Migrating from HMTk to MBTk for smoothed seismicity

[Carlos Herrera]

Dear OpenQuake community,

Regarding smoothed seismicity, I have some experience working with HMTk functions thanks to this still useful tutorial. However, other threads in this forum mention that the hazard team now uses the MBTk as the main tool for this. So, I just started familiarizing myself with the MBTk to migrate to it. I installed it following these instructions without any apparent error. Right now, it's OpenQuake 3.26.0 (devel), GMT 6.6.0, and PSHAModelBuilder in Julia, all installed on Linux.

​My goal is to "translate" my HMTk scripts to MBTk, eventually gaining capability for both fixed and adaptive kernels. The current MBTk documentation is a good start, but I still have some unanswered questions about its workflow:

1. Should I still run the completeness and Gutenberg-Richter steps in the documentation, even if I already know the completeness table of the catalogue and an estimated regional b-value? Coming from HMTk, I started working on that assumption and for now I skipped those two steps of the SSC workflow.

2. What would be the equivalent parameters in the MBTk for the following parameters?
    - "grid" and "bvalue" in the HMTk  SmoothedSeismicity() function.
    - "Length_Limit", "BandWidth", and "increment" in the HMTk run_analysis() function.
​3. In this documentation section, how should the "zoning" be formulated? For now, I created a single zone (polygon) in QGIS covering the study area (equivalent to the grid bounds in the HMTk function).
4. When running oqm wkf create_subcatalogues_per_zone or oqm wkf set_h3_to_zones, I get the following warning on the terminal (not sure about the reason):
Warning 3: Cannot find header.dxf (GDAL_DATA is not defined)

5. If I save the polygon that I delineated in QGIS in GeoJSON format, I get the following error when running oqm wkf set_h3_to_zones:

TypeError: polygon_to_cells() got an unexpected keyword argument 'geo_json_conformant'

But if I save the same polygon as a shapefile, I don't get that error and keeps running.

I'm currently stuck in the box counting for smoothing (this documentation section). This step runs without errors, but the script creates an empty "mapping_h5.csv" file. After this point, both fixed kernel and adaptive smoothing methods fail with errors.

I prepared an example catalogue (already complete and declustered) and a sample script with HMTk functions (attached), which runs well and returns a grid of smoothed a-values. I've also attached the script with MBTk functions and the toml parameter file that I used for testing, which reads the same catalogue and the polygon shapefile, but it's failing to carry out the smoothing.

I hope these example scripts and data help to easily track errors of my workflow. I'd appreciate the help to make the MBTk script work properly (to output smoothed a-values and the XML file) and to better understand its parameters.

Best regards,

Carlos