Research student profile: Michael Nelson
Project title
Spatial Generalized Linear Mixed Models in Digital Soil Class Mapping
Project overview
My project aims to apply the framework of Spatial Generalized Linear Mixed Models to the mapping of soil classes.
While previous research has concentrated on applying the Spatial Generalized Linear Mixed model framework to mapping soil properties, this is the first work to apply it to mapping multinomial soil classes. The Spatial Generalized Linear Mixed model framework gives previously unavailable a rigorous theoretical basis for quantifying prediction uncertainty and designing sampling schemes for digital soil class mapping.
To tackle this project my research involves a thorough understanding of the Spatial Generalized Linear Mixed model framework to develop a valid and workable model for spatially correlated multinomial data. To implement my methodology I am using the R Programming language, as well as coding in FORTRAN.
Findings form this work will enable digital soil class maps to be produced with quantifiable levels of prediction uncertainty. Sampling schemes for soil class mapping will able to be produced that optimize sampling locations so as to minimize cost and maximize the information available.
Background
I am an local student from Sydney and completed my Bachelor of Land and Water Science at the University of Sydney (University Medal) in 2007. I have been granted a Vice Chancellor’s Research Scholarship.
I received the University Medal on graduating from B L W Sc at the University of Sydney in 2007. I was placed on the Dean's List for Academic Excellence for the Faculty of Agriculture, Food and Natural Resources in 2004, 2005, 2006 and 2007 and was awarded the Belmore Scholarship for Soil Science in 2005, Continuing Undergraduate Scholarship in 2005, Dean's Scholarship for Academic Excellence in 2006.
In the first year of my postgraduate studies I had the fortune to travel to Palmerston North, New Zealand to attend the Joint Conference of the Australian and New Zealand Soil Science Societies - 'Soil - the living skin of Planet Earth" where I presented a paper on a genetic algorithm approach to the creation of digitial soil class maps using legacy soil data and ancillary information.
Publications
- Nelson, M. A. & Odeh, I. O. A. (2008) Creation of digital soil class maps using legacy soil data and ancillary information: a genetic algorithm approach, Proceedings of Soil 2008 - the living skin of Planet Earth.