Student research opportunities

Blending multiple data sources into a single higher quality one CSIRO PhD top-up $15000 per year available for application

Project Code: CECS_741

This project is available at the following levels:
PhD
Please note that this project is only for higher degree (postgraduate) applicants.

Keywords:

Intelligent computation, remote sensing, applications

Supervisors:

Dr Warren Jin
Professor Tom Gedeon

Outline:

In environmental science, an object can often be regularly measured or estimated by different sources. For example, soil moisture in Australia can be retrieved from sensors onboard several remote sensing platforms, including ASCAT onboard the EUMETSAT METOP satellite, AMSR-E onboard NASA's Earth Observing System (EOS) Aqua Satellite, and ASAR onboard the Evensat satellite. Water quality may be assessed by accurate but temporally sparse grab sampling and also by high frequency optical sensors. Many environmental applications however demand a single high quality data set, often covering the greatest time extent or having the highest temporal/spatial resolution. For example, climate models and hydrological models may both require long term daily high resolution soil moisture information. This project will focus on developing effective computation or modelling techniques for combining several sources of data into a high quality blended product that respects the inherent accuracies and uncertainties of the component products. These techniques need to handle several challenges, including different temporal and spatial resolutions or alignments, heterogeneous uncertainties, different collection periods, contrasting measurement scales, and so on. Approaches will initially be investigated with Australian soil moisture data.

Goals of this project

The objective of this project is to develop and implement effective statistical computation techniques for incomplete observational data which can be widely applied to environmental problems.

Several directions can be examined to achieve the objective. One possibility is to smooth/blend one or more time series into a complete series via a state space model (Shumway & Stoffer, 2006) before assimilating with some biophysical model. For example, daily soil moisture time series can be blended from data sensed from AMSR-E (Advanced Microwave Scanning Radiometer - Earth Observing System) and ASAR (Advanced Synthetic Aperture Radar). Daily precipitation can be blended from weather station observations with remotely sensed data. A key challenge is to choose an appropriate state transition equation which can reflect underlying dynamics. Another possibility is to embed a missing data processing mechanism directly into the filtering procedures of a state space model. For example, the observation equations can be adjusted according to different availability of observations (Durbin & Coopman, 2001). If there exists spatial or cross variables correlations, spatio-temporal models should be constructed appropriately before efficient computation methodology.

Requirements/Prerequisites

Applicants are expected to have a major in computer science, information technology, or applied mathematics/statistics.

Preferably with strong background in data mining, machine learning, or statistics.

Preferably with excellent programming skills (C/C++, R, and/or MatLab)

Student Gain

A student working in this project can expect

to learn state-of-art of state space modelling techniques;

to be involved in developing cutting-edge soil moisture or evaporation products, especially an operational system for Australia;

to gain experiences on solving real-world challenges while working with a research group delivering great science and innovative solutions for Australian society and economy;

Supplementary PhD scholarship available from CSIRO $15000 per year for three years, subject to a separate application to CSIRO

Background Literature

P. Kokic and H. Jin. Parameter estimation using the EM algorithm in a multivariate state-space model with missing data. Under review

H. Jin and B. Henderson. Towards a Daily Soil Moisture Product based on Incomplete Time Series
Observations of Two Satellites. MODSIM'2011, Dec 2011, pp. 1959-1965. ISBN: 978-0-9872143-1-7

Cressie, N., T. Shi, and E. L. Kang (2010), Fixed Rank Filtering for Spatio-Temporal Data, Journal of Computational and Graphical Statistics, 19(3), 724-745, DOI 10.1198/jcgs.2010.09051;

Liu, Y. Y., R. M. Parinussa, W. A. Dorigo, R. A. M. de Jeu, W. Wagner, A. I. J. M. van Dijk, M. F. McCabe, and J. P. Evans (2010), Developing an improved soil moisture dataset by blending passive and active microwave satellite-based retrievals, Hydrology and Earth System Sciences Discussions, 7(5), 6699-6724, doi:10.5194/hessd-7-6699-2010;

Robert H. Shumway, David S. Stoffer (2006). Time Series Analysis and Its Applications: With R Examples. Time Series Analysis and Its Applications: With R Examples (Springer Texts in Statistics)