Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

Changes in water and carbon in Australian vegetation in response to climate change

Liu, Ning (2017) Changes in water and carbon in Australian vegetation in response to climate change. PhD thesis, Murdoch University.

PDF - Whole Thesis
Download (12MB) | Preview


Australia has experienced pronounced climate change since 1950, especially in forested areas where a reducing trend in annual precipitation has occurred. However, the interaction between forests and water at multiple scales, in different geographical locations, under different management regimes and in different forest types with diverse species is not fully understood. Therefore, some interactions between forests and hydrological variables, and in particular whether the changes are mediated by management or climate, remain controversial. This thesis investigates the responses of Australia’s terrestrial ecosystems to both historical and projected climate change using remote sensing data and ecohydrological models. The thesis is structured in seven chapters, and contains five research chapters.

Vegetation dynamics and sensitivity to precipitation change on the Australian continent for the past long drought period (2002-2010) are explored in Chapter 2 using multi-resource vegetation indices (VIs; normalized difference vegetation index (NDVI) and leaf area index (LAI)) and gridded climate data. During drought, precipitation and VIs declined across 90% and 80% of the whole continent, respectively, compared to the baseline period of 2000-2001. The most dramatic declines in VIs occurred in open shrublands near the centre of Australia and in southwestern Australia coinciding with significant reductions in precipitation and soil moisture. Overall, a strong relationship between water (precipitation and soil moisture) and VIs was detected in places where the decline in precipitation was severe. For five major vegetation types, cropland showed the highest sensitivity to water change, followed by grassland and woody savanna. Open shrublands showed moderate sensitivity to water change, while evergreen broadleaf forests only showed a slight sensitivity to soil moisture change. Although there was no consistent significant relationship between precipitation and VIs of evergreen broadleaf forests, forests in southeastern Australia, where precipitation had declined since 1997, appear to have become more sensitive to precipitation change than in southwestern Australia.

The attribution of impacts from climate change and vegetation on streamflow change at the catchment scale for southwestern Australia are described in Chapter 3. This region is characterized by intensive warming and drying since 1970. Along with these significant climate changes, dramatic declines in streamflow have occurred across the region. Here, 79 catchments were analyzed using the Mann-Kendall trend test, Pettitt’s change point test, and the theoretical framework of the Budyko curve to study changes in the rainfall-runoff relationship, and effects of climate and vegetation change on streamflow. A declining trend and relatively consistent change point (2000) of streamflow were found in most catchments, with over 40 catchments showing significant declines (p < 0.05, -20% to -80%) between the two periods of 1982-2000 and 2001-2011. Most of the catchments have been shifting towards a more water-limited climate condition since 2000. Although streamflow is strongly related to precipitation for the period of 1982 to 2011, change of vegetation (land cover/use change and growth of vegetation) dominated the decrease in streamflow in about two-thirds of catchments. The contributions of precipitation, temperature and vegetation to streamflow change for each catchment varied with different catchment characters and climate conditions.

In Chapter 4, the magnitude and trend of water use efficiency (WUE) of forest ecosystems in Australia, and their response to drought from 1982 to 2014, were analyzed using a modified version of the Community Atmosphere Biosphere Land Exchange (CABLE) land surface model in the BIOS2 modelling environment. Instead of solely relying on the ratio of gross primary productivity (GPP) to evapotranspiration (ET) as WUE (GPP/ET), the ratio of net primary productivity (NPP) to Transpiration (ETr) (NPP/ETr) was also adopted to more comprehensively understand the response of vegetation to drought. For the study period, national average annual forest WUE was 1.39 ± 0.80 g C kg−1 H2O for GPP/ET and 1.48 ± 0.28 g C kg−1 H2O for NPP/ETr. The WUE increased in the entire study area during this period (with a rate of 0.003 g C kg−1 H2O yr-1 for GPP/ET; p < 0.005 and a rate of 0.0035 g C kg−1 H2O yr-1 for NPP/ETr; p < 0.01), whereas different trends were detected in different biomes. A significantly increasing trend of annual WUE was only found in woodland areas due to higher magnitudes of increases in GPP and NPP than ET and ETr. The exception was in eucalyptus open forest area where ET and ETr decreased more than reductions in GPP and NPP. The response of WUE to drought was further analyzed using 1-48 month scales standardised precipitation-evapotranspiration index (SPEI). More severe (SPEI < -1) and frequent droughts (over ca. 8 years) occurred in the north than in the southwest and southeast of Australia since 1982. The response of WUE to drought varied significantly regionally and across forest types. The response of WUE to drought varied significantly regionally and across forest types, due to the different responses of carbon sequestration and water consumption to drought. The cumulative lagged effect of drought on monthly WUE derived from NPP/ETr was consistent and relatively short and stable between biomes (< 4 months), but notably varied for WUE based on GPP/ET, with a long time lag (mean of 16 months).

As Chapters 2-4 confirmed that climate change has been playing an important role in the water yield and vegetation dynamics in Australia, the response of water yield and carbon sequestration to projected future climate change scenarios were integrated using the Water Supply Stress Index and Carbon model (WaSSI-C) ecohydrology model in Chapter 5. This model was calibrated with the latest water and carbon observations from the OzFlux network. The performance of the WaSSI-C model was assessed with measures of Q from 222 Hydrologic Reference Stations (HRSs) in Australia. Across the 222 HRSs, the WaSSI-C model generally captured the spatial variability of mean annual and monthly Q as evaluated by the Correlation Coefficient (R2 = 0.1-1.0), Nash-Sutcliffe Efficiency (NSE = -0.4-0.97), and normalized Root Mean Squared Error by Q (RMSE/Q = 0.01-2.2). Then 19 Global Climate Models (GCMs) from the Coupled Model Intercomparison Project phase 5 (CMIP5), across all Representative Concentration Pathways (RCPs) (RCP2.6, RCP4.5, RCP6.0 and RCP8.5), were used to investigate the potential impacts of climate change on water and carbon fluxes. Compared with the baseline period of 1995-2015 across the 222 HRSs, the temperature was projected to rise by an average of 0.56 to 2.49 ˚C by 2080, while annual precipitation was projected to vary significantly. All RCPs demonstrated a similar spatial pattern of change of projected Q and GPP by 2080, however, the magnitude varied widely among the 19 GCMs. Overall, future climate change may result in a significant reduction in Q but may be accompanied by an increase in ecosystem productivity. Mean annual Q was projected to decrease by 5 - 211 mm yr-1 (34% - 99%) by 2080, with over 90% of the watersheds declining. On the contrary, GPP was projected to increase by 17 - 255 g C m-2 yr-1 (2% - 17%) by 2080 in comparison with 1995-2015 in southeastern Australia. A significant limitation of WaSSI-C model is that it only runs serially. High resolution simulations at the continental scale are therefore not only computationally expensive but also present a run-time memory burden.

In Chapter 6, using distributed (Message Passing Interface, MPI) and shared (Open Multi-Processing, OpenMP) memory parallelism techniques, the model was parallelized (and renamed as dWaSSI-C), and this approach was very effective in reducing the computing run-time and memory use. By using the parallelized model, several experiments were carried out to simulate water and carbon fluxes over the Australian continent to test the sensitivity of the model to input data-sets of different resolutions, as well as the sensitivity of the model to its WUE parameter for different vegetation types. These simulations were completed within minutes using dWaSSI-C, and this would not have been possible with the serial version. Results show that the model is able to simulate the seasonal cycle of GPP reasonably well when compared to observations at 4 eddy flux sites in Australia. The sensitivity analysis showed that simulated GPP was more sensitive to WUE during the Australian summer as compared to winter, and woody savannas and grasslands showed higher sensitivity than evergreen broadleaf forests and shrublands. With the parallelized dWaSSI-C model, it will now be much easier and faster to conduct continental scale analyses of the impacts of climate change and land cover change on water and carbon.

Overall, vegetation and water of Australian ecosystems have become very sensitive to climate change after a considerable decline in streamflow. Australian ecosystems, especially in temperate Australia, are projected to experience warmer and drier climate conditions with increasing drought risk. However, the prediction from different models varied significantly due to the uncertainty of each climate model. The impacts of different forest management scenarios should be studied to find the best land use pattern under the changing climate. Forest management methods, such as thinning and reforestation, may be conducted to mitigate the impacts of drought on water yield and carbon sequestration in the future.

Item Type: Thesis (PhD)
Murdoch Affiliation(s): School of Veterinary and Life Sciences
United Nations SDGs: Goal 13: Climate Action
Supervisor(s): Harper, Richard, Dell, Bernard and Liu, Shirong
Item Control Page Item Control Page


Downloads per month over past year