The static datasets of USGS NSHMs prior to 2014 were computed using Fortran (see
The static datasets of USGS NSHMs prior to 2014 were computed using Fortran (see
[nshmp-haz-fortran](https://github.com/usgs/nshmp-haz-fortran]). The static datasets for the
[nshmp-haz-fortran](https://github.com/usgs/nshmp-haz-fortran]). The static datasets for the
2014 Conterminous U.S. NSHM were computed using the Fortran codes and
2014 Conterminous U.S. NSHM were computed using the Fortran codes and
[OpenSHA](https://opensha.org/) for the California portion of the model. The dynamic versions
[OpenSHA](https://opensha.org/)(for the California portion of the model). The dynamic versions
of the 2008 and 2014 Conterminous U.S. models were then implemented in the 1st version of
of the 2008 and 2014 Conterminous U.S. models were then implemented in the 1st version of
[nshmp-haz](https://github.com/usgs/nshmp-haz)(on GitHub). This updated Java codebase uses XML
[nshmp-haz](https://github.com/usgs/nshmp-haz)(on GitHub). This updated Java codebase uses XML
source models and supports the web services behind the dynamic calculations of the [Unified Hazard
source models and supports the web services behind the dynamic calculations of the [Unified Hazard
...
@@ -19,7 +19,7 @@ underlying logic trees and support uncertainty analysis.
...
@@ -19,7 +19,7 @@ underlying logic trees and support uncertainty analysis.
NSHMs are very detailed and migrating from one format to another is not trivial and prone to error.
NSHMs are very detailed and migrating from one format to another is not trivial and prone to error.
Moreover, approximations (e.g. using 3.1415 for Pi rather than the the value built into most
Moreover, approximations (e.g. using 3.1415 for Pi rather than the the value built into most
languages) can yield different results. When multiple such small changes exist, deciphering what
languages) can yield different results. When multiple such small changes exist, deciphering what
si giving rise to differences in results can be challenging.
is giving rise to differences in results can be challenging.
To document the transition from nshmp-haz v1 to v2, we here attach comparison maps at four return
To document the transition from nshmp-haz v1 to v2, we here attach comparison maps at four return
periods (475, 975, 2475, and 10,000 year) for the 2018 Conterminous U.S. model. Maps are included
periods (475, 975, 2475, and 10,000 year) for the 2018 Conterminous U.S. model. Maps are included
...
@@ -30,7 +30,8 @@ additional epistemic uncertainty added to the the NGA-West2 ground motion models
...
@@ -30,7 +30,8 @@ additional epistemic uncertainty added to the the NGA-West2 ground motion models
We continue to investigate the cause of other differences but they are small enough that we are
We continue to investigate the cause of other differences but they are small enough that we are
comfortable moving forward deploying nshmp-haz v2 codes and models to our public web services and
comfortable moving forward deploying nshmp-haz v2 codes and models to our public web services and
applications.
applications. This repository includes end-to-end tests for supported NSHMs that may be run
on demand.
[Download v1 to v2 difference and ratio maps](https://code.usgs.gov/ghsc/users/pmpowers/nshmp-haz/-/raw/code-version-doc-506/docs/pages/images/comp_JSON_vs_XML_0p2-grid-20220216-BC.pdf?inline=false)
[Download v1 to v2 difference and ratio maps](https://code.usgs.gov/ghsc/users/pmpowers/nshmp-haz/-/raw/code-version-doc-506/docs/pages/images/comp_JSON_vs_XML_0p2-grid-20220216-BC.pdf?inline=false)
