Large-scale response of ecosystem production to GRACE derived dynamic water storage

Author: Robbie Andrew

Andrew, Robbie, 2017 Large-scale response of ecosystem production to GRACE derived dynamic water storage, Flinders University, School of the Environment

This electronic version is made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.

Abstract

Terrestrial ecosystems play a large role in the global carbon cycle as one of the two natural carbon sinks on Earth, along with oceans. In comparison to the ocean sink, the terrestrial carbon sink is much more variable, and often driven by temporal variations in hydro-meteorological conditions. Thus, it is important to monitor, understand, and model the hydrologically driven vegetation dynamics as a premise for improving our understanding of the global carbon cycle. Terrestrial primary production from vegetation is driven by water and in some parts of the globe is almost entirely dependent on water availability. Thus there is a clear link between terrestrial water availability and vegetation dynamics. Our ability to estimate water storage over the globe has increased over recent decades, with the launch of remote sensing tools such as the Gravity Recovery and Climate Experiment (GRACE). GRACE has proven to be an extremely useful satellite mission for hydrological studies. The body of this research encompasses developments in our understanding of the way vegetation responds to water availability, and expands the use of GRACE data for hydrological estimations. GRACE data is analysed in an innovative way such that more information can be extracted from it than ever before. The aim of this PhD is to improve our understanding of relationships between terrestrial water and vegetation on a continental and global scale. This is in conjunction with the aim of extending the potential application of GRACE by using it in innovative and previously unused ways. Specifically, this work investigates: (1) the use of wavelet decomposition of GRACE data to comprehensively ‘split’ GRACE total water storage (TWS) into shallow and deep subsurface components; (2) the use of wavelet decomposition of GRACE data in conjunction with the Normalised Difference Vegetation Index (NDVI) to examine the temporal variability and moisture dependence of vegetation cover across Australia; and (3) the use of GRACE TWS amplitude to represent dynamic water storage and to examine how it is a key driver of biomass production in terrestrial water limited ecosystems globally. A potential limitation of GRACE is that the TWS storage it estimates have no vertical segregation. In the first component of this research, a new method was developed to create estimations of deep and shallow subsurface water storage from GRACE TWS estimations. To achieve this, a wavelet decomposition is used to ‘split’ GRACE into components of different temporal frequencies, hypothesising that various vertical water storage components have different temporal frequencies. For example, deep groundwater has a low frequency, slow moving signal, while the storage of soil moisture near the surface is more dynamic. The Australian Water Resources Assessment (AWRA) model is used as a reference for the decompositions of total water storage across Australia. A stepwise regression compares the wavelet decomposed components of GRACE TWS to the AWRA model. Results show a clear improvement in using decomposed GRACE data instead of raw GRACE data when compared against the outputs from the AWRA model. GRACE TWS has recently been used to investigate moisture dependence of vegetation cover. However, part of GRACE TWS is beyond the reach of the root zone and thus irrelevant to vegetation function. In the second part of this research, this issue is addressed by using shallower water storage signals to examine temporal variability of NDVI. Wavelet decomposed components of GRACE TWS anomalies are analysed against NDVI anomalies in a stepwise regression. The results show that combinations of different frequencies of decomposed GRACE TWS data explain NDVI temporal variations better than raw GRACE TWS alone. Different types of vegetation show distinct differences in how they respond to the changes in water storage which are generally consistent with our physical understanding. GRACE TWS of each cell is referenced to (offset by) a prescribed mean of itself, leading to difficulties to compare TWS across cells or use TWS to investigate spatial variability of vegetation cover. In the third part of this research, the hypothesis is posed that terrestrial ecosystem production is driven by effective water fluxes going through the system at a pace relevant to vegetation functioning. Hence, the relationship between the annual amplitude of GRACE TWS and gross primary productivity is examined. The GRACE amplitude represents the dynamic water storage in a year. The results show that the dynamic water storage is a significant driver of biomass production. Strong correlations between gross primary production and annual amplitudes of total water storage exist in water limited ecosystems globally. The use of total water storage amplitude provides a novel approach linking the dependence of vegetation production to water that is available and actually used by ecosystems, and extend the applicability of GRACE data in explaining large-scale spatial variability of vegetation cover. This PhD research presents advances in our understanding of largescale water-vegetation relations which are of global significance. The innovative analysis of GRACE data as developed and, tested and applied in this research helps to shape further scientific developments in the application of such data.

Keywords: Hydrology, GRACE, Water storage, Wavelet, Regression, Biomass, Primary productivity, NDVI
Subject:

Thesis type: Doctor of Philosophy
Completed: 2017
School: School of the Environment
Supervisor: Huade Guan