Ron Hoeke1, Claire Trenham2, Julian O’Grady3, Robert Davy4, Rebecca Gregory5, Kathy McInnes6, Mark Hemer7
1CSIRO Oceans & Atmosphere, Aspendale, Australia, Ron.Hoeke@csiro.au
2CSIRO Oceans & Atmosphere, Aspendale, Australia, Claire.Trenham@csiro.au
3CSIRO Oceans & Atmosphere, Aspendale, Australia, Julian.O’Grady@csiro.au
4CSIRO IM&T, Canberra, Australia, Robert.Davy@csiro.au
5CSIRO Oceans & Atmosphere, Aspendale, Australia, Rebecca.Gregory@csiro.au
6CSIRO Oceans & Atmosphere, Aspendale, Australia, Kathleen.Mcinnes@csiro.au
7CSIRO Oceans & Atmosphere, Hobart, Australia, Mark.Hemer@csiro.au
Coastal communities and maritime activities face a number or risks from coastal hazards associated with storm waves and flooding. These risks are generally increasing due to sea level rise, but other factors define local risk. These factors include: astronomical tides; severe weather-induced storm surges (caused by low atmospheric pressure and strong onshore winds); wind and swell waves caused respectively by local and remote weather systems; and other variations in local sea levels that occur on intraseasonal to interannual time scales. Efforts to better understand and predict coastal sea levels has led to a proliferation of both modelled and observed data. The Sea-level Rise, Waves, and Coastal Extremes team in CSIRO’s Climate Science Centre (CSC) maintain and archive many of these datasets. However, some datasets were stored on tape and in a data structure that made time-series analysis prohibitively slow, and analysis code was held by individual researchers and not managed in a version control system. Also, the datasets have not been regularly updated with the most recent data. Therefore, there was no systematic and computationally economical means of accessing the data for the purposes of better understanding their contributions to extreme coastal sea levels for multiple locations or for large gridded datasets.
As part of the National Environmental Science Programme’s (NESP) Earth Systems and Climate Change Hub, and with the support of the CSIRO eResearch Collaboration Projects programme, we are developing a suite of software (based on python and R) in a shared code repository and improving our data structures to deliver improved science and services (see Figure 1). The scope of the eResearch project is that it will also facilitate the delivery of future projects. Initially, we have focussed on the CAWCR Wave Hindcast [1,2], but are currently adding in other datasets pertaining to tides, sea level, and storm surge, both based on observations and models.
Figure 1: Flow chart showing path from original data structure to value-added data delivery.
In this presentation, we discuss how we have solved a number of problems that held us back from rapid and repeatable analysis and show a sea level prediction tool we have developed for use by the wider community. In particular we will discuss:
- Data restructuring (wave hindcast)
- Data aggregation: co-locating all relevant datasets into disk-based cloud devices
- Software development (migration of code into a version control system and subsequent restructuring)
- Updating the data with automated monthly jobs.
- Examples of how the resulting rapid access data can be quickly and effectively analysed though statistics/extreme value analysis and viewed using online tools (e.g. Rshiny)
These improvements allow maintenance of up-to-date scientific datasets and rapid assessment of output for various applications . For example, we now have the capability to perform extreme value analysis (e.g. full generalised Pareto fitting) of gridded wave hindcast output in parallel. An example of this is illustrated by Figure 2, which shows the extent of extreme waves during an East Coast Low event that occurred in June 2016 . Prior to starting this work, the computational effort to extract this data, calculate the statistics, and combine it over the grid was infeasible. As a result of improving data structure and performance, we are now able to produce these results in approximately 20 minutes.
Figure 2: Example of extreme storm event analysis for the East Coast Low that caused extensive erosion and coastal flooding along much of east coast NSW and Tasmania between June 4-6, 2016. The map to the left represents the maximum significant wave height (Hs) over the duration of the event (several days), normalised by the 50-year average return interval, i.e. map colours greater than 1 indicates storm wave heights greater than the 1-in-50 year event.
We also discuss how such improvements are being integrated into a number of coastal hazard assessment tools. An example of how we will make data available to users is through the Rshiny software package. Using Rshiny we are developing an interactive webpage to access the extreme water levels dataset for coastal locations around Australia which are generated in the rapid-access pipeline software.
- Durrant T, Greenslade D, Hemer M, Trenham C. Global Wave Hindcast focussed on the Central and South Pacific. CAWCR Tech. Rep. 70, 1–54. 2014.
- Hemer M, Zieger S, Durrant T, O’Grady J, Hoeke RK, McInnes KL, Rosebrock U. A revised assessment of Australia’s national wave energy resource. Renewable Energy. 2016. doi:10.1016/j.renene.2016.08.039
- Davy R; Hoeke RK; Trenham C, O’Grady, J, Hemer M, Gregory R, McInnes K. Fast time series analysis of wave hindcast data. In: Collaborative Conference on Computational and Data Intensive Science; Melbourne. 2018.
- Hoeke RK, Stephenson A, McInnes K, Davy R, O’Grady J, Hemer M, Williams G. A multivariate statistical retrospective of two high-impact coastal events. Australasian Coasts & Ports 2017: Working with Nature. 2017:593.
Claire joined CSIRO Marine & Atmospheric Research in 2011 working on ocean wave climate modelling. After a period working as a Senior Research Data Services Specialist for the National Computational Infrastructure (NCI) in Canberra between 2014 and 2017, she returned to CSIRO’s Oceans & Atmosphere division in 2017. She currently works in the Sea level rise, waves and coastal extremes team alongside the regional climate team as part of the Climate extremes and projections group. Claire is heavily involved in climate data and preparation for CMIP6, as well as regional climate modelling, data processing, making improvements to data and software to enhance science capabilities, and supervising student volunteers in data digitisation. In a past life she was a radio astronomer, and is also a qualified high school maths/science teacher.