Multi-Phenomenology Explosion Monitoring (MultiPEM) Toolbox
The MultiPEM approach involves the statistical modeling of observed signatures from multiple phenomenologies at distributed sensor networks to infer characteristics of interest from new explosive events. Recent publications have considered processed waveforms from seismic (e.g. P-wave displacement, peak particle velocity), acoustic (e.g. pressure impulse, duration), and optical (e.g. irradiance first minimum, second maximum) phenomenologies for this purpose. Expected signatures are modeled empirically or from first principles utilizing two basic components: source physics which generates initial waveforms from explosive events, and path physics which attenuates these waveforms along their propagation route from source to sensor. Systematic departures of observed signatures from these two model components can be modelled dynamically.
Historically, device parameters were characterized by single phenomenology analyses. When signatures from multiple phenomenologies are available, MultiPEM assessments will naturally compute a compromise among single phenomenology results while often producing rigorously quantified uncertainty reduction as the desired benefit.
Consult the technical report provided with this repository for a description of the statistical methodology and example application implemented in this repository:
./multipem-la-ur-23-21950.pdf
Consult the user manual provided with this repository for descriptions of the input decks provided with the example application and instructions on how to run single-phenomenology and MultiPEM analyses:
./multipem-um-la-ur-23-30117.pdf
Directories in this repository:
Applications/
Contains application-specific code, data, and verification tests
Code/
Contains functions utilized across applications, such as the log-likelihood, log-prior, and log-posterior distributions and their associated gradients
Runfiles/
Set up application directories to run MultiPEM analyses using R
Runfiles-Docker/
Set up application directories to run MultiPEM analyses using R from Docker
Test/
Contains code for running general verification tests using R
Test-Docker/
Contains code for running general verification tests using R from Docker
These directories, and associated sub-directories, contain additional README files with information and instructions for running the code.
This toolbox is open source under the BSD-3 License. Its LANL-internal identifier is O4673.