Flux towers measure the net exchange of carbon dioxide (NEE) between the land surface and the boundary-layer of the atmosphere but Gross Primary Productivity (GPP) and Ecosystem Respiration (ER), the components of NEE, are often derived from flux tower data. GPP and ER can then be used by modellers and remote sensing practitioners to parameterise and validate their models and retrieval systems.
Recent work using TERN data to explore the temperature response of GPP showed that there was a high level of noise in GPP at some TERN OzFlux sites, most notably at TERN’s Cumberland Plain Woodland SuperSite. Additionally, two methods of estimating ER (neural network and Lloyd-Taylor) gave opposite temperature responses. A further puzzle, possibly related, was the poor performance of friction velocity (u*) threshold detection techniques at this site. Examination of spectra, cospectra and the quality control of the flux data led to a significant reduction in the noise of the CO ₂ flux data and a consequent improvement in the processed results.
This talk uses the experiences at the Cumberland Plain Woodland SuperSite to explore methods of improving the quality of TERN OzFlux data, indicators of data quality and provides insight into the uncertainty inherent in our partitioning techniques. Reducing this uncertainty has contributed to better understanding of the sensitivity of Australian ecosystems to further climate warming.
Dr Peter Isaac
Prof Elise Pendall