<-- Back to schedule

Python for Water Forecasting Services

Project: Water Resources Information of the Bureau of Meteorology

Since 2008, the Bureau of Meteorology has developed several modelling systems to support its streamflow forecasting services. These systems include MSDM (Modified Statistical Downscaling Method) and WAFARi (Water Availability Forecasts of Australian Rivers) for the Seasonal Streamflow Forecasting service (http://www.bom.gov.au/water/ssf), STAR (Streamflow Toolkit for Australian Rivers) for the 7-Day Streamflow Forecasting service (http://www.bom.gov.au/water/7daystreamflow) and HRS toolkit for the Historical Reference Stations (http://www.bom.gov.au/water/hrs). These systems routinely ingest recent observation data, fetch climate forecasts, run rainfall-runoff models and provide updated forecasts through publicly available websites. We chose Python as the primary programming language to build the main components of these systems, and used open source packages for scientific computing including NumPy, Pandas, Matplotlib, PyTables and IPython. Python was used as the glue to integrate different system components, and as a bridge to connect scientists and IT programmers. This approach resulted in a highly productive collaboration with CSIRO and university partners. It also fostered effective communication between hydrologists and system developers. In this presentation, we will describe how these modelling systems were built up and currently operate within the Bureau, and also explain how the use of Python was a key factor for successful development and operation of these forecasting systems.

Daehyok Shin

Daehyok Shin is a hydrological modeller and also a professional Python programmer for scientific and engineering computing. He studied electronics and computer programming at the Engineering Department and remote sensing and geographic information system at the Environment Department of the Seoul National University, South Korea. He earned a Ph.D degree with a study about watershed hydrological modelling at the Geography Department of the University of North Carolina, Chapel Hill, USA. He joined the Bureau of Meteorology in 2008 and since the year, he has led several project teams to develop operational modelling systems for new streamflow forecasting services. The new service includes the Seasonal Streamflow Forecasting Service, the Hydrological Reference Stations Service and the 7-day Streamflow Forecasting Service. As a scientific Python programmer since 2004, he conducted a key role in the Bureau's adoption of Python for the development of its operational streamflow forecasting systems. He gave many presentations about the application of Python and organised a couple of Python training programs particularly for environmental modellers. He is currently working at the Bureau's Head Office in Melbourne.