HR: 1340h
AN: A13A-0898    [Abstracts]
TI: Prospects for Problems Associated with Integrative and Inter-comparative Analysis of Eddy Flux Data Sets
AU: * Kobayashi, Y
EM: yosikazu@niaes.affrc.go.jp
AF: National Institute for Agro-Environmental Sciences, Kannondai 3-1-3, Tsukuba, 305-8604 Japan
AU: Miyata, A
EM: amiyat@niaes.affrc.go.jp
AF: National Institute for Agro-Environmental Sciences, Kannondai 3-1-3, Tsukuba, 305-8604 Japan
AU: Nagai, H
EM: iagan@niaes.affrc.go.jp
AF: National Institute for Agro-Environmental Sciences, Kannondai 3-1-3, Tsukuba, 305-8604 Japan
AU: Mano, M
EM: mmano@niaes.affrc.go.jp
AF: National Institute for Agro-Environmental Sciences, Kannondai 3-1-3, Tsukuba, 305-8604 Japan
AU: Yamamoto, S
EM: yamas@cc.okayama-u.ac.jp
AF: Graduate School of Environmental Science, Okayama University, Tsushimanaka 3-1-1, Okayama, 700-8530 Japan
AB: In last decade, numerous long-term eddy flux measurements have been conducted worldwide to assess annual/seasonal energy, water and carbon exchanges between terrestrial ecosystem and the atmosphere. And FLUXNET communities now seem to come into a next phase with the objectives: integration of flux data observed at various ecosystems and/or inter-sites comparative studies. For example, a big research project "S-1" is ongoing in Japan and other eastern Asian region to set up terrestrial carbon management of Asia in the 21st century. One of the highlights of S-1 project is to provide a carbon budget map of all over Asia based on integrated and inter-compared eddy flux data collected at 15 sites of S-1 membership. FLUXNET communities including S-1 project have recognized that integration and inter-comparison of eddy flux data are the key issues to understand aspects of energy, water and carbon budgets at regional scale. However, the issues have difficulties to be settled because each flux site applies own data processing methods and gap-filling methods with site-specified classification and threshold values. In order to conduct appropriate integrative and inter-comparative analysis for eddy flux data effectively, we made it clear that how the differences in the data processing method affect the obtained flux values and searched for suitable and common gap-filling methodology. The differences in the data processing methods affect the obtained flux data in the present study was discussed based on a comparative experiment in S-1 project. We prepared one-month common test data sets, which consisted of 10 Hz eddy covariance raw data and related half-hourly meteorological data obtained at a larch forest site and a paddy site, in the comparative experiment. The 15 sites of S-1 memberships processed the test data by using their own processing methods. The results indicated that combined influences of coordinate rotation, detrending and frequency response correction brought about up to 10% of flux discrepancy, and that the forest sites were more sensitive to differences in the data processing methods than the non-forest sites. Multiple imputation method (MI), one of the statistical operations for analyzing incomplete multivariate data set, is likely to be an easy-to-use and objective gap-filling method to account for missing eddy flux data. We also discussed validity of application of MI to fill missing flux data by comparing a gap-filled complete eddy flux data set obtained by MI with that by nonlinear regression method and look-up table method. It was revealed that, with suitable separation of the periods to be filled and proper selection of reference variables, MI has potential to be applied commonly to gap-filling missing flux data, and that MI can be a useful tool for FLUXNET communities to make inter-site comparison of long-term flux data.
DE: 3270 Time series analysis (1872, 4277, 4475)
DE: 3275 Uncertainty quantification (1873)
DE: 3294 Instruments and techniques
DE: 3307 Boundary layer processes
DE: 3322 Land/atmosphere interactions (1218, 1631, 1843)
SC: Atmospheric Sciences [A]
MN: Fall Meeting 2005