Integrating remote sensing and in-situ observations to analyse vegetation greenness in Australia's changing climate

Abstract

Climate change as driven by rising atmospheric CO2 has increased the variability of precipitation and temperature, thereby impacting fundamental ecological processes in water-limited ecosystems. Projecting the impacts of climate change in an Australian environment requires an accurate understanding of its vegetation dynamics. Although remotely sensed data offer long-term monitoring of vegetation on a continental scale, on-ground occurrences and vegetation dynamics are not always captured clearly. Recent survey efforts have expanded and produced nationwide in-situ vegetation data including species identification and abundances, soil and topographic properties, and fractional cover (FC), providing ample opportunities to integrate with spatiotemporally extensive remotely sensed data to capture these vegetation dynamics. In this thesis, I examined spatiotemporal patterns in remotely sensed FC from Digital Earth Australia (DEA FC) time series (1987-2023) at 735 ground monitoring sites established by AusPlots Survey. In Chapter 2, I investigated whether the timings of survey visits can reproduce long-term trends in FC as captured by DEA FC and the agreement between DEA and AusPlots FC. Accordingly, I explored the long-term trends in DEA FC. In Chapter 3, I sought to identify what environmental factors (e.g. climatic, topographic, and soil variables) are important for modelling FC and how these importances differ between biomes. The modelling results demonstrated good utility for understanding the effects of environmental variables on different biomes and the development of dynamic vegetation models. The partial effects of CO2 in different biomes appeared to be consistent with the differences in greening trends between the bioclimatic classes, clarifying how elevated atmospheric CO2 affects different regions of Australia. However, the sensitive nature of Desert ecosystems to the intra-annual variability of precipitation and their interaction with widespread increases in VPD should be considered when projecting the greening trends in these ecosystems. Additionally, as both the DEA-AusPlots FC agreement and my modelling performance were limited in the Desert region of Australia, this region should be considered carefully when developing FC products and modelling procedures.

Date of Award	2024
Original language	English
Awarding Institution	Western Sydney University
Supervisor	Belinda Medlyn (Supervisor) & Juergen Knauer (Supervisor)