[Shake-dev] USGS ShakeMap V3.5 is available

[Kuo-Wan Lin]

Greetings,

This email is intended for interested parties of the ShakeMap 
application.  On Tuesday, 12/29/2009, a new version (3.5) of the 
software was released by USGS after three months of beta-testing.  We 
recommend that current ShakeMap users upgrade their installations after 
evaluating the software.  To access the code, you will need the 
subversion client and issue the following command from a shell,

svn checkout 
https://vault.gps.caltech.edu/repos/products/shakemap/tags/release-3.5/ 
local_directory_name

Unlike the beta version, user authentication is not required for 
accessing the code, except for granting an exception to accepting the 
secure layer certificate during connection.  Current beta users should 
checkout a new copy of ShakeMap from the repository and merge any user 
modifications from the beta code.

The new version continues to be a branch release based on the original 
software but changes the way ShakeMaps are computed. It includes several 
major upgrades and enhancements regarding data format and processing.  
As results, for example, user-defined regression modules and 
configuration files for programs "grind" and "mapping" must be upgraded 
in order to work with the version 3.5 code.  A LOT has changed so please 
make sure you read the Release Notes and the Software Guide before 
trying to install the software (both of these documents may be found in 
the "doc" directory).

Special thanks to Peter Lombard (creator of plotregr) and Lijam Hagos 
for their contributions on improving the software during beta-testing.

Listed below are highlights of the new functionality of Version 3.5.

1. MMI (or Intensity, in general) is now a valid input data type. 
Intensity data uses the same format of input XML file as other ground 
motion data, but we've added two new attributes to the station tag: the 
intensity attribute should be set to the decimal intensity for the 
"station;" the netid attribute should be set to "MMI," "CIIM," "DYFI," 
or "INTENSITY" (all four are currently equivalent).  If netid is set to 
one of these values, any amplitude data (i.e., data enclosed in a comp 
tag) will be ignored and grind will use the mmi2pgm function to derive 
the ground motions.  Likewise, if netid is not one of these values, the 
intensity attribute will be ignored and grind will compute intensity 
using the pgm2mmi function.

Here is an example of a station tag using Did You Feel It? data:

<station code="91042" name="ZIP Code 91042 (Intensity VII, 38 
responses)" insttype="USGS (Did You Feel It?)" lat="34.282604" 
lon="-118.237943" source="USGS (Did You Feel It?)" netid="DYFI" 
commtype="USGS (Did You Feel It?)" intensity="7.4">

2. ShakeMaps now include a new class of data: converted observations.  A 
converted observation would be, for example, MMI converted to PGA, PGV, 
or PSA, or one of those ground motion observations converted to MMI.  On 
a map of PGA, for example, PGA data are considered native observations, 
and MMI converted to PGA are considered converted observations.

3. The new version of grind produces a weighted combination of native 
observations, converted observations, and estimated data at every point 
in the output grid.  Formerly, grind combined observations (data) with 
GMPE-derived estimates computed on a coarse grid, then interpolated 
(i.e., filled gaps) with the GMT program surface to a finely spaced 
grid.  Estimates near observations were rejected in favor of the 
observation. Weighting is now determined by the formal uncertainty of 
each datum.  See (Composite ShakeMap, 2010) for more information on the 
way weighting is computed.

4. In the new grind macroseismic intensity is treated as any other 
ground motion parameter. MMI (or Intensity) was formerly computed only 
from PGA and/or PGV. For intensity maps, intensity observations are now 
native, PGM is converted to intensity for converted observations, and 
estimates are made directly from an Intensity Prediction Equation (IPE).

5. The conversion from PGM to Intensity (and Intensity to PGM) is now 
performed in configurable modules.  The operator may thus select (or 
develop) the conversion function most appropriate for his or her 
region.  The modular approach also allows the use of Intensity measures 
other than MMI.

6. Bias is now a magnitude adjustment, rather than an amplitude 
multiplier.  (I.e., the magnitude given to the GMPE is adjusted to give 
the best fit to the observed and converted data (with converted 
observations being down weighted relative to native observations.)  This 
approach should be more consistent with removing the inter-event 
variance from the GMPE.

7. grind now offers the option to compute site amplification using the 
GMPE's native site amplification term(s).  Borcherdt-style site 
corrections are still the default, however if the operator calls grind 
with the new -gmpesc flag, grind will opt instead for the site 
correction formula defined by the GMPE (if one exists.)  If no site term 
is defined by the GMPE, or the site term is unusable with Vs30, the 
module may complain, or it may quietly use the Borcherdt corrections (if 
the operator has configured Borcherdt-style corrections in grind.conf.)

8. Along with the improved handling of observational intensities, we 
have included a program called 'getdyfi' to extract intensities from the 
online DYFI (Did You Feel It?) web page. This is suitable for inclusion 
in the 'retrieve' configuration file, or you can run it for a individual 
event from the command line.

We highly encourage all ShakeMap users to utilize the shake-dev mailing 
list below to discuss various aspects of the functionality of the new 
ShakeMap application.  At the minimum level of implementation, the 
ShakeMap version 3.5 will behave the same as the current version 3.2 
code and still benefits from improved accuracy of ground shaking 
estimates at the grid level and improved performance.


Best Regards,

Bruce, Dave, Kuo-Wan, and Vince


_______________________________________________
Shake-dev mailing list
shake-dev at geohazards.usgs.gov
https://geohazards.usgs.gov/mailman/listinfo/shake-dev