Resolve "[CRS Update Step 5] Spatially Varying Forecast"
Closes #137 (closed)
Added spatial forecast to the Coulomb Rate-State forecast model. The parameter tab, forecast rate map, probability of exceedance map, and temporal forecast at 1000days for Decatur example are shown in the snapshots below (tkinter version on MacOS):
Temporal forecast using the STRIVE front end is shown below:
The spatial forecast uses the following logic to compute static coulomb stress:
- Perform temporal clustering using rate density: (Each color represents a temporal cluster)
- Perform spatial clustering for each temporal cluster: The logic flow chart
- Calculate Coulomb stress change for each spatial-temporal cluster for the host cell and affected cells An example from temporal clusters to spatial-temporal clusters and Coulomb stress change calculation
In this version of the forecast model. Coulomb stress changes are calculated for all grid cells, this could potentially be inefficient if there are millions grid cells but works fine with 224 grid cells with the Decatur example. A more efficient method exist (only calculate Coulomb stress change for affected cells), and can be implemented as an alternative for large problems in the future.