Crawler of the natural hazard events data.

Natural hazard events data provided by external authorities is loaded into the hazard_events layer located in PosgreSQL/PostGIS database.

1. Install PostGIS database.
2. Install GDAL.
3. Install Python 3.
4. Install GDAL python bindings (check your version by using gdalinfo --version)
   python3 -m pip install gdal==2.4.1
5. Install a PostgreSQL adapter for the Python
   python3 -m pip install psycopg2-binary

Initialize database connection settings and create the hazard_events table
python3 scripts/pgsetup.py

Import the hazard events data from external providers
python3 scripts/import_all.py

All data is loaded into hazard_events table located in crawler schema.

ATTRIBUTE LIST
geom        geography(Point,4326)
fid         integer
event_id    text
event_type  text
event_date  timestamp
alert_level text
name        text
description text
origin      text
origin_id   text

During pgsetpu.py process there are created views for specific hazard types:

• earthquake_events
• severe_storm_events
• flood_events
• volcano_events
• drought_events

All views have the same attributes as hazard_events table.

• Global Disaster Alerting and Coordination System (GDACS)
• Earth Observatory Natural Event Tracker (EONET)
• USGG Earthquake Hazards Program
• EPOS Seismic Portal

• Each provider has different set of attributes. In current implementation missing attributes are set to NULL. This behaviour can be changed eg. by limiting list of attributes to common denominator.
• Some attributes can be calculated from others. Eg. alert_level may be calculated using earthquake magnitude value.
• Some providers have single feature per event (eg. GDACS). Other have multiple features per single event (eg. EONET). In current implementation when multiple features occurs, only last feature is used. This behaviour can be changed eg. by moving from Point to MultiPoint.
• Some providers record single category (event type) for event. Other allow multiple categories per single event (eg. earthquake and volcano). In current implementation only first category is used. This behaviour can be changed by allowing multiple categories or by duplicating features with many categories.