Skip to the content | Change text size

Climatology - Research - GES

Australian Savanna Landscapes: Past, Present and Future

  • Investigators
  • Summary
  • Project Overview
  • Aims and Background
  • Approach
  • Data
  • Links
  • Acknowledgements


    • Professor Jason Beringer (Monash University, Melbourne, Australia)
    • Associate Professor Lindsay Hutley (Charles Darwin University, Darwin, Australia)
    • Professor Qiang Yu (University of Technology, Sydney, Australia)
    • Dr Vanessa Haverd (CSIRO, Canberra, Australia)
    • Professor Steven Higgins (Goethe University Frankfurt, Frankfurt am Main, Germany)
    • Dr Youngryel Ryu (Harvard University, Cambridge, USA)
    • Dr Stan Schymanski (Max Planck Institute for Biogeochemistry, Jena, Germany)
    • Dr YingPing Wang (CSIRO, Hobart, Australia)
    • Professor Mathew Williams (University of Edinburgh, Edinburgh, UK)
    • Dr Gab Abramowitz (University of New South Wales, Sydney, Australia)


    Australian savannas are productive and culturally and biologically significant landscapes, but are vulnerable to climate change. We will determine savanna function (carbon and water balance) for the present and assess how sensitive they have been to past climate variability. We then address how they may respond to future climate change.

    Project Overview

    Australian savannas cover 25% of the continent and are significant for grazing, cultural heritage and wilderness value. They provide important ecosystem services but are highly sensitive to climate change and variability, especially precipitation. This program will assess how Australian savanna systems function (carbon and water cycles) in the present, how sensitive they have been to past climate variability and how they may respond to future change. We identify gaps in models used to project climate and improve them to ultimately dynamically generate current, past and future savanna vegetation patterns and responses. Our models will provide tools for land managers to account for the consequences of climate change on ecosystem variability.

    The project is funded from 2013 to 2015.

    The full project proposal can be found here (PDF ).

    Aims and Background

    Currently, savannas comprise 15% of the global terrestrial surface and 25% of Australia (almost 1.9 million km2). Climate change is likely to alter the structure and function of savanna through shifts in available moisture (rainfall-evaporation), fire regime and CO2 concentration. This project seeks to explore the sensitivity of savanna carbon and water fluxes to climate at century time scales to better predict future responses. Land managers require tools and models that can account for the consequences of climate change and elevated atmospheric CO2 concentrations on ecosystem viability. Savannas are closely coupled with the atmosphere through both biophysical processes and biogeochemical cycles. Mechanistic models, that simulate this coupling, are required to understand the impacts of climate change on savanna structure and function. Yet, our ability to project ecosystem responses to climate change is constrained by a lack of field studies and capable models to project the long-term consequences of changing climatic variability (both past, present and future). We will build on a recently-completed ARC funded multidisciplinary field campaign "Savanna Patterns of Energy and Carbon Integrated Across the Landscape (SPECIAL)" undertaken during the dry season to understand the spatial patterns and processes of land surface-atmosphere exchanges (radiation, heat, water, CO2 and other trace-gasses). Australian savannas cover the top 1/3 of the continent and host much of the grazing industry in Australia, although conservation and Indigenous land covers a significant fraction of this area and this region has been described as one of the world's last great wildness areas and thus plays a significant function in providing ecosystem services (carbon cycle, water balance, biodiversity). The structure, composition and functioning of Australian savannas is largely controlled by climate, particularly rainfall and therefore they are highly sensitive to climate change, especially changes in the timing, quantity and variability of precipitation). Seasonal variability and inter-annual variability in rainfall are the critical drivers of structure and density of tree stands in these savannas. This is especially evident across the continent from the northern Australian coast (rainfall 1700 mm.yr-1 mm and woodland savanna) to arid interior Australia (rainfall 285 mm.yr-1 at Alice Springs and grassland). Past rainfall in this region has been highly variable due to varied monsoonal activity and variability associated with modes in climate such as El Niño Southern Oscillation (ENSO) and Inter-decadal Pacific Oscillation (IPO). Future anthropogenic climate change is projected to include increasingly variable precipitation regimes, as well as atmospheric warming. General circulation models forecast a higher frequency of extreme rainfall events from intense convective storms, a lower frequency of rainfall days, and longer intervening dry periods. Increasing atmospheric CO2 will have differential effects on wood plants (C3 pathway) vs grasses (C4 pathway) that comprise savanna vegetation types. Elevated CO2 will shift tree:grass ratios with the magnitude of the response likely to differ at high and low rainfalls. Increasing temperature and changing growing season length will also modify fire regimes. Given the poor understanding of the savanna ecosystem and its vulnerability to change, further systematic scientific study of the region is needed to sustainably manage these ecosystems.


    The overall aim of this project is to assess 1) how Australian savanna systems function (carbon and water cycles) in the present, 2) how sensitive they have been to past climate variability and 3) how they may respond to future change. This builds on SPECIAL which was an observational snap shot in time. We will use the strong rainfall gradient (Fig. 2) along the NATT where there is an associated change in savanna structure, composition and function as a framework for our observations and modelling. This transect is one of several transects originally established under the International Geosphere-Biosphere Program (IGBP) program to study the effects of changes in climate, land use, and atmospheric composition on biogeochemistry, surface-atmosphere exchange and vegetation dynamics of terrestrial ecosystems. The NATT covers a gradient of decreasing mean annual rainfall from 1700 mm in the north to 350 mm in the south. The Australian monsoon weakens with distance inland (quantity of rainfall declines), where rain comes from small, isolated convective storms (is more variable) rather than from extended periods of monsoon activity. Hence we can use a 'space for time substitution' to examine what may happen under a changing rainfall regimes (time) in the future by comparing it with sites along the transect (space). Most rainfall occurs during the summer wet season (November to April). The rainfall gradient is associated with changes in the structure, composition and function of the savannas. Savannas consist of an evergreen overstorey dominated by eucalypts (C3 photosynthetic pathway) with an understorey of tall grasses (C4) that senesce in the dry season. Accounting for the distribution and composition of tree and grass components is critical but is currently inadequate in models. Grasses subsequently become fine fuels for the frequent fires in the dry season. This proposal will not directly quantify emissions from fires but will use past estimates of biomass burning emissions and information of the impact of fire on savanna productivity.

    We focus on two key ecosystem indicators; 1) savanna productivity, which is defined as the Gross Primary Productivity (GPP) via photosynthesis and is the key input to the carbon cycle, and 2) savanna evapotranspiration (ET), which is the flux of water via evaporation from the land surface and transpiration by vegetation. ET is also known as latent heat flux (LE) is a key part of the hydrological cycle and water budget, with precipitation being the key input. In order to achieve our aim we will complete the following steps:

    1. Synthesise observations: We will utilise existing flux tower observations of GPP and ET and provide support for the ongoing measurement of carbon and water fluxes across the savanna landscape. This will be supplemented by ongoing flux measurements along the NATT.
    2. Model parameterisation:We will again utilise existing measurements of ecosystem characteristics from leaf to canopy to provide the necessary input data for models that will in turn be used to examine past, present and future variability of savanna fluxes (GPP and ET). Land surface models require input information (state variables) that describes the physical structure, species/functional composition and physiological performance. We will synthesise available information describing savanna ecosystem characteristics along the NATT.
    3. Present gaps and challenges for models: Model gaps will be identified by validating them against available flux observations (above 1). A suite of LSMs will be used to simulate the present carbon, water and energy fluxes for all seven sites along the transect. We will use multiple 'types' of models to assess strengths and weakness of each as each model 'type', each of which has a different philosophy and algorithms.
    4. Model improvements: Develop new model parameterisations for missing processes and improve current model limitations. Identify biases in diurnal, seasonal and interannual performance and diagnose for improvement. The different models listed above have their own strengths and weaknesses that we will evaluate and use to develop the best land surface models for savanna and seasonally water-limited ecosystems generally.
    5. Past savanna variability: We will utilise employ the improved models to assess the past variability of tree/grass population dynamics and concomitant variability of carbon and water fluxes due to climate variability during the 20th Century. In particular, we will focus on the effects of historical rainfall variability as quantified by the use the Shannon index (I), standard deviation (SD), coefficient of variation (CV), and number of dry weeks (V). The models will be run at least for each of the seven NATT sites, and selected models will be applied to the entire tropical savanna region at 0.05° (~ 5 km) resolution, using BAWAP gridded meteorological data as used by the Australian Water Availability Project ( BAWAP includes daily precipitation, radiation, maximum temperature, minimum temperature and vapour pressure at 0.05o resolution from 1900 to present, derived by sophisticated spatial interpolation of weather stations, which accounts for topographic influence. For selected models, we will use a weather generator to produce hourly data from daily surfaces.
    6. Sensitivity to future change: To assess the possible future states of Australian savanna landscapes we will use output from the CSIRO climate model (Mk2) as input for the improved models. The CSIRO Mk2 model is a fully coupled atmosphere/ocean model that includes dynamic sea ice and soil-canopy models and has been shown to provide good agreement with historic Australian rainfall dynamics. The atmospheric model is spectral with rhomboidal truncation at wave number 21, and has nine vertical levels, while the ocean is a grid point model with 21 levels. Horizontal resolution is 5.6o longitude by 3.2o latitude. We will use several IPCC climate change scenarios (SRES) including IS92a "business-as-usual" scenario, the "best case," B2, and "worst case," A2.


    Real-time data from flux tower observations will be made available through our Monash University web server and will help foster inter-site collaborations. The project will be linked and promoted via the Ozflux and FLUXNET web sites and we will initiate/contribute to any collective OZFLUX meetings. Final data will be archived regularly through TERN. An educational web site will be developed for high school students and teachers that will include basic scientific principles and a project kit that utilises real time data. Model results will be made available through Monash e-Research to the Australian National Data Service and to the PALS. We will strengthen research interactions through inter-disciplinary studies and involvement in FLUX networks. The value of this ARC proposed project is that it allows for innovation, but it will also contribute to a network that will allow us to address larger questions and develop general principals of vegetation interaction with the atmosphere. We will attend international European/American Geophysical meetings to share results, innovative measurements and analysis methods, and develop inter-site coordination. We will produce key findings in prominent national and international journals that will cover aspects of controls over fluxes, annual carbon balance, model simulations and development and long-term observations and variability. We will target ERA A* and A ranked journals. We will communicate results in a timely manner through national and international conferences. We will hold a science workshop on "Combining Ecosystem Flux Measurements and modelling" in year two and invite colleagues involved in other flux-based research and other interested parties from university, government and industry to participate. The workshop will be held over three days in Darwin and include site visits along the NATT.

    Useful Links


    This project is funded from 2013 to 2016 from the Australian Research Council under project number DP130131566, with Prof Beringer funded through an ARC Future Fellowship FT110100602.

    The site is produced by the Climate Group, School of Geography and Environmental Science and proudly part of the Australian Flux Network (OzFlux) and the NCRIS Terrestrial Ecosystem Research Network (TERN).