“Benthos” is a project to explore and generate fulldome visualisations of volumetric and multidimensional datasets derived from Australian Antarctic Division, CSIRO and international oceanographic and marine science programmes.
This page details some of the work-in-progress towards those visualisations.
Enormous amounts of data have been generated during the International Polar Year (IPY) during the extensive international CAML (Census of Antarctic Marine Life) and ongoing marine research activities by the AAD and other marine-science research institutions in Australia and internationally. These include datasets measuring physical features such as ocean currents, temperature gradients, thermo-haline cycles as well as ecological features such as marine biodiversity; benthic and pelagic ecology (deep ocean; continental shelf; shoreline) and so forth. Part of this project involves identification of suitable datasets for volumetric and three-dimensional visualisation.
Volumetric visualisation in a fulldome format explores both aesthetic and empirical features of datasets, creating a fertile intersection of the sciences and arts. Within the field of contemporary media arts little has been done with fulldome data visualisation, however, there is an emergent field of data-arts, which draws upon scientific datasets and explores them in new and insightful ways – the viewer’s immersion in a visualised dataset reveals not only the inherent beauty of normally unseen structures but can also elicit an understanding of complex connections between parameters that may not have been revealed via conventional analytic approaches.
Benthos is supported by a Synapse Art-Science residency via the Australian Network for Art and Technology and the Australia Council for the Arts. It is also generously supported by the Australian Antarctic Division and the Western Australian Supercomputer Program at the University of Western Australia.
Dr Peter Morse, Visualisation Consultant and Media Artist. www.petermorse.com.au
Dr Martin Riddle, Senior Principal Research Scientist (Programme Leader, Environmental Protection and Change), Australian Antarctic Division.
Dr Steve Nicols, Senior Research Scientist (Program Leader, Southern Ocean Ecosystems), Australian Antarctic Division.
Dr Ben Raymond, Data Mining & Visualisation, Australian Antarctic Division.
2.0 Data Visualisation:
Data visualised includes resources such as global topography/bathymetry (eg. GEBCO, ETOPO1); volumetric ocean models; sea ice data; krill observation datasets; deep sea ocean trawl video; CTD profiles (conductivity/temperature/depth); sub-surface buoys; sonar information; long-range remote sensing data; voyage track data; marine life track data; electron microscopy data.
The sheer amount of data available via Australian and international scientific programmes is overwhelming: it is a truly remarkable intellectual and observational achievement of humanity. When you begin to dig deeper into available resources you begin to realise what a credit it is to the public nature of science that thousands and thousand of individuals all over the world have given of their time, not only professionally, but personally as well, to share their hard-won and vastly complex knowledge about the world that we live in. This is also a function of the time we live in, when information can be made globally accessible via the internet and when download speeds enable the transaction of huge datasets, many of which are measured in terabytes and, in the near future, in petabytes and exabytes. Needless to say very fast networked supercomputers with large amounts of RAM and high-end GPUs are also crucial to this developing area.
In this project I can barely scratch the surface of what is possible – so this presents an initial problem – where to start? Having given this substantial thought I decided upon a deceptively simple question:
3.0 “What is the shape of the sea?”
It has become apparent to me that we have a deeply terrestrial view of the world – ordered by a whole series of anthropocentric conventions – and that in looking at and attempting to understand the sea – much of this apparatus must be discarded. It is similar to the feeling of looking at a world-map upside down: the familiar landmarks, the topology, are estranged and made unfamiliar – we lose our bearings. The paradox then, in order to “find” the sea as a globally encompassing body of water, is precisely to explore this ‘seeing the aspect‘ as an advantage.
When we look at global topographic and bathymetric models of the Earth, what we see is the surface of Earth – the rock surface, the geology: not the fluid world. The dataset is constructed from millions of point readings of the elevation of that surface in relation to ‘nominal sea level’ – itself determined in relation to the Earth geoid (WGS84 and other models) that mathematically describe the complex shape of our planet. This is a very complex area that I can only gloss here – as a non-expert – so I encourage you to read the links for clarification.
Remarkably, in my research into this question, I have found no images at all of what the sea would look like if we simply saw it as a self-contained volume – the largest biosphere on the planet, in which most of the world’s life exists, in an entirely different volumetric and physical way to which we perceive and interact with the world. So I posed this question to Paul Bourke at the WASP: how would it be possible to construct a volume representation of the global ocean? Isn’t it a matter of ‘simply’ subtracting the GEBCO topography/bathymetry from the WGS84 geoid? The answer is, principally, yes, but as Paul figured out, the solution is more complex than it seems – and involved writing some software that could accurately calculate the intersection of “mean sea level” with global coastlines as well as retaining the bathymetric structure of the ocean floor – at many resolutions. This has led to the development of a new dataset derived from GEBCO that creates an accurate high-resolution polygonal mesh of the global ocean. In itself this is not, strictly speaking, volumetric (we will cover this later on in detail), as the model derived is a polygonal closed solid that represents the surficial shape of the ocean – however, it is a remarkable object and gives us a unique view of the world’s global oceans.
Here’s a short first-attempt quicktime movie visualising this polygonal volume, as if the global ocean was frozen at a point in time and cast in blue glass (which is precisely what we’re going to do):
A visualization of the global ocean derived from the GEBCO dataset. The entire earth has been removed and only the ocean volume remains.
See Paul’s notes upon deriving the model: http://local.wasp.uwa.edu.au/~pbourke/miscellaneous/oceans/
Important note: the depth exaggeration is a factor of 200 – the ocean, like the atmosphere, is a thin veil across the surface of the earth, despite its prodigious depths. If it was not exaggerated we would barely see it in this model.
This model has been constructed at a range of resolutions ranging from 1 degree to 1 minute (1/60th degree); the 3D meshes range from ~8MB to 270MB+ (fairly substantial for a 3D mesh object): this enables animations and visualisations at a wide range of levels of detail, such that it is possible to zoom in closely on certain parts of the ocean at sufficient resolution to retain detail, whilst accurately locating finer regional datasets.
An important point to make here is that we are looking at current scientific data that accurately represents the shape of the global ocean. It is not a mock up or special effect – it is as close as can be currently achieved using extant data resources and those facilities available to me. The underlying data is always undergoing revision and development as more and more data becomes available, so at some point this view will be superceded.
Another important consideration is scale – the ocean is vast and deeply complex – so, despite there being large scale inspection of the ocean via satellite and aerial observation, on finer levels and in significant areas and in significant ways it is also barely known. Sampling points made on ship tracks and via subsurface buoys and CTD samplers across the ocean form minuscule traces on this surface and hardly any in relation to its volume – this is not because the work isn’t being done, but simply because of the fact that the ocean is so huge and ships and samplers and people are relatively so small. So it is often meaningless to interpolate data between datapoints – inferring that the sea is often understood as a series of discrete systems of which the relationship between certain large scale interactions and structures and regional and small-scale manifestations are poorly understood.
From a systems point of view – how does one relate the very large to the very small, the discrete to the contiguous/conterminous? This is made more difficult – from my point of view – in that it raises the question of data synthesis. How meaningful is it to layer different types of data within a model? What are the correlations between them? Do they follow the same coordinate and projection geometries? How is the data formed? What are the filetypes that encapsulate the data and how can they be transformed so they are workable-with – does this ‘translation’ maintain their integrity?
A consequence of these questions is to develop a methodology to work through the morass of information and computational processes required – and, hopefully, in the end, achieve a workable synthesis that does not diverge too far from the reality. I have no expectation that the results will picture the ‘real’ ocean – this is properly the domain of scientific specialists – it will be more a picture of the ‘real’ data. However, if it demonstrates a productive way of thinking that enables new scientific and cultural/aesthetic intuition about the ocean and its processes – upon which our life is ultimately contingent – then that will be a satisfying result.
4.0 Volumetric Visualisation of the Global Ocean.
More information coming soon, but in the meanwhile – a preliminary visualisation:
Preliminary visualisations of the Global Ocean as a volume. Derived from the CARS 2009 data (Jeff Dunn, CSIRO), massaged by Paul Bourke (WASP), for a current project I am working on to visualise the world’s oceans using a new synthesis of scientific datasets, for fulldome display. The objective is to map as many types of datasets as possible (within my limited resouces) into a spherical volumetric visualisation of the world’s oceans – to visualise a kind of “digital ocean.”. Hopefully something interesting will emerge – it’s blue sky or blue ocean or blue data stuff.
5.0 Volumetric Visualisation of CEAMARC Trawl Video
Remapping video from time series into spatial series demonstrates the possibility of creating volumetric visualisations of video. The video is exported as a series of frames (selecting every 10th frame from a short sequence in this example); then using volume visualization software the frames are image processed and remapped into a volume representation, using transfer functions to elicit structure and detail.
This is very much at an experimental stage.