Emlyn Jones

Assessing the impact of observations on ocean forecasts and reanalyses: Part 2, Regional applications

The value of global (e.g. altimetry, satellite sea-surface temperature, Argo) and regional (e.g. radars, gliders, instrumented mammals, airborne profiles and biogeochemical) observation-types for monitoring the mesoscale ocean circulation and biogeochemistry is demonstrated using a suite of global and regional prediction systems and remotely-sensed data. A range of techniques is used to demonstrate the value of different observation-types to regional systems and the benefit of high-resolution and adaptive sampling for monitoring the mesoscale circulation. The techniques include Observing System Experiments, Observing System Simulation Experiments, adjoint sensitivities, representer matrix spectrum, observation footprints and spectral analysis. It is shown that local errors in global and basin-scale systems can be significantly reduced when assimilating observations from regional observing systems.

Bayesian learning and predictability in a stochastic nonlinear dynamical model

Bayesian inference methods are applied within a Bayesian hierarchical modeling framework to the problems of joint state and parameter estimation, and of state forecasting. We explore and demonstrate the ideas in the context of a simple nonlinear marine biogeochemical model. A novel approach is proposed to the formulation of the stochastic process model, in which ecophysiological properties of plankton communities are represented by autoregressive stochastic processes. This approach captures the effects of changes in plankton communities over time, and it allows the incorporation of literature metadata on individual species into prior distributions for process model parameters. The approach is applied to a case study at Ocean Station Papa, using particle Markov chain Monte Carlo computational techniques. The results suggest that, by drawing on objective prior information, it is possible to extract useful information about model state and a subset of parameters, and even to make useful long-term forecasts, based on sparse and noisy observations.

Applications of heterogeneous computing in computational and simulation science

As the size and complexity of scientific problems and datasets grow, scientists from a broad range of discipline areas are relying more and more on computational methods and simulations to help solve their problems. This paper presents a summary of heterogeneous algorithms and applications that have been developed by a large research organization (CSIRO) for solving practical and challenging science problems faster than is possible with conventional multi-core CPUs alone. The problem domains discussed include biological image analysis, computed tomography reconstruction, marine biogeochemical models, fluid dynamics, and bioinformatics. The algorithms utilize GPUs and multi-core CPUs on a scale ranging from single workstation installations through to large GPU clusters. Results demonstrate that large GPU clusters can be used to accelerate a variety of practical science applications, and justify the significant financial investment and interest being placed into such systems.

Marine Biogeochemical Modelling and Data Assimilation

The inclusion of biogeochemistry into the Global Ocean Data Assimilation Experiment systems represents an exciting opportunity that involves significant challenges. To help articulate these challenges we review marine biogeochemical modeling and the existing applications of biogeochemical data assimilation. The challenges of biogeochemical data assimilation stem from the large model errors associated with biogeochemical models, the computational demands of the global data assimilation systems, and the strong non-linearity between biogeochemical state variables. We use the ocean state estimation problem to outline an approach to adding biogeochemical data assimilation to the Global Ocean Data Assimilation Experiment systems. Our approach allows the biogeochemical model parameters to be spatially and temporally varying to enable the data assimilation system to track the observed biogeochemical fields. The approach is based on addressing the challenges of biogeochemical data assimilation to improve both the state estimation of the biogeochemical fields and the underlying biogeochemical model.

Development of operational data-assimilating water quality modelling system for South-East Tasmania

With the rapid advances in on-line observing system applications, the paradigm in environmental modelling is shifting from one-off models for specific purposes, to operational models, sequentially assimilating data streams from in situ and remote sensors. Such models can provide products and services to support a wide range of applications, from short-term forecasting to long-term scenarios, and are expected to deliver superior performance much more cost-effectively. In the marine field, this is most advanced for circulation models at large ocean scales. The potential benefit from these advances is even greater in the coastal zone, where human uses, impacts and ecosystem services are concentrated. However, there are substantial challenges to be overcome. Coastal applications typically require biogeochemical, ecological, and ultimately socioeconomic models. These additional models are more complex, with higher uncertainty, and require different approaches to data assimilation and uncertainty analysis. The uncertainties arise from a number of sources including poorly known parameters, structural errors and stochastic forcing. When model realisations are sufficiently fast, Monte Carlo techniques can be used to improve the model performance and assess its quality, otherwise alternative estimation techniques are required. This paper describes the development of an operational, data-assimilating coastal model for SE Tasmania, integrating across hydrodynamics, sediment dynamics and biogeochemistry. Inputs and outputs from the model are expected to be integrated into the regional information system (INFORMD), and to be used directly in multiple management applications, and as input into ecosystem models. A hydrodynamic model, nested inside an operational global model, will be assimilating data from the coastal sensor network and other sources, including remote sensing. The model is based on an operational modelling platform developed by CSIRO through the BlueLink project (ROAM), and will be used to implement and test data-assimilation techniques for coastal models under development in BlueLink. Operational sediment dynamic and biogeochemical models, will be coupled to the hydrodynamic model, either directly or through intermediate transport models. Data-assimilating techniques for these models currently are under development in Computational and Simulation Sciences theme, CSIRO. This paper outlines preliminary results from these developments. A number of candidate techniques including Kalman Filter, Particle Filter and MCMC are discussed. The utility of fast and cheap statistical surrogates of complex models (emulators) for sequential data assimilation is illustrated through the trial application of emulators to one-dimensional sediment/pollutant and 3-d sediment transport models.

Hierarchical Emulation & Data Assimilation into the Sediment Transport Model☆

Abstract Synthetic observations of the suspended sediment concentration in an idealised macro-tidal estuary are assimilated into the 3d sediment transport model. The assimilation scheme relies on fast and cheap surrogates of the complex model (called emulators) to update the models state variables and its 2 parameters. A scenario with a hierarchically structured emulator is contrasted to the scenario with a more conventional non-hierarchical emulator. Numerical experiments indicate that for a given size of the ensemble an emulator which replicates a hierarchical structure of the model tends to provide a better approximation of that model. Improving the quality of the emulator translates into the improved quality of the assimilation products.

Identification of Patterns over Regional Scales Using Self-Organising Maps on Images from Marine Modelling Outputs

The Self-Organizing Feature Map (or SOM), has been used to analyse a dataset consisting of oceanographic modelling output images, in order to identify patterns in the hydrodynamic behaviour of the south-east Tasmanian (SETas) coastal region over a 360-day period between August 2009 and August 2010. The SOM provided a visualization of the dataset, distributed across a 5x7 two- dimensional grid, which enabled an oceanographer to identify significant hydrodynamic patterns being exhibited by the SETas region over that period. Four prototype (typical) states were identified by the oceanographer, who then interpreted each of these states in terms of the major ocean currents which impact on the region, the East Australian Current and the Zeehan Current. These results indicate that SOM analysis can be a useful technique for identifying patterns in large oceanographic datasets, such as those now being provided by remote sensing, ocean modelling and marine sensor network technologies.