Trigger Model Automation

Automation of Triggr model for Forecast-based Financing in the Philippines

Description :

Organize process of running priority models. The following steps have been implemented for the pilot Phase automation

Step 1:

Check for active typhoon events in the Philippines area. This is done first by checking for disaster events listed on GDACS website https://www.gdacs.org/ , from those active events select typhoon events.

Step 2:

From those typhoon events check if they are located in the Philippines responsibility (PAR) area defined by the following binding box PAR=[[145, 35], [145, 5], [115, 5], [115, 35],[145, 35]] which is approximately the area shown in the figure below.

Step 3:

Extract name and other relevant information for the typhoon event located in PAR

Step 4:

Download Typhoon forecast data from UCL and Rainfall forecast Data from NOAA

Step 5:

Run the trigger model based on input data for the new typhoon

Step 6:

Send an automated email with information on forecasted impact of the new typhoon for users identified by the local FbF team. For the pilot phase this information will be sent via a temporary email account created for the trigger model. The tentative list for automated email recipients has five email addresses.

Step 7:

Repeat the above steps 1 to 6 every 6 hours – until landfall.

Instructions to Run

Prerequisites

• Docker installed
• Docker-settings: set memory of containers to at least 2GB

Retrieve code and move in repository:

git clone https://github.com/rodekruis/Typhoon-Impact-based-forecasting-model.git
cd Typhoon-Impact-based-forecasting-model

Copy secrets-file:

cp secrets.py.template secrets.py

.. and retrieve the correct credentials from someone who knows.

Start up application:

docker build -t fbf-ph .
docker run --name=fbf-ph -v ${PWD}:/home/fbf -it fbf-ph

If entering the container a 2nd time or later:

docker exec -it fbf-ph /bin/bash

or (if unstarted)

docker start -i fbf-ph

To start code manually from inside container

python3 automation_code_automation.py

To inspect the logs (e.g. when getting an email about errors), run from inside the container:

nano /var/log/cron.log

(Scroll down with Ctrl+V)

Imitate typhoon-scenario

Most times, there will be no ongoing typhoon. If you want to simulate a typhoon-scenario for testing/development purposes, you can change the following lines in automation_code_automation.py:

  delete_old_files()
  create_ucl_metadata()
  # Activetyphoon=['KAMMURI']

to

  # delete_old_files()
  # create_ucl_metadata()
  Activetyphoon=['KAMMURI']

and run

  cp Rodekruis-example.xml forecast/Rodekruis.xml