5 th Carbo EuropeIP Integrated Project Meeting Pozna

5 th Carbo. Europe-IP Integrated Project Meeting, Poznań 2007 Diurnal cycles of fossil fuel CO 2: Comparison of model results with observations at Heidelberg and Schauinsland Felix Vogel 1 including work of: I. Levin U. Karstens C. Rödenbeck 2, M. Krol 3, S. Houweling 3, P. Peylin 4, P. Bousquet 4, C. Aulagnier 4, C. Geels 5, A. Vermeulen 6 1, 2, Institut für Umweltphysik, Universität Heidelberg 2 Max-Planck-Institute for Biogeochemistry, Jena 3 National Institute for Space Research Utrecht 4 Laboratoire des Sciences du Climat et de l'Environnement Gif-sur-Yvette 5 National Environmental Research Institute Roskilde 6 Energy research Center of the Netherlands, Petten 1

Outline Ø Introduction Ø Methods - Calculating FFCO 2 from observations - Statistics Ø Comparison - Heidelberg - Schauinsland Ø Summery & Outlook

Calculating FFCO 2 from observations [Levin et al. 2003 GRL Vol. 30/23] weekly excess D 14 C weekly excess FFCO 2 weekly excess CO continuous excess CO X weekly excess ratio FFCO 2/CO = calculated continuous excess FFCO 2

Calculating FFCO 2 from observations Mean diurnal cycle - Heidelberg 2002 Central European Time

Outline Ø Introduction Ø Methods - Calculating FFCO 2 from observations - Statistics Ø Comparison - Heidelberg - Schauinsland Ø Summery & Outlook

Methods - Statistics Mean diurnal cycle - Heidelberg summer 2002 Wrong phasing of the diurnal cycle significantly decreases the correlation coefficient !

Methods - Statistics Time shift = 0 h R² = 1 Time shift = 1 h Time shift = 2 h R² = 0. 63 R² < 0. 2 Ø Find maximal correlation coefficient Ø Determine time shifts in model data Ø Possibly validate diurnal cycle of the emission inventories Ø No measure for variability! (Amplitude x 2)

Methods - Statistics Why diurnal analysis and not pure comparison of the time series? Advantages: Ø Less sensitive to pollution events Ø Reduction of uncertainties Ø Less computational effort Implicit assumptions: Ø Similar emission statistics for each season Ø Diurnal cycle is significant compared to the noise

Outline Ø Motivation Ø Methods - Calculating FFCO 2 from observations - Statistics Ø Comparison - Heidelberg - Schauinsland Ø Summery & Outlook

Results - Heidelberg unshifted Excess FFCO 2 diurnal cycle – Heidelberg Winter 2002

Results - Heidelberg Excess FFCO 2 diurnal cycle – Heidelberg Winter 2002 FFCO 2, corr = FFCO 2, mod x 0. 8 0. 7 Rnmeas Rnmod 0. 3 0. 8 0. 3 shifted 1. 2 Amplitude

Results - Heidelberg Excess FFCO 2 diurnal cycle – Heidelberg Winter 2002 0. 3 0. 8 0. 7 1. 2 0. 6 Amplitude 1. 2 shifted 1. 5

Results - Heidelberg Excess FFCO 2 diurnal cycle – Heidelberg Summer 2002 Amplitude 1. 4 shifted 0. 9 2. 3 0. 8 0. 7 1. 8 0. 7 2. 2 1. 3 2. 1

Results - Heidelberg Excess FFCO 2 diurnal cycle – Heidelberg summer 2002

Outline Ø Introduction Ø Methods - Calculating FFCO 2 from observations - Statistics Ø Comparison - Heidelberg - Schauinsland Ø Summery & Outlook

Results - Schauinsland unshifted Excess FFCO 2 diurnal cycle - Schauinsland Winter 2002

Results - Schauinsland shifted Excess FFCO 2 diurnal cycle - Schauinsland Winter 2002

Results - Schauinsland Excess FFCO 2 diurnal cycle - Schauinsland Winter 2002

Outline Ø Motivation Ø Methods - Calculating FFCO 2 from observations - Statistics Ø Results - Schauinsland - Heidelberg Ø Summery & Outlook

Summery Ø Differences between IER and EDGAR are not significant but IER seems to perform better in Heidelberg Ø Large spread in models Ø Ensemble analysis shows that the significant shifts in the diurnal cycle are not likely due to the inventories Ø Winter better than summer - less variable boundary layer height in winter - vertical mixing in summer is still a problem

Outlook Ø Analysis: - Extensive variability analysis - Running R² on continuous records Ø Methods: - Studies on CO diurnal cycle - Include Δ 14 Cbio - Further studies on the radon source - Measurements at more representative sites