On Monday 8 April 2013 the Working Group on Hydrological Prediction (WG6) had its first meeting by videoconference. The accessibility, quality and sharing of data, in particular atmospheric forecast data, but also flow gauge observations and catchment boundaries, were earmarked as an important issue in the WG6 area. Engagement with the Bureau of Meteorology, as the authorized provider of hydrological forecasts in Australia, is considered key in this respect, either by learning the Bureau’s research needs or by suggesting research areas of interest to the Bureau. It was noted priorities in a recently communicated list of identified research needs by the Bureau (e.g. unified model/system across decadal, seasonal and short-term forecasts, use of probabilistic atmospheric forcing, inundation modelling, coastal flooding) are currently insufficiently articulated. The Bureau’s operational hydrological forecasting service is considered an advantage for this WG as a point of reference, although it also may complicate data sharing issues. To make a case for a hind-cast experiment with Bureau involvement it was decided to aim for a small, common activity of the WG6 participants with a clearly articulated purpose. To this end, all WG6 participants are invited to list their interests in a shared drop box document to be initiated shortly. The next WG6 meeting will be announced pending the finalization of this shared document of purpose, but preferably within a time frame of 1-2 months. Full minutes follow below.

===

MINUTES

Ozewex WG6 Hydrological Prediction start-up meeting 

Monday 8 April 2013 2-4pm by video conference

Partcipants:

• Adelaide: Okke Batelaan (Flinders University)
• Canberra: Albert van Dijk (ANU), Ben Gouweleeuw (CSIRO, meeting/WG6 co-chair)
• Melbourne: Dongryeol Ryu (UoM), David Robertson (CSIRO), Tom Pagano (BoM, WG6 co-chair)
• Perth: Mohammed Bari (BoM)

Apologies: Jason Evans (UNSW), Bellie Sivakumar (UNSW), Beth Ebert (BoM), Rachel Gilmore (BoM), Justin Robinson (BoM), Narendra Tuteja (BoM), Prasantha Hapuarachchi (BoM), Fiona Johnson (UNSW, WG6 co-chair)

 

AGENDA

1. Recap research areas of participants in Working Group 6

2. What do people want to get out of OzEWEX?

• Ben Gouweleeuw at CSIRO is interested in modeling and remote sensing, particularly satellite estimates of flood inundation/open water extent to constrain hydrodynamic forecasts in a data assimilation framework. 
• Albert van Dijk at ANU is interested in forecasting, large scale monitoring and forecasting, weather forecasting, intermediate to seasonal forecasting, data assimilation; process based forecasting; He is involved with GEWEX, which has a forecasting working group that is going to be combined into HEPEX. 
• Okke Batelaan at Flinders is interested in distributed physically based modeling; remote sensing input; intercomparison of distributed models; WetSpa model developed in Brussels (yes this is freely available)
• Mohammad Bari at Bureau; hydrological prediction, developing operational services; landscape scale water balance; distributed modeling for large scale water balance and climate change impact evaluation. Lucicat model.
• Tom Pagano at Bureau; has an operational flood forecasting background; short-term forecasting research in Australia while at CSIRO, recently joined BoM to help transition continuous modeling to flood forecasting operations; interested in all aspects of forecasting from data to modeling to communication. 
• David Robertson at CSIRO; head of CSIRO short-term forecasting research, has experience in seasonal forecasting using statistical methods; ensemble forecasting, NWP.
• Dongryeol Ryu at University of Melbourne; quantifying uncertainty of predictions and its ensemble realisation; remote sensing; forecasting, modeling and observing soil water.

In total, the group’s interests covered a wide range of temporal and spatial scales, from short-term to seasonal. There was not much representation of the climate change sector but climate change was considered a special topic in forecasting. 

3. Identify needs/constraints on hydrological prediction research in Australia.

4. What are the key challenges/opportunities for hydrologic prediction for the future? Does our current research plan cover them?

Researchers in forecasting face many of the same challenges that modelers in general face, such as issues of data sharing, data quality control, data provenance, data availability (particularly sub-daily data), metadata management, a lack of sharing common tools for modeling and analysis, a lack of intercomparisons of techniques and so on. A particular problem for forecasting researchers is a lack of retrospective datasets of rainfall forecasts. 

van Dijk asked Bari about Bureau of Meteorology operational forecasting needs. He responded that drought prediction is an emerging issue. Ultimately there will be a need for spatially distributed drought/soil moisture modeling at a continental scale. Bari placed heavy emphasis on the need for distributed modeling with river routing but conceded that this is not the state of the operational practice at the moment. There was also a need for a unified modeling system across a range of timescales (decadal, seasonal, and short-term forecasts). Probabilistic forecasting, inundation modeling and coastal flooding were also identified as priorities by Bari and others. Inundation modeling is not very much on the Bureau’s radar at the moment, though. When asked if Robertson knew about the Bureau’s priorities around short-term forecasting research with WIRADA, he replied that he received a list of priorities that morning to be discussed at a meeting next week. 

While the Bureau is a source of data and forecasts, it has an uneven policy on access to external partners. For example, POAMA and TIGGE retrospective forecasting datasets are open and freely available. ACCESS retrospective forecasting datasets are available for $1000/year after a one-time setup fee and recipients must abide by a distribution agreement. Other datasets such as the sub-daily data being used to develop the short-term streamflow forecasting service are not available under any conditions except to certain CSIRO researchers. The Bureau even has issues with exchange of data within the organization itself. There were (unanswered) questions about the availability of SEACI data. French researcher Maria Helena Ramos also has useful forecasting datasets that she has offered to share. 

The group discussed the use of OzEWEX to negotiate data sharing agreements for researchers. There is the possibility of using OzEWEX as a platform for redistributing data but the first preference would be to get data from the providing agencies. For example, the OzEWEX data workgroup is first focused on data discovery and advertising (but not necessarily hosting) existing datasets. 

In the original group charge, it is mentioned that “this WG will design and carry out hind-cast experiments to compare the skill of alternative statistical and/or dynamic prediction techniques. This will build on the hind-cast and verification experiments that have occurred in WIRADA and the BoM Extended Hydrological Prediction section, and likely involve the same or a similar set of experiment catchments.” Considering that other working groups are focusing on model evaluation and benchmarking, this working group should place special emphasis on a retrospective forecasting experiment, not just model simulations. 

The experiment should be small, with small datasets, narrowly defined science questions and a clear purpose. It should also be relevant to the Bureau’s interests, such as by using their models of interest and/or studying catchments of high interest. The group considered attracting international participants to contribute to the experiments. Also, what are the overlaps of this experiment with HEPEX activities? 

The group questioned if the right people were at the table- are there any forecasting researchers that we don’t know about? The ARC webpage has information about grants to universities. Melbourne, Newcastle, New South Wales, Monash and Flinders are relevant universities. WIRADA CSIRO is active in forecasting. Private companies such as SKM, Deltares, Entera, SMEG, SeqWater are involved with forecasting. 

In terms of interfacing with the Bureau, the short-term and flood forecasting groups have a strong emphasis on implementation of Deltares’ FEWS forecasting system. It is hoped, as it was in the US with the Community Hydrologic Prediction System (CHPS), that FEWS would enable better integration of research products into operations. However, some researchers are finding that, if anything, FEWS is another barrier (not an enabler) that researchers must overcome to be relevant to operational forecasters. Operational forecasters in the US have complained that FEWS does not have the global support community that they thought it did. 

To this end, all WG6 participants are invited to list their interests in a shared drop box document to be initiated shortly. The next WG6 meeting will be announced pending the finalization of this shared document of purpose, but preferably within a time frame of 1-2 months.