ANDS Logo

Project promotion materials:

Data collections can be seen on:

http://researchdata.ands.org.au/extreme-weather-events-database

Software is available at:

http://swish-climate-impact-assessment.github.io/

Project Members:

Dr. Keith Dear (Project Manager, keith.dear@anu.edu.au)

Ivan Hanigan (Data Manager, ivan.hanigan@anu.edu.au)

Dr. Charmian Bennett (Epidemiological System Analyst, charmian.bennett@anu.edu.au)

Mr Ian Szarka (Software Developer, )

ANDS Contact:

Mingfang Wu (mingfang.wu@ands.org.au)

Project Status:

Completed

A Scientific Workflow System for Assessing and Projecting the Health Impacts of Extreme Weather Events

Australian National University

Project Description:

The Environment, Climate and Health (ECH) group at the National Centre for Epidemiology and Population Health poses research questions that seek to explain the influence of environmental phenomena on human health and wellbeing. The group does much of its research using existing data sourced from governmental organisations. However, the use of these datasets is limited by the individual abilities of a researcher to access and use the existing computational and data infrastructure, which relies on high-level programming knowledge and skills. A researcher who wishes to merge several datasets needs to be conversant in database management systems (DBMS), script-writing and programming, network-based communication with remote services, data visualisation and statistical computing (at a minimum). Additionally, the steps taken must be recorded in sufficient detail for the results to be examined and reproduced by other researchers.

This project aimed to create a scientific workflow system to facilitate access to, and manipulation of, large datasets held on remote databases. We called our system SWISH: Scientific Workflow and Integration Software for Health.

The SWISH system provides a user-friendly, drag-and-drop interface to locate and extract data from the Extreme Weather Events Database (EWEDB). Once familiar with the SWISH environment, users can link together multiple steps around data extraction, manipulation and analysis using ‘actors’ to form an executable workflow that can be modified and updated repeatedly as the analysis progresses. These workflows document the exact steps taken in any analysis, thus combining documentation and execution into a single step. Workflows can also be easily shared with other researchers to improve the reproducibility of results and facilitate collaborative research projects. SWISH workflows can also be used to standardise processes within our research team, such as data cleaning steps and updating existing datasets with more recent data.

SWISH workflows can incorporate preferred programs, procedures and analyses by providing actors that run statistical packages such as R and Stata within a workflow. The use of scientific workflows like SWISH can also minimise accidental errors and omissions, especially with more complex analyses, as the step-by-step process of creating the workflow forces the user to break up complicated analyses into small, logical, well-documented steps.

In summary, the SWISH system:
- Uses a familiar, drag-and-drop user interface rather than complex programming scripts
- Integrates documentation and analysis using executable scientific workflows.
- Makes it easy to share workflows with other researchers, especially in collaborative projects.
- Creates scientific workflows that are easy to modify, update or extend as analysis progresses.
- Can incorporate existing, trusted procedures, analyses and statistical programs.
- Can automate repetitive or frequently used data acquisition and preparation steps.


Research Champion:

A/Prof. Keith Dear, Dr. Charmian Bennett and others


Data Type:

Input Data:
Meteorology; drought indices; heat stress indices; mortality; demographics; administrative boundaries; infrastructure; SEIFA; remoteness; climate change scenario projections.

Output Data:
Extreme weather events

ANZSRC-FOR code:

111706 Epidemiology
160508 Health Policy