- ARC Centre of Excellence in Mathematical & Statistical Frontiers (ACEMS), Associate Investigator David Gunawan, 2017 – 2021
- ARC Centre of Excellence in Mathematical & Statistical Frontiers (ACEMS), Associate Investigator Matt Moores, 2018 – 2021
- ARC Discovery Early Career Research Award (DECRA), Deep space-time models for modelling complex environmental phenomena, Andrew Zammit-Mangion, 2018 – 2021
- ARC Discovery Project, Bayesian inversion and computation applied to atmospheric flux fields, Noel Cressie and Andrew Zammit-Mangion, 2019 – 2022
- ARC Special Research Initiative, Securing Antarctica's environmental future, Noel Cressie and Andrew Zammit-Mangion, 2021 – 2028
- ARC Industrial Transformation Research Hub (ITRH) for Transforming Energy Infrastructure Through Digital Engineering, Andrew Zammit-Mangion, Matt Moores, and David Gunawan, 2021 – 2025
- ARC Discovery Project, Bayesian inference for psychological theories with intractable likelihood, David Gunawan, 2021 – 2024
In the pages below, we provide a brief and somewhat informal description of our current research in environmental informatics. All research topics are linked to pages with more details and references.
- Statistical remote sensing
- Flux inversion of atmospheric trace gases
- Space-based research in spectroscopy
- Multivariate stochastic modelling
- Other projects
Statistical remote sensing
Remote sensing of geophysical variables almost always involves indirect measurements of energies in relevant bands of the electro-magnetic spectrum, taken by an instrument on board a satellite or some other flight vehicle. For any one measurement at a given location and time, inference on states of the atmosphere is carried out in the presence of uncertainty.
Flux inversion of atmospheric trace gases
Flux inversion is the process through which we locate and quantify the sources and sinks of a gas using measurements of that gas at different points (or along different paths) in space and time. All flux inversion procedures need to take into account meteorology that can be used to predict how gas particles move in the atmosphere when released from a source. At CEI we are working on flux inversion at three scales, which we classify as local, regional, and global.
Local flux inversion
Local flux inversion is concerned with locating and quantifying point sources and sinks in a small area of interest, generally on the order of a few thousand square metres or a few square kilometres. At these scales, simple atmospheric models can be used, and precise quantification of emissions can be done. (Image credit: Stephan Muller)
Regional flux inversion
Regional flux inversion is concerned with locating and quantifying sources and sinks at much larger scales (spanning whole countries, or a continent). This inversion setup is useful for validating a country's emission portfolio, for example, to ensure that it is complying with any emission targets in place. (Image credit: NASA)
Global flux inversion
Global flux inversion is concerned with locating and quantifying sources and sinks at the global scale. Inversion at this scale requires the use of global atmospheric transport models, as well as sophisticated computational machinery to deal with massive data sets. (Image credit: NASA)
Space-based research in spectroscopy
Understanding the chemical composition of mineral and biological samples is vital for many practical applications, including the search for evidence of past life on Mars, environmental monitoring of coral health and water quality, and medical diagnostics of cancer and other diseases. (Image credit: NASA)
Multivariate stochastic modelling
Multivariate geostatistics is based on modelling all covariances between all possible combinations of two or more variables at any locations in a continuously indexed domain. Multivariate spatial covariance models need to be built with care, since any covariance matrix that is derived from such a model has to be nonnegative-definite. In our research we are trying to find new ways to easily construct valid complex multivariate covariance models. Two such ways that we present below are through conditioning, and through domain warping.
When using conditioning for model construction one only needs univariate covariance functions to yield valid multivariate models. The approach we develop has a vast array of applications since it features conditional dependencies on a potential causative process. The project on regional methane flux mapping given on the page, Regional flux inversion, is one such application.
Models on warped spaces
Complex multivariate spatial covariance models can be constructed by taking a very simple, valid, multivariate spatial covariance model, and then warping the spatial domain on which it is defined. We find that warping can easily induce properties such as nonstationarity and asymmetry.