All Content © CRC for Forestry 2007

Mr David Lazaridis

David LazaridisMr David Lazaridis

PhD student

Topic: Using shrinkage regression to help predict tree mortality

University of Melbourne

I graduated from the University of Melbourne with a BSc (Hons) in statistics, in which I learned how useful statistics can be to solve many really interesting problems. My main interest is shrinkage regression, which is the use of constraints or penalties to shrink model estimates.

My project involves the use of such regression techniques to measure and predict tree mortality using time-series data obtained from MODIS (MODerate resolution Image Spectroradiometer) mounted on the two Earth observation satellites, Terra and Aqua.

Due to the nature of the data and infrequent aerial inspections of forest health, analysis is not straightforward and much further research is required to produce powerful and reliable tools or models that can be used to diagnose and predict forest health.

Our key challenge is to explore and develop techniques that can best use this data to predict forest health. This has us looking at many statistical techniques that we have to adapt to best suit our problem.

In the initial phase of my project I am looking at various regression techniques, model selection strategies, time-series analysis and correlation structures.

Two overarching concerns are prediction performance and explanatory power. We need to work with tools that give us good prediction characteristics and tools that help us understand the underlying problem or phenomenon.

My supervisors are Dr Andrew Robinson from the Department of Mathematics and Statistics at the University of Melbourne, and Dr Jan Verbesselt, Remote Sensing Team, CSIRO Sustainable Ecosystems.

My research contributes to the CRC for Forestry’s Research Program One: ‘Managing and monitoring for growth and health’.

David Lazaridis - hiking