Diurnal cycles of fossil fuel CO2: Comparison of model results with observations at Heidelberg and Schauinsland Felix Vogel 1

including work of:I. Levin 1, U. Karstens 2, C. Rdenbeck 2, M. Krol 3, S. Houweling 3, P. Peylin 4, P. Bousquet 4, C. Aulagnier 4, C. Geels 5, A. Vermeulen 61 Institut fr Umweltphysik, Universitt Heidelberg2 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-Yvette5 National Environmental Research Institute Roskilde6 Energy research Center of the Netherlands, Petten

5th CarboEurope-IP Integrated Project Meeting, Pozna 2007

Introduction

Methods- Calculating FFCO2 from observations

- Statistics

Comparison- Heidelberg- Schauinsland

Summery & OutlookOutline

Introduction

Methods- Calculating FFCO2 from observations

- Statistics

Comparison- Heidelberg- Schauinsland

Summery & OutlookOutline

Calculating FFCO2 from observationsweekly excess D 14C

weekly excess CO2continuous excess COcalculated continuous excess FFCO2weekly excess ratioFFCO2/CO X=weekly excess COweekly excess FFCO2[Levin et al. 2003 GRL Vol.30/23]

Mean diurnal cycle - Heidelberg 2002Central European TimeCalculating FFCO2 from observations

Introduction

Methods- Calculating FFCO2 from observations

- Statistics

Comparison- Heidelberg- Schauinsland

Summery & OutlookOutline

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

R < 0.2R = 0.63R = 1 Find maximal correlation coefficient Determine time shifts in model data Possibly validate diurnal cycle of the emission inventories No measure for variability! (Amplitude x 2)Methods - StatisticsTime shift = 2hTime shift = 1hTime shift = 0h

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

Motivation

Methods- Calculating FFCO2 from observations

- Statistics

Comparison- Heidelberg- Schauinsland

Summery & OutlookOutline

Results - HeidelbergExcess FFCO2 diurnal cycle Heidelberg Winter 2002unshifted

Results - HeidelbergExcess FFCO2 diurnal cycle Heidelberg Winter 20020.80.70.30.80.31.2FFCO2, corr = FFCO2,mod xRnmodRnmeasAmplitudeshifted

Results - HeidelbergExcess FFCO2 diurnal cycle Heidelberg Winter 20020.71.20.61.5Amplitudeshifted0.80.70.30.80.31.2

Results - HeidelbergExcess FFCO2 diurnal cycle Heidelberg Summer 20022.21.30.72.30.72.10.90.71.40.81.8shiftedAmplitude

Results - HeidelbergExcess FFCO2 diurnal cycle Heidelberg summer 2002

Introduction

Methods- Calculating FFCO2 from observations

- Statistics

Comparison- Heidelberg- Schauinsland

Summery & OutlookOutline

Results - SchauinslandExcess FFCO2 diurnal cycle - Schauinsland Winter 2002unshifted

Results - SchauinslandExcess FFCO2 diurnal cycle - Schauinsland Winter 2002shifted

Results - SchauinslandExcess FFCO2 diurnal cycle - Schauinsland Winter 2002

Motivation

Methods- Calculating FFCO2 from observations

- Statistics

Results- Schauinsland- Heidelberg

Summery & OutlookOutline

Differences between IER and EDGAR are not significantbut IER seems to perform better in Heidelberg

Large spread in models

Ensemble analysis shows that the significant shifts in the diurnal cycle are not likely due to the inventories

Winter better than summer- less variable boundary layer height in winter- vertical mixing in summer is still a problem

Summery

Analysis:- Extensive variability analysis- Running R on continuous recordsMethods:- Studies on CO diurnal cycle- Include 14Cbio- Further studies on the radon source- Measurements at more representative sites

Outlook