new makefiles, as yes, they're different from OS2 v2. The files in
earthworm/working/bin, named to_os2v2.cmd and to_os2v3.cmd copy the
corresponding versions of the makefiles into the standard name
"makefile.os2" found in each module's source directory.


MODULES IN PROGRESS:
********************

A picker module based on the Rex Allen algorithm.  It's new features are:
	* process the new waveform format for single channel data
	* allow each channel to have it own picking parameters.


Enhancement to import/export to handle binary messages.



