Figure: Geographical distribution of observation sites in Australia (click on the small figure to view a larger version)

Data Citation

Cite this data set as follows:

Barrett, D. J. 2001. NPP Multi-Biome: VAST Calibration Data, 1965-1998. Data set. Available on-line [http://www.daac.ornl.gov] from Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, U.S.A.

Description

The VAST (Vegetation and Soil-carbon Transfer) dataset contains observations of net primary productivity (NPP), biomass, litter mass, surface horizon soil carbon concentration (i.e., mass fraction) and bulk density, and soil carbon and bulk density measurements at various depths. The data consist of 33 estimates of above-ground NPP based on cut grass swards and visual assessment of growth, 150 measurements of litterfall (leaf and fine twig), 76 measurements of above-ground biomass (phytomass), 91 determinations of fine litter mass, 341 measurements of soil carbon concentration in surface layers (to 15 cm depth), and 50 determinations of soil bulk density (to 15 cm depth). All these data were derived from 174 original literature references describing study sites throughout Australia.

The carbon content (mass fraction) of dry plant matter was assumed to be 0.42 for herbaceous vegetation and 0.45 for woody vegetation. Each observation is associated with a latitude and longitude, values of climate, soils and vegetation parameters for the 0.05-degree grid cell within which the observation lies (derived from various Australian national datasets) and the literature reference from which the observation was obtained.

Quality control on measurements was imposed by ensuring that data had undergone scientific peer review prior to publication. Data were assumed to correspond to "minimally disturbed" vegetation according to the following criterion: the author's description of overstory vegetation structure and species composition was the same as the Australian Survey and Land Information Group (AUSLIG) historical data set of 1788 vegetation classification, i.e. the growth form of vegetation remained constant in the period between the historical data set and the date of the study. Georeferencing of point observations utilised the latitude and longitude of each study site provided by authors or, where absent, obtained from an identifiable topographic feature described within the study and located on 1:1,000,000 and 1:2,500,000 scale AUSLIG maps. For those studies where authors presented an age sequence of biomass or littermass, the oldest sites were chosen on the assumption that these measurements were more likely to be near steady state. Measurements were averaged for those studies where multiple sites were described by authors but listed under a single latitude and longitude within one 0.05-degree grid cell. As a consequence a small amount of information on spatial variability at local scales was lost. In cases where multiple years of NPP or litterfall measurements were available at a single site, each year's data was included because year-to-year deviations of NPP measurements from the long term average are an important component of total variability in NPP.

Soil carbon concentrations (mass fractions) range from 0.1 to 720 mgC g-1 (0.01% to 72%), the highest value being for a moss field site in Tasmania. Average soil carbon concentration (to 15 cm depth) is about 33 mgC g-1 (3.3%). Most arid soils in Australia contain less than 10 mgC g-1 (<1%).

The data set and the VAST model were compiled and developed to address the limited information available on the magnitude, variability and spatial distribution of mean residence time of carbon in the terrestrial biosphere. The geographical distribution of continental-scale variability in steady state mean residence time of biosphere carbon stocks was determined using statistical models developed from spatially distributed data sets of point observations and a one-dimensional carbon cycle model. The steady state carbon cycle model (the VAST model) was "inverted" and solved for rate constants governing the turnover of live biomass, litter, soil carbon and biosphere carbon for each and every 0.25 x 0.25 degree terrestrial grid cell over the Australian continent. Statistical models of maximum potential net primary productivity (NPP) and carbon stocks were developed from spatially distributed, geo-referenced point observations. Confidence intervals of inferences from the statistical models were then used in combination with Monte Carlo techniques to generate the prediction intervals of these turnover constants.

Area-weighted Australian-continental sums of the median value (and range of prediction intervals, ±1.0 standard deviation) for inferences of NPP, live biomass, litter, and total soil carbon to 1.0 m depth were, respectively, 0.96 GtC yr-1 (0.70 - 1.34), 8.4 GtC (4.1 - 29.4), 4.4 GtC (3.7 - 5.6) and 26.9 GtC (23.8 - 30.5). The reciprocal median residence times and ranges were, respectively, 53.4 years (35.0 - 76.4), 9.2 years (4.2 - 22.7), 6.8 years (4.8 - 8.7) and 66.0 years (47.1 - 86.1). Sources of uncertainty contributing to the magnitude of the prediction intervals were grouped into three categories: (1) errors originating from national climatic, edaphic and vegetation data sets and from insufficient data to quantify covariance terms between NPP and carbon stocks; (2) biases that arise from insufficient knowledge of the allocation of photosynthate to below-ground tissues, from insufficient data to determine the proportion of decomposing litter carbon that is oxidised and returned to the atmosphere, and from the small sample number with which to develop statistical models; and (3) natural heterogeneity that arises from variation in microenvironment and from interspecific differences in plant and soil microbe responses to this variation.

An additional source of uncertainty in reconciling continental and global soil carbon stocks, soil carbon fluxes and modeled soil carbon residence time is specifying the soil depth beyond which heterotrophic respiration contributes an "insignificant" proportion to soil surface CO2 efflux to the atmosphere. It was found that, in order to reconcile soil carbon residence time for the Australian continent with literature values of soil carbon turnover, a decrease in the "effective" lower bound of soil CO2 production was necessary, from 30-50 cm at temperate latitudes to around 30 cm in the tropics.

The distribution of observations in the geographic domain displays a bias. Comparison of the distribution of samples across vegetation classes against the continental area-distribution of these classes showed that a predominance of sites occurred in Tall Closed Forests (vegetation classification = T) and Medium Forests (M) even though these forests occupy <1% and 23% of continental land area, respectively. Additionally, NPP and phytomass measurements for Low Forests (L) and Shrublands (S, Z) were under-represented given that these vegetation classes occupy the majority land area of the continent (85%). Soil carbon concentration is better distributed across vegetation classes due to incorporation of data from the CSIRO National Soil Database.

The distribution of observations in the climatic domain is more representative of the continental range of mean annual temperature and mean annual rainfall than the distribution of these observations in the geographic domain. However, an there is an absence of observations for NPP, phytomass, and littermass from regions of mean annual rainfall >1200 mm and mean annual temperature <15 C. This region corresponds primarily to the remote western Tasmanian forests where no observations were available. Additionally, relatively few observations were available for regions having mean annual rainfall > 1000 mm and mean annual temperature > 20 C which corresponds to tropical rainforest and savannas at latitudes < 15 S.

References

Barrett, D. J. 2001. Steady state net primary productivity, carbon stocks and mean residence time of carbon in the Australian terrestrial biosphere. Submitted to Global Biogeochemical Cycles.

Barrett, D.J. 1999. Steady state carbon mean residence time in the Australian terrestrial biosphere. EOS (Supplement), Transactions of the American Geophysical Union 80(46): F51.

Contact Information

Contact: Dr. D. J. Barrett
CSIRO Plant Industry
GPO Box 1600
Canberra ACT 2601
AUSTRALIA

Tel. +61 (2) 6246-5558
Fax: +61 (2) 6246-5800
E-mail: damian.barrett@pi.csiro.au