Aircraft and ground-based observations of boundary layer CO2 concentration in anticyclonic synoptic condition



[1] Concentrations of atmospheric CO2 and aerosol were measured in a field campaign conducted in winter 2006 around Mt. Tsukuba, Japan using ground-based CO2 analyzers, a lidar, and sky radiometers as well as CO2 analyzers onboard an aircraft. Vertical measurements revealed occasional similarity between the profiles of CO2 and aerosol concentrations, though their temporal variations are not always coordinated because of the effects of local sources or sinks. A sudden increase of downward winds, due to the approach of an anticyclonic synoptic flow, resulted in a rapid decrease in both the CO2 and aerosol concentrations in the boundary layer. These observation results have demonstrated that simultaneous measurements with airborne and ground-based instruments set on the summit/foot of a mountain are useful for the study of variability of CO2 concentration in the boundary layer.

1. Introduction

[2] The variability of carbon dioxide (CO2) concentration in the atmospheric boundary layer (ABL) has been intensively studied [e.g., Sarrat et al., 2009; Shashkov et al., 2007; Gibert et al., 2006]. The CO2 fluxes from the surface vegetation have often been observed with observation towers, yielding data used for model development [e.g., Peters et al., 2007] and validating satellite observations [e.g., Barkley et al., 2007]. Meanwhile, synoptic weather such as high/low pressures also affects the CO2 concentration in the ABL through the replacement of airmass vertically or horizontally on a much larger scale than local influences [e.g., Adamson et al., 2006; Brooks et al., 2003]. The synoptic variability of CO2 concentration has been simulated by means of transport models and/or general circulation models [e.g., Law et al., 2008; Chan et al., 2004], and observed in cyclones [Hurwitz et al., 2004]. In contrast, detection of the synoptic variability of CO2 in anticyclones has not yet been fully studied.

[3] In this study we describe the results of a field campaign conducted in winter 2006 around Mt. Tsukuba, Japan using ground-based CO2 analyzers, a lidar, and sky radiometers as well as CO2 analyzers onboard an aircraft. Although aircraft profiling has been reported previously [e.g., Shashkov et al., 2007; Styles et al., 2002], such a coordinated approach has been undertaken for the first time in the temporal/spatial variability study of the CO2 concentration in and just above the ABL. To detect the synoptic- or mesoscale effect on the variability of CO2 concentration, here we pay particular attention to the correlation between aerosol and CO2 concentrations in the ABL as well as in the free troposphere, since the concentrations of these two species occasionally reflect air-mass variation on regional or larger scales [Gibert et al., 2008]. In addition, the observation of an aerosol profile serves to estimate the ABL height. As compared with the local variability, the synoptic weather can produce more extensive effects on both concentrations.

[4] The aerosol concentration in the ABL can be estimated from the aerosol optical depth (AOD) measured with a sky radiometer by appropriately separating the contributions from ABL and free troposphere [Aoki and Fujiyoshi, 2003]. Since the ABL height is in the range of 500–1500 m in the mid-latitudes, here we employ the sky-radiometer and meteorological data taken at the station [Hayashi, 2006] on the summit of Mt. Tsukuba (Figure 1a). Since this is an isolated peak in the midst of the Kanto Plain (white area in Figure 1a), this station is one of the best locations in Japan to observe the data pertinent to the part of the atmosphere that is mostly above the ABL. The column AOD in the ABL is derived from the data of two sky radiometers installed on the summit and at the mountain foot (Makabe, Figure 1a), giving the vertical column length of 836 m.

Figure 1.

Sites on the Kanto Plain for CO2 and aerosol measurements: Mt. Tsukuba and Makabe. Numbers 1–12 are the AMeDAS sites for surface wind measurements. (a) Topography with a contour interval of 100 m and (b) land use derived from the Landsat 7 Enhanced Thematic Mapper Plus image on 11 December 2006; forest (deep green), grassland/shrub (light green), agricultural (yellow), urban (red), and water (dark blue) areas.

2. Instrumentation and Observation

[5] From 1 to 18 December 2006, the ground-based instruments were operated at the summit of Mt. Tsukuba (T-site: 140.098°E, 36.226°N, 881 m AMSL) and the city hall in Makabe (M-site: 140.092°E, 36.278°N, 45 m AMSL), located 5.8 km north of Mt. Tsukuba (Figure 1a). At each site a non-dispersive infrared gas analyzer (NDIR: LiCOR, LI840), a sky radiometer (Prede, POM02), and meteorological instruments were used to monitor the CO2 concentration, radiation amount, and meteorological parameters (e.g., wind speed/direction), respectively. For CO2 measurement, outside air was drawn into the instrument by a diaphragm pump, dried while passing through a dryer filled with magnesium perchlorate, and supplied to the NDIR. Concentration values were saved to a data-storage (Campbell, CR1000) every 5 s. Each NDIR was calibrated periodically for 2 min every hour by flowing standard diluted gas (354.55 and 393.53 ppm CO2 in dry air).

[6] The sky radiometers were used to measure the direct solar irradiance and diffuse sky radiance during daytime at five wavelengths (400, 500, 675, 870 and 1020 nm). The AOD was obtained from the aureole (aerosol mode) and direct sunlight (sun-photometer mode) measurements under clear sky conditions in every 3 and 7 min, respectively. Since there is an altitude difference of 836 m between the T- and M-sites, the observed difference in the AOD between these two sites gives a good measure of the aerosol concentration in the lower troposphere (which is roughly equivalent to the ABL, depending on the atmospheric conditions).

[7] The airborne measurements were carried out on clear-sky days (December 4 and 18) to observe the CO2 concentration, humidity, and temperature profiles inside and above the ABL using an aircraft (Cessna 172). Each flight path consisted of spirals with ∼500 m radius, descending over the T-site and ascending over the M-site. The altitude range was between ∼300 to ∼3000 m above ground, in accordance with the regional flight regulations. The flight duration was 11:45–13:05 local time (UTC+9) on December 4 and 11:05–13:12 on December 18, each providing ∼7 profiles of the CO2 concentration. A compact version of the NDIR [Machida et al., 2008] was employed for the measurement, as in the case of another flight campaign [Saito et al., 2008]. This onboard NDIR was calibrated using the standard gas with two different concentrations for 1 min every 6 min during the flights. The in-situ data of CO2 concentration and outside temperature/humidity that was measured with a thermometer (Vaisala, HMP230) were saved to a data-storage every 2 s.

[8] A Mie-scattering lidar (Leosphere, ALS300) was operated on the rooftop of the city hall at the M-site. The vertical aerosol profile was measured at the laser wavelength of 355 nm using a 20-cm diameter telescope. The altitude range below ∼200 m was not measured because of the lack of overlap between the laser beam and telescope field-of-view. Since the emitted wavelength was in the eye-safe UV range, the airborne measurement could be performed concurrently with the lidar measurement. The lidar data, with a vertical resolution of 15 m and temporal resolution of 36 s, were analyzed by Fernald's method to find the vertical distribution of the aerosol backscatter coefficient. In the analysis, aerosol-free atmosphere was assumed at 3.5 km altitude, and the aerosol extinction-to-backscattering ratio was estimated from the AOD obtained from the two sky radiometer measurements.

[9] For the analysis of the wind field data, we employ the data derived from Meso Regional ANALysis Data Set (MANAL) provided by the Japan Meteorological Agency (JMA). This MANAL dataset is based on 10 km by 10 km resolution reanalysis computed every 3 h. For 12 sites indicated in Figure 1a, 10-minite-average surface wind data from Automated Meteorological Data Acquisition System (AMeDAS) operated by JMA were also used to analyze the surface wind field.

3. Results and Discussion

[10] Figure 2 shows the results of the lidar and aircraft measurements of the vertical profiles of the aerosol backscatter coefficient (βa), CO2 concentration, and relative humidity (RH) observed on the two flight days (December 4 and 18), with the wind field data from the MANAL. The lidar measurements and in-situ airborne profiles in different colors denote the results in different sets of ascending/descending flights, as indicated in Figures 2a and 2e).

Figure 2.

Vertical profiles observed during each ascending/descending flight duration on December 4 and 18: (a, e) aerosol backscatter coefficient (βa) from the lidar measurement at the M-site (local time is indicated with color bars), (b, f) carbon dioxide (CO2) and (c, g) relative humidity (RH) from the airborne and (d, h) wind direction and speed (vectors) and vertical motion (colored circles) at 09:00, 12:00, and 15:00 from the MANAL.

[11] On the basis of these data, we compare the profiles of CO2 and aerosol concentrations. Around noon (11:45–12:06, red dots) on December 4, both the aerosol backscatter coefficient (Figure 2a) and CO2 concentration (Figure 2b) were large in the ABL (below ∼1.1 km) with noticeable peaks at ∼0.9 km. After 1 h (blue dots), the “sharp” peaks disappeared, and below 0.8 km, the aerosol backscatter coefficient increased whereas the CO2 concentration decreased. Figure 2d shows the vertical motion (ω: colored circles) and horizontal wind direction and speed (arrows) at 09:00, 12:00, and 15:00 on December 4. The dissipation due to upward winds around noon presumably explains the disappearance of “sharp” peaks, while the increase in the aerosol concentration near the ground level suggests the increased contribution from local sources on the Kanto Plain.

[12] As compared with the data on December 4, the results on December 18 (Figures 2e2h) indicate more drastic changes that occurred in the atmospheric conditions. After 12:30, CO2 and aerosol concentrations decreased rapidly, accompanied with the decrease in RH (Figure 2g), while the values in the free troposphere (above ∼1.5 km) were nearly intact. During the last flight (12:54–13:12), it turned out that both the aerosol and CO2 concentrations were almost homogeneous below 3 km, indicating a sudden disappearance of the ABL structure. Figure 2h indicates that wind speed was stronger in the afternoon on December 18, with ∼3 times increase in horizontal wind speed (arrows) and ∼6 times increase in the downward wind (ω) below 3 km. It is evident from Figure 2h that a layer characterized by remarkable downdrafts (more than 30 hPa/h) had gradually descended: ∼3 km at 09:00, ∼2 km at 12:00, and ∼0.6 km at 15:00. Moreover, northwest winds were persistent throughout the day. Below, we will discuss the remarkable phenomena in relation to the ground observation data.

[13] Figures 3a and 3e show the daytime variations of the aerosol backscatter coefficient; Figures 3b and 3f show column value of AOD; Figures 3c and 3g show CO2 concentration, and Figures 3d and 3h show wind speed. On clear days such as December 4, aerosol concentration usually increases from the morning to late afternoon in the ABL, as indeed observed at the M-site (Figures 3a and 3b). In contrast, the temporal variation of column optical depth was much smaller at the T-site (Figure 3b). The CO2 concentration, on the other hand, is affected by photosynthetic activities in the forests around Mt. Tsukuba (Figure 1b), and usually decreases during the daytime under clear sky conditions, as actually seen in Figure 3c. The appearance of a certain sort of advected plume after 11:00 between 0.7–1.0 km (Figure 3a) and the increase in the CO2 concentration observed at around 13:00 at the T-site (Figure 3c) are consistent with the change in the concentration profiles shown in Figures 2a and 2b.

Figure 3.

Time series from 09:00 to 15:00 local time on December 4 and 18 of (a, e) aerosol backscatter coefficient (βa) from the lidar measurement at the M-site, (b, f) aerosol column optical depth at 400 nm from the sky radiometers, (c, g) CO2 concentration from the ground-based NDIR, and (d, h) surface wind speeds at the sites indicated in Figure 1a. Circles (black) and triangles (gray) represent the results at the M- and T-sites, respectively. Blanks indicate lack of data.

[14] Just after 12:30 on December 18, sudden decrease in the aerosol (Figures 3e and 3f) and CO2 concentrations (Figure 3g) were observed also in the ground data, as in the lidar and airborne measurements shown in Figures 2e2g. With these changes, a sudden increase in the surface wind speed (Figure 3h) was observed from ∼1 m/s at 12:00 to ∼4 m/s at 13:00 at the M-site. The values of the aerosol backscatter coefficient and of the AOD measured at the M- and T-sites were nearly the same after 12:30 (Figures 3e and 3f), indicating the decrease of the aerosol concentration in the ABL to the level of the free-troposphere concentration. Thus, we have a result from the ground observation that is fully consistent with the concurrent lidar observation.

[15] Here we consider the effect of the wind field on the simultaneous and sudden decrease in the aerosol and CO2 concentrations. The rapid increase in local surface winds during the afternoon on December 18 at almost all the sites (Figure 3h) was possibly originated from the synoptic weather on the Kanto Plain. Figure 4 shows the pressure patterns, horizontal wind vectors, and downdraft distributions on December 4 and 18. Figure 4 indicates that the dominance of northwesterly winds in the lower troposphere in East Japan, while inhomogeneous topography and land cover generally disturb and weaken streamlines near the surface (insets for Figures 4a and 4c). On the afternoon of December 18, directions of surface winds (inset for Figure 4d) were the same as those of lower tropospheric winds (Figure 4d), due to the subsidence of the wind fields (Figures 2h and 3e). Figures 4b and 4d suggests that this subsidence was a phenomenon with a scale larger than mesoscale (i.e., larger than the scale of land/sea breezes, for example). Thus, it is likely that just after the noon, a sudden increase in the wind speed occurred on a synoptic scale as part of clockwise streamlines of surface winds persisting over the Kanto Plain. The increase in downward winds (Figure 2h) that co-occurred with the increase in northwesterly winds (inset for Figure 4b) led to drastic variability of CO2 and aerosol concentrations in the ABL after 12:30.

Figure 4.

Weather map on (a, c) December 4 and (b, d) December 18 from the JMA: downdrafts (ω) and horizontal wind vectors (green arrows) of the 700 hPa, and geopotential height (contours, interval 60 hPa) of the 500 hPa. Insets show streamlines of surface winds over the Kanto Plain. Symbol “M” represents the M-site.

[16] The synoptic phenomenon observed on December 18 over the Kanto Plain is ascribable to anticyclones (marked as ‘H’ in Figure 4) located on the west of Japan. From these figures, it is apparent that there was no noticeable effect of weather front during the daytime on the day. The region with strong downdrafts moved from western Japan at 09:00 (Figure 4b) to the Kanto Plain area at 15:00 (Figure 4d). Accordingly, the downdrafts occurred also at the M-site (Figure 2h) in the afternoon, causing the downward movement of dry air from the higher to lower altitudes in the troposphere, as indeed evidenced in the humidity profile in Figure 2g. Also the CO2 and aerosol concentrations that might have existed in the higher troposphere during the previous night had probably been squeezed downward, leaving the ABL nearly as clean as the free troposphere. It is expected that this type of changes in the atmosphere may often occur in the east of an anticyclone.

4. Conclusions

[17] Variability of CO2 concentration in the ABL was observed by coordinated measurements with airborne and ground-based instruments set on the summit/foot of the mountain. Also, we have observed rather exceptional replacement of boundary-layer air with much clear air in the free troposphere, presumably due to rapid changes in wind fields associated with approaching anticyclones. In particular, the in-situ airborne profiles on December 18 were obtained before and after the rapidly increasing downdraft observed on a synoptic scale. When such synoptic scale phenomena hit to the sites, other local disturbances from surface sources/sinks on inhomogeneous land cover (Figure 1b) exerts negligible effects on the ABL. Similar influence of synoptic changes on the CO2 concentration in the ABL was pointed out also in a previous study [Shashkov et al., 2007]. As compared with this previous study, however, the present observation has revealed a very rapid homogenization (in less than 30 min) of the CO2 concentration in relation to the anticyclonic synoptic situation. In contrast, the concentrations in the ABL on December 4 were more affected by local advections, in spite of anticyclones located alike on both the days. The region of downdrafts was limited by the trough (geopotential-height contours in Figures 4a and 4c) along the east of the cyclones (marked as ‘L’).

[18] Further long-term measurements are desirable to understand the synoptic- or mesoscale variability of CO2 concentration. The synoptic variability can also be found in mid-latitudes in association with remarkable baroclinic activity such as generation and running of high/low pressures. Under good as well as such unfavorable weather conditions, measurements are attempted using commercial aircrafts (the CONTRAIL [Machida et al., 2008]), providing information on CO2 profiles in the free troposphere.


[19] The authors would like to thank T. Machida of NIES and H. Suto of JAXA for support in using the airborne instrumentation and K. Aoki of University of Toyama for analyzing results from the sky radiometers. We are also grateful for the use of Meteorological Observation Station at Mt. Tsukuba supported by Intramural Research Project (S) for 2005 at the University of Tsukuba.