Project B.2:  Uncertainty and Variability in the Carbon Cycle

• Team Members
• Objectives
• Strategy
• Relevance
• Outputs
• Outcomes
• Photo Gallery

Research objectives

• To develop an improved theoretical understanding of the interactions between the C cycle and climate, nutrients, disturbance and land cover change at a range of time-space scales from seasons to centuries and catchments to continent.
• To develop novel parameter estimation methods for incorporating multiple and disparate datasets of observations as constraints on models of the terrestrial C cycle.
• To develop quantitative methodologies for reducing uncertainty in net C exchange rates between land surfaces and the atmosphere arising from (1) seasonal to multi-year variability in climate, (2) natural disturbance and (3) land use change.
• To develop a case-study of the application of these methods to the National Carbon Accounting System of the Australian Greenhouse Office.
• To extend the sampling strategies developed and validated by an original CRC project (3.1) for coarse-textured soils to other soil types in Australia.
• To develop robust empirical relationships between soil organic carbon (SOC) density and its d ¹³C-isotopic composition and their relationships with climate to provide input for model-derived estimates of SOC carbon stocks at the regional/continental scale.
• To monitor developments in remote sensing and develop methods of incorporating remote sensing products into large scale C cycle models

Strategy

Resolving past, present and future net fluxes of terrestrial carbon at large scales is difficult. There exist large uncertainties arising from 1) inadequate knowledge of system processes and the relationships between processes at different spatial and temporal scales, 2) insufficient data, 3) natural heterogeneity, and 4) stochastic forcing by weather, climate and anthropogenic events.

One of the key objectives of this project is to combine an improved theoretical understanding of key processes governing the terrestrial C cycle with advanced computational techniques (specifically Numerical Programming, Genetic Algorithms and Bootstrap Methods) that allow integration of data from a variety of different sources to estimate poorly known parameters. This approach has four benefits

• it ensures consistency between model structure and observations thereby reducing the potential for bias in model predictions;
• the information content of data sources is quantified through parameter statistics;
• it identifies where major data limitations interact with model sensitivities; and
• it enables a quantitative measure of the range in model predictions given data uncertainties.

Most biological and ecological research is conducted at local scales, but for carbon accounting purposes, the results often need to be scaled to larger areas. In Project B.2, we are formulating a new approach to this long-standing problem. The essence of our approach is to recognise explicitly local-scale variability in the basic description of the system.

For example, the photosynthetic properties of a vegetation canopy are represented by leaf-scale averages and the variability about the mean. Following that, changes in the flows and stocks of carbon depend on both the average and the variability. Without consideration of both the average and the variability, large scale estimates developed from process understanding at fine scales may be biased.

Soil C stocks and turnover are key uncertainties in large scale C cycle modelling. Another key objective of Project B.2 is to use the stratified sampling methodology originally developed in a previous CRC project (3.1) to sample soils having a range of textures (clay-poor to clay-rich) in selected regions that cover the climate conditions of Australia (four regions encompassing wet tropical, dry tropical, arid and temperate).

The current project will build on the results of this earlier project by extending the datasets already produced to the full range of soil textures. It is to be closely linked to the modelling effort in Project B.2. With the extension of the stratified sampling approach to the full range of soil textures in Australia, a complete empirical description of SOC stocks and the d ¹³C-isotopic composition of SOC in Australian soils will be possible.

Relevance

The outcomes of this research have direct relevance to carbon accounting systems at project, national and international levels. Transparent and verifiable measurement methodologies use direct measurement of C stocks at local spatial scales and quantifying uncertainties at this scale is a sampling problem. On moving from local to landscape scales, unbiased scaling of process models is required. At large spatial scales, limited information on system processes and poor data availability make the task of measuring and accounting for C stocks much more difficult.

A requirement of the verification process is quantitative measures of the uncertainty at all these scales to judge confidence in estimated sizes of C sources and sinks. Within that framework, uncertainty also exists in projections of C-cycle dynamics. For example, multi-year climate fluctuations coincident with Commitment Periods of international climate change treaties could have dramatic effects on sink activity over large regions, particularly of soil decomposition processes where most uncertainty in the terrestrial C-cycle currently resides.

Alternatively, changed disturbance regimes (fire, wind or land use change) may act to reinforce or offset C sequestration programs. Project B.2 deals explicitly with quantifying and reducing these uncertainties in measuring and modelling the terrestrial C cycle at different scales.

Outputs

• Development of a theoretical framework for the generic description of the state of plant and soil systems, and the C-fluxes to and from those systems as constrained by terrestrial biogeochemistry
• Use of this theoretical framework as the basis for scaling C-fluxes in space and time.
• Development of novel routines for estimating parameters from multiple data sources.
• Establishment of a case-study application of new computational methods for the National Carbon Accounting System
• Incorporation of insights and developments in process understanding and data assimilation from other CRC projects into Project B.2.
• Generation of variance propagation algorithms (First Order and Monte Carlo methods) for calculating uncertainty in predicted C-fluxes.
• Deployment of Numerical Programming, Genetic Algorithms and Bootstrap Methods of parameter estimation.
• Generation of quantitative ranges of predictions of C-emissions arising from both land cover and land use change.
• Estimation of the likelihood of net terrestrial C-exchange rates arising from episodic climate events and examine the change in C-fluxes expected under the CSIRO Climate Change Projections.
• Extension of the stratified sampling methodology validated in the original Project 3.1 to other soil types in Australia to provide data on soil carbon densities, stable-isotope composition and turnover time in terms of prime determinants – temperature, rainfall, soil texture and tree-grass distribution.
• Development of predictive empirical relationships for the stable-carbon isotope composition of soil organic carbon in coarse-textured soils as a function of climate which can predict both the isotopic composition of SOC and of overlying vegetation for use in budget, turnover and tracer studies.
• Provision of a consistently sampled and analysed dataset for testing and validating continental scale models of soil carbon stocks and fluxes and isotope distributions at a range of scales.
• Estimation of SOC variability that is proposed to assist in the development of quantitative statistical methodologies to calculate uncertainties in continental net C emissions from vegetation clearing and agricultural activity.
• Development of a rotary soil-coring device suitable for the rapid collection of soil cores (with bulk density determination) from hard, finer-textured soils that cannot be sampled by hand coring devices.
• Awareness of emerging developments in remote sensing technology and their application in measurement and modelling of carbon stocks and fluxes.

Outcomes

• A stronger basis for national inventory carbon trading by increasing our knowledge of the spatial and, in particular, the temporal variability and uncertainties in the emission and uptake of greenhouse gases.