6B.7 EXPERIMENAL VALIDAION OF HE WEBB CORRECION FOR CO2 FLUX WIH AN OPEN-PAH GAS ANALYZER Osamu sukamoto* an Fumiyoshi Kono Okayama University, Okayama, Japan 1. INRODUCION urbulent flux by the ey covariance technique is the only irect measurement over earth surfaces. CO2 an water vapor are the only trace gases that can measure turbulent flux with the open-path gas analyzer. he open-path CO2/H2O gas analyzer oes not measure non-imensional CO2 an water vapor concentrations as mixing ratios but it measures CO2 an water vapor ensities. For this reason, the trace gas flux using this analyzer nees to be correcte for the mean vertical flow ue to air ensity fluctuation. Webb et al. (1980) suggeste that the flux ue to the mean vertical flow cannot be neglecte for trace gases such as water vapor an CO2. o evaluate the magnitue of the influence by the mean vertical flow, Webb et al. (1980) assume that the vertical flux of ry air shoul be zero for the mass conservation. Practically, sensible an latent heat fluxes evaluate by the ey covariance technique are use to calculate water vapor an CO2 fluxes by the mean vertical flow (Webb or WPL correction). he Webb correction is important over sea surface CO2 flux as it is much smaller than the vegetate surface. Ohtaki et al. (1989) suggeste that the Webb correction term of CO2 flux is larger than the raw CO2 flux an changes the transport irection of the CO2 flux over the coastal sea surface. An recently, Kono an sukamoto (2007) also confirme the importance of the Webb correction over the open ocean surface. he basic Webb correction theory has been iscusse well incluing the recent iscussions (e.g. Leuning,2007), but this correction is still poorly unerstoo from actual observe ata in various conitions. herefore, it is important to evaluate the original Webb correction scheme using the experimental ata.. For this purpose, we measure the turbulent flux over an asphalt surface (large parking lot), where CO2 an water vapor fluxes were consiere almost zero an sensible heat flux was significant. ------------------------------------------------------------------- *Corresponing author ; Osamu sukamoto, Okayama University, Okayama, Japan e-mail : tsuka@cc.okayama-u.ac.jp 2. EXPERIMENAL MEHOD he observation was mae in the large parking lot on August 24, 2006. his site is locate at a coast in Okayama, Japan. his parking lot extens 110 m x 180 m an provie a uniform an ry asphalt surface. he fine weather continue an the surface was very ry. Sea breeze ominate uring this observation perio. Fluctuations of three-imensional win velocities an soun virtual temperature were measure using a sonic anemometer thermometer (KAIJO, DA-600-3V). he SA probe was irecte against the ominant sea breeze. CO2 an water vapor ensity fluctuations were measure using an infrare open-path CO2/H2O gas analyzer (LI- COR, LI-7500). Both instruments were mounte on a tripo at a 1.62 m height above the asphalt surface an separate horizontally by 0.25 m. he observation point was locate near the northwest ege of the parking lot to ensure the sufficient fetch from the sea breeze. he fetch to the southeast coast was more than 180 m. he analog output signals from these instruments were sample an igitize at the rate of 10 Hz by a ata acquisition system (LabVIEW, National Instruments) on a PC. 3. EDDY FLUX EVALUAION INCLUDING HE WEBB CORRECION H = he sensible heat flux (H) is given by C ρ w ' ' p p = C ρ ( w ' ' sv 0.514 sv w' q') (1) = sv ( 1 0.514q) (2) where, C p : specific heat of air at constant pressure (J kg -1 K -1 ) ρ : ensity of ry air (kg m -3 ) w : vertical win velocity (m s -1 ) : air temperature (K) sv : soun virtual temperature (K) Q : specific humiity (kg kg -1 ).
Win velocities measure by the sonic anemometer were rotate into a coorinate system aligne with the mean win (McMillen, 1988). he soun virtual temperature by the sonic thermometer inclue the crosswin correction (Hignett, 1992). Humiity correction was applie to the soun virtual temperature to evaluate the real sensible heat flux (Eq. 1) an air temperature (Eq. 2). In this paper, the specific humiity fluctuation in Eq. 1 was calculate as the ratio of water vapor ensity fluctuation measure by the infrare open-path CO 2 /H 2 O gas analyzer to air ensity. As shown in Fig. 6, the correcte sensible heat flux was almost ientical to the raw sensible heat flux, as the humiity correction was negligibly small in the present experiment. hen, the total latent heat (λe) an CO 2 (F c ) fluxes incluing the Webb correction terms were calculate as follows. ρv λ E λ ( 1+ μσ )( w' ρ' v + w' ') ρc ρc F = w ρ c + μ w ρ v + 1+ μσ ) w ρ = (3) ( (4) c where, λ : latent heat of vaporization (J kg -1 ) μ : ratio of the molecular weights of ry air an water vapor σ : ratio of water vapor an ry air ensities ρ v an ρ c : H 2 O an CO 2 ensities by the open-path CO 2 /H 2 O gas analyzer (kg m -3 ). he first an secon terms of the right parenthesis on the right-han sie in Eq. 3 are the raw water vapor flux an the water vapor Webb correction term ue to sensible heat flux. he first, secon, an thir terms on the right-han sie in Eq. 4 are the raw CO 2 flux an the CO 2 Webb correction terms ue to latent an sensible heat fluxes, respectively. Detaile erivations of Eqs. 3 an 4 can be foun in Webb et al. (1980). 4. RESULS 4.1 urbulent fluctuations Figure 1 shows a typical example of turbulent fluctuations for vertical win velocity, soun virtual temperature, an water vapor an CO 2 ensities measure over the asphalt surface uring the aytime. hese fluctuations are plotte as 10 Hz raw ata as the eviation from the 5 minutes mean value. he soun virtual temperature showe positively skewe fluctuation with an amplitue of 5 K, an was positively correlate with vertical win velocity. he CO 2 ensity showe negatively skewe fluctuation with an amplitue of 10 mg m -3 (= 6 ppm), an was negatively correlate with vertical win velocity. Fluctuations of soun virtual temperature an CO 2 ensity were note to be istinctly an negatively correlate with each other using an inepenent instrument. hese results showe that the ey transport of CO 2 was apparently ownwar from the atmosphere to the asphalt surface with the large upwar sensible heat flux. he fluctuation of water vapor ensity was small an i not show correlation with the fluctuation of soun virtual temperature or CO 2 ensity. hese results showe that there is no cross-sensitivity between H 2 O an CO 2 gases by an infrare open-path CO 2 /H 2 O gas analyzer. In the present stuy, the total CO 2 flux was etermine by the raw CO 2 flux an the CO 2 Webb correction term ue to sensible heat flux, because the Webb correction term ue to latent heat flux can be negligible in the present ata. 1.2 Vertical Win Velocity (ms-1) 0.6 0.0-0.6-1.2 8 Soun Virtual emperature (K) 4 0-4 1.2 Water Vapor Density (g m-3) 0.6 0.0-0.6-1.2 8 4 0-4 -8-12 Carbon Dioxie Density (mg m-3) 0 600 1200 1800 2400 3000 IME (0.1s) 10:45-10:50 Aug. 24, 2006 Figure 1 urbulent fluctuations of vertical win velocity, soun virtual temperature, an water vapor an CO2 ensities 4.2 Power spectra an co-spectra Figure 2 shows the normalize power spectra of the vertical win velocity, soun virtual temperature, an water vapor an CO 2 ensity fluctuations observe by the open-path instruments between 10:45 an 11:00. he normalize power spectral ensities S xx (n)/σ 2, were plotte as functions of nonimensional frequency f = nz/u. hese power spectra ha similar frequency structure, an followe the 5/3 power law in the inertial subrange above 0.11. As shown in Figure 3, the
normalize power spectra of CO 2 ensity fluctuation show similar frequency structures uring this observation perio. the CO 2 Webb correction term ue to sensible heat flux showe positive for the whole frequency range, an was negatively correlate with that of raw CO 2 flux. In contrast, the cospectrum of the CO 2 Webb correction term ue to latent heat flux was negligible for the whole frequency range. he cospectrum of the resulte total CO 2 flux showe a positive value for the whole frequency range. Figure 2 Normalize power spectra of the vertical win velocity, soun virtual temperature, an water vapor an CO 2 ensity Figure 4 Cospectra of CO 2 ensity an vertical win velocity normalize by the raw CO 2 flux as functions of non-imensional frequency Figure 3 Normalize power spectra of the CO 2 ensity fluctuations as functions of non-imensional frequency Figure 4 shows the cospectra of CO 2 ensity an vertical win velocity normalize by raw CO 2 flux as functions of the nonimensional frequency. hese were analyze using the same runs as shown in Fig. 3. hese cospectra also ha similar frequency structure between each run, an the flux ominant frequency ranges were aroun 0.05. Figure 5 compares the ensemble-average cospectrum of raw CO 2 flux with those of the Webb CO 2 correction terms ue to latent an sensible heat fluxes, an total CO 2 flux. he cospectrum of Figure 5 Ensemble-average cospectra of the raw CO 2 flux, CO 2 Webb correction terms ue to latent an sensible heat fluxes, an total CO 2 flux 4.3 Webb correction on CO 2 flux over the asphalt surface Figure 6 represents the time series of raw an correcte sensible heat fluxes(eq.1), an raw an total latent heat fluxes(eq.3). he raw an correcte sensible heat fluxes are almost ientical
as the surface was very ry. he latent heat flux represent some positive flux when inclue the Webb correction as to be iscusse later. Figure 6 ime-series ata of raw an correcte sensible heat fluxes, an raw an total latent heat fluxes. he raw an correcte sensible heat fluxes are almost ientical. between 87 an 243 Wm -2 ). he Webb correction term ue to latent heat flux was between 0.008 an 0.016 mg m -2 s -1 (raw latent heat flux was between 21 an 41 Wm -2 ). otal(correcte) CO 2 flux was always positive between 0.001 an 0.097 mg m -2 s -1 (the average was 0.064 mg m -2 s -1 ), an i not show a clear iurnal variation, as i raw CO 2 an sensible heat fluxes. Ham an Heilman (2003) performe a similar experiment over an asphalt surface, an reporte total CO 2 flux by the ey covariance technique of 0.02 mg m -2 s -1 (= 1.8 g m -2 ay -1 ) an CO 2 flux by the chamber technique of 0.03 mg m -2 s -1 (= 2.8 g m -2 ay -1 ) over the asphalt surface. he average total CO 2 flux in the present stuy was about two or three times larger than their result. Although total CO 2 flux measure by the ey covariance technique as reporte by Ham an Heilman (2003) inclue negative values, total CO 2 flux in the present stuy retaine a significantly positive value. 5. DISCUSSIONS As shown in Fig. 6, total latent heat flux incluing the Webb correction term was between 6 an 63 Wm -2 (the average was 32 Wm -2 ), an larger than the result reporte by Ham an Heilman (2003), showing systematic iurnal variation. Over the ry asphalt surface latent heat flux is consiere almost zero, total latent heat flux resulte in a significantly positive value. Figure 7 ime-series ata of the raw CO 2 flux, Webb CO 2 correction terms for the sensible an latent heat fluxes, an total CO 2 fluxes (uncorrecte an correcte the horizontal separation). Figure 7 shows the time-series ata of the raw CO 2 flux, CO 2 Webb correction terms ue to sensible an latent heat fluxes, an total CO 2 flux. he raw CO 2 flux showe a iurnal variation, an was always negative (ownwar from the atmosphere to the asphalt surface) between 0.095 an 0.368 mg m -2 s -1 (the average was 0.283 mg m -2 s -1 ). he CO 2 Webb correction term ue to sensible heat flux also showe a iurnal variation, an was always positive between 0.167 an 0.462 mg m -2 s -1 (the sensible heat flux was Figure 8 Ratio of cospectrum of the raw CO 2 flux to that of the CO 2 Webb correction term ue to sensible heat flux as a function of frequency. he error bars are the stanar eviation. In Fig. 5, the ensemble-average cospectrum of total CO 2 flux ha ominant frequency ranges aroun 0.1 Hz, an the range was consistent with
those of raw CO 2 flux an the CO 2 Webb correction term ue to sensible heat flux. Above 0.62 Hz, the cospectrum of raw CO 2 flux attenuate faster than that of the CO 2 Webb correction term ue to sensible heat flux. As a result, the cospectrum of total CO 2 flux kept a positive value in this frequency range. his can be cause by horizontal separation between the sonic anemometer an the open-path CO 2 /H 2 O gas analyzer. Kristensen et al. (1997) showe that the flux causes a loss of 13% at a height of 1 m with a sensor horizontal separation of 0.2 m. o improve the unerestimation of raw CO 2 flux by this separation, we have assume cospectral similarty. Figure 8 shows the ratio of the cospectrum of raw CO 2 flux to CO 2 Webb correction term ue to sensible heat flux as a function of frequency. he ratio is almost -1 between 0.04 an 0.62 Hz, suggesting negative cospectral similarity. While above 0.62 Hz the ratio increases from -1, suggesting the issimilarity. he issimilarity comes from the sensor separation mentione above. Assuming that the cospectral similarity is kept in the high frequency above 0.62Hz, the CO2 raw flux was ajuste to keep the ratio as -1. As the result, the raw (negative) CO2 flux increase by 0.021 mg m -2 s -1 an correcte total (positive) CO 2 flux ecrease by 0.021 mg m -2 s -1 on average as shown in Fig.7. his correction value was about 33% of the separation uncorrecte total CO 2 flux (0.064 mg m -2 s -1 ) or less than 8% of the raw CO 2 flux ( 0.283 mg m -2 s -1 ). However, the average total CO 2 flux kept a positive value (0.043 mg m -2 s -1 ). CO 2 flux by the ey covariance technique using an infrare open-path CO 2 /H 2 O gas analyzer may inclue some uncertainties of the Webb correction, cross-sensitivity, energy balance closure, an so on. Our result suggeste that the Webb correction may cause an overestimate correction of CO 2 an H 2 O fluxes over a ry asphalt surface where there is no cross-sensitivity between H 2 O an CO 2 gases by an infrare open-path CO 2 /H 2 O gas analyzer. Alternative metho of the flux valiation can be chamber technique an/or close-path eycovariance. hese further stuies are require to evaluate the accuracy of the Webb correction by the ey covariance technique using an open-path CO 2 /H 2 O gas analyzer. REFERENCES Ham, J.M., an Heilman, J.L., 2003: Experimental test of ensity an energy-balance corrections on carbon ioxie flux as measure using open-path ey covariance, Agron. J., 95, 1393-1403 Hignett, P., 1992: Corrections to temperature measurements with a sonic anemometer, Bounary-Layer Meteorol., 61, 175-187 Kono, F., an sukamoto, O., 2007: Air-sea CO 2 flux by ey covariance technique in the equatorial Inian Ocean, J. Oceanogr., 63, 449-456 Kono,F. an O.sukamoto, 2007: Evaluation of Webb correction on CO2 flux by ey covariance technique using open-path gas analyzer over asphalt surface, J.Agric. Meteorology, 64, 1-8. Kristensen, L., Mann, J., Oncley, S.P., an Wyngaar, J.C., 1997: How close is close enough when measuring scalar fluxes with isplace sensors?, J. Atmos. Oceanic echnol., 14, 814-821. Leuning, R., 2007: he correct form of the Webb, Pearman an Leuning equation for ey fluxes of trace gases in steay an non-steay state, horizontally homogeneous flows, Bounary-Layer Meteorol., 123, 263-267 McMillen, R.., 1988: An ey correlation technique with extene applicability non-simple terrain, Bounary-Layer Meteorol., 43, 231-245 Ohtaki, E., sukamoto, O., Iwatani, Y., an Mitsuta, Y., 1989: Measurements of the carbon ioxie flux over the ocean, J. Meteorol. Soc. Jap., 67, 541-554. Webb, E.K., Pearman, G.I., an Leuning, R., 1980: Correction of flux measurement for ensity effects ue to heat an water vapour transfer, Q. J. R. Meteoro. Soc., 106, 85-100