STEP Model

Overview

STEP (Short-Term Earthquake Probability) is a Java-based application that generates earthquake forecasts using catalog data. The system can be run either directly on a host machine or via Docker.

System Components

Key Files

• input/step_config.yaml: Main configuration file for forecast parameters
• STEP/run.py: Main execution script
• STEP/params_handler.py: Python script that updates STEP parameters from config
• STEP/catalog_handler.py: Python script for processing earthquake catalogs
• STEP/OpenSHA/: Core application directory containing Java classes and libraries

Configuration Files

1. input/step_config.yaml: Contains primary settings for:

   • Data paths and file locations
   • Catalog date ranges
   • Magnitude thresholds
   • Forecast parameters
   • Grid specifications

2. input/defaults_template.properties: Contains default application settings, and is handled internally by STEP/params_handler.py

Input Data Format

Catalog Format

The system accepts earthquake catalogs in two formats:

1. Input CSEP ascii Format:

   • Columns: longitude, latitude, magnitude, origin_time, depth, cat_id(optional), event_id(optional)
   • Time format: ISO format (e.g., "2023-07-03T10:25:00.000Z")

2. STEP Format (converted automatically by modify_catalog.py):

   • Space-separated values
   • Format: longitude latitude year month day magnitude hour minute second depth 0 0 0 publicid
   • Example: 175.234 -41.123 2023 7 3 4.5 10 25 0.0 12.5 0 0 0 2023p123456

Catalog Processing

The modify_catalog.py script handles:

• Converting from CSV to STEP format
• Filtering events based on:
  • Geographic bounds
  • Magnitude thresholds
  • Depth constraints
• Output filename format: Cat_YYYY-MM-DD-HH-MM_STEP.txt (based on latest event in catalog)

Running STEP

Method 1: Direct Execution

1. Install python<=3.11 and system dependencies:

   sudo apt-get update && sudo apt-get install -y --no-install-recommends gmt openjdk-8-jre tcsh ant

2. Setup virtual environment and install STEP package

   python -m venv venv
   source venv/bin/activate
   pip install -e .

3. Configure the forecast in step_config.yaml. Paths must be either relative to where step_config.yml is located or absolute:

   paths:
     data_dir: ./        
     catalog_file: catalog.csep   # located in ./input
     bg_rates: NZZeroRate05.dat  # located in ./input 
     output_dir: ../forecasts/
     output_file: NGSTEP_90days.txt
   
   catalog:
     start_date: "2016-09-23T13:00:00"
     end_date: "2018-05-07T00:00:00"
     min_magnitude: 3.0
     max_magnitude: 8.0
     delta_magnitude: 0.1
   
   forecast:
     length_days: 90
     min_magnitude: 5.0
     region_code: 90
     parameters:
       a_value: -1.59
       b_value: 1.03
       p_value: 1.07
       c_value: 0.04
   
   grid:
     spacing: 0.05
     precision: 0.01
     anchor: 0.0
     cutoff: 0.5
     bounds:
       lat_min: -43.5
       lat_max: -40.7
       lon_min: 171.7
       lon_max: 175.5

4. Execute the forecast:

   run

   which will use input/step_config.yml. Alternatively, a specific configuration file can be passed as:

   run config.yml

Method 2: Docker Execution

Install the Docker engine for your distribution

1. Build the Docker image:

   docker build \
   --build-arg USER_UID=$(id -u) \
   --build-arg USER_GID=$(id -g) \
   -t step-model .

2. Run the container:

   docker run --rm \
   -u $(id -u):$(id -g) \
   -v "$PWD/input:/app/input" \
   -v "$PWD/forecasts:/app/forecasts" \
   step-model

Output Files

The system generates several output files in the forecasts/ directory:

• NGSTEP_90days.txt: Main forecast output file

File Structure

.
├── Dockerfile
├── runSTEP.sh
├── step_config.yaml
├── Update_parameters.py
├── ModifyCatalog.py
└── STEP/
    └── OpenSHA/
        ├── build/
        │   ├── lib/
        │   └── bin/
        └── config/
   ├── params_handler.py
   ├── catalog_handler.py
   └── run.py

Requirements

Java Dependencies

• commons-httpclient.jar
• commons-logging.jar
• commons-math-1.1.jar
• dom4j.jar
• jargs.jar
• jquakeml-1.0.1-2.0.1-RC1.jar
• log4j-1.2.4.jar
• step-aftershock.jar

System Requirements

• Java Runtime Environment
• Python 3.x (for parameter updates and catalog processing)
• GMT (for visualization)