Comparison to literature data

the cloud due to a flight level change, the LMS was entered once again. At about 23:00 UTC the aircraft flew back into UT air and entered the third cloud (extended up to a pressure altitude of about 125 hPa) before descent. The occurrence of all three clouds was confirmed by the CWC measurements (cf. Fig. 6.2). To decide whether a measurement was obtained inside a cloud, KNMI provide a cloud cover fraction model output with a temporal resolution of one minute along the flight route (Fig. 6.5b). These data were averaged for the same three minute bins the particle size distributions were integrated for. If the cloud cover fraction was found to be above 0.1 and the model relative humidity was above 90 % for an individual bin, the measurement was marked as “obtained inside cloud” and was excluded for further analyses.

6.1.3 Comparison to literature data

As already mentioned in Chapter 1, only few measurements of the particle size distribution were carried out in the UT/LMS region up to 2011. The data published in the last 20 years are summarized in Tab. 6.1. Most of the data were obtained in geographical regions where the OPC unit did not measure up to May 2011. Thus all CARIBIC OPC data (obtained until flight LH344) were divided into three different major air mass data subsets: the mid-latitude UT, the mid-latitude LMS, and the tropical and subtropical mid- and upper troposphere (MUT). Due to the relatively long life time of accumulation mode particles in the UT/LMS (several weeks; cf. Sec. 2.3),

Table 6.1: Summary of previous measurements of the particle size distribution in the UT/LMS region (until 2010).

location and altitude time campaign published in west of Tenerife

(9-11 km)

July 1997 Second Aerosol Characterization

6.1 Regular data analysis and evaluation

Table 6.2: Classification of the CARIBIC measurement regions for data comparison. To differ between UT and LMS the ECMWF PV along the flight track was used. (1 PVU = 10^-6 Km²kg^-1s^-1)

region latitude longitude UT/LMS

separation mid-latitude UT > 40° N averaged over all

longitudes

PV < 1 PVU mid-latitude LMS > 40° N averaged over all

longitudes

PV > 3.5 PVU tropical and

subtropical MUT

35°S < lat < 35°N averaged over all longitudes

PV < 1 PVU

measurements in the same air mass at different locations can be compared. As in the UT the longitudinal mixing is faster than the meridional one, the compared regions have to be within the same latitudes.

The air mass classification is given in Tab. 6.2. Before averaging the OPC data for each class, cloud measurements, as well as data obtained during ascend and descend were excluded. Because O3 data were not available for all measurement flights, the PV was used to distinguish between tropospheric and stratospheric air masses. The measurement was attributed to the troposphere if the PV was lower than 1 PVU (1 PVU = 10^-6 Km²kg^-1s^-1) and to LMS if PV was larger than 3.5 PVU. Data with PV 1 PVU to 3.5 PVU were not used to have a clear separation.

Because the measurements (CARIBIC and literature) were obtained at different flight levels, a quantitative comparison is reasonable only by normalizing all data to standard pressure and temperature (STP: p=1013.25 hPa; T=273.15 K). Unfortunately only de Reus et al., [2001] and Krejci et al., [2003] presented their data at STP conditions. All other data were assumed to be given at ambient conditions and were converted to STP by using the US standard atmosphere [Seinfeld and Pandis 1998, p. 7, 1293] and the given altitude. The mean height was used if an altitude range was given.

Uncertainties caused by the difference of the true measurement conditions compared to the US standard atmosphere values were estimated to be smaller than 20% (based on a pressure and temperature variation of ±30 hPa and ±15 K at a pressure altitude of 250 hPa).

Another uncertainty in the data comparison arises from the different measurement instruments and platforms. The published data were measured onboard research aircraft with usually relatively short sampling lines whereas the CARIBIC sampling line is quite long with 4 m length. Between inlet and the OPC unit the sample air is warmed by about 90 K from -55°C to +35°C and an initial relative humidity (rH) of 70% decrease to nearly 0% at +35°C. Consequently, nearly all water evaporates from the aerosol particles. Simulations performed by Jens Voigtländer (IfT/TROPOS) with the Computational Fluent Dynamics (CFD) program “FLUENT” showed that H2SO4 – H2O particles with a wet diameter of dp,wet = 1000 nm and a corresponding dry

87  

Table 6.3: Modeled drying of H2SO4 – H2O particles at the CARIBIC aerosol inlet system (inlet + sampling line). At the aerosol inlet the temperature and relative humidity was set to -55°C and 70%, respectively. At the sample line exit the temperature was set to +35°C. The residence time of the measurement air from the inlet towards the OPC unit was calculated to be 0.65 seconds. The dry diameter (dp,dry) corresponding to a certain wet diameter (dp,wet) gives the particle size if only the H2O would evaporate and the H2SO4 would remain. The simulations were performed by Jens Voigtländer (IfT/TROPOS) with the Computational Fluent Dynamics (CFD) program “FLUENT”.

dp,wet – at inlet

diameter of dp,dry = 682.1 nm shrink during the sampling to a wet diameter of dp,wet = 806.8 nm (dp,dry = 681.7 nm)²³. For other particle diameters the results of the CFD simulations are listed in Tab. 6.3.

The size distributions given in the literature were mostly measured using (much) shorter aerosol sampling lines with (much) shorter measurement air transport times.

Unfortunately, the relative humidity at which the measurements took place was not given in the literature. However, the CFD simulations for the CARIBIC inlet system showed that the water evaporates instantaneously from the particles after entering the inlet, because of the strong temperature increase. Therefore it is assumed the particle size distributions given in the literature were measured at comparable humidity.

Figure 6.6 shows the averaged CARIBIC OPC size distributions as well as the literature data for the mid-latitude UT (a), the mid-latitude LMS (b), and the tropical and subtropical MUT (c). Within each graph red squares represent the CARIBIC mean value and the gray area indicates the 10 % and 90 % percentile. The OPC integration time was set to 900 s (cf. Sec. 5.3.1) for this comparison. Sometimes no particles were counted within 15 min in certain size channels (especially those counting particles larger than 800 nm), leading to a jump of the 10 % percentile to zero. For the routinely used 180 s resolution the 10% percentile drops to zero at much smaller particle diameters (cf. Fig. 6.4). However, the 180 s resolution was chosen for the routine analysis to detect also small scale changes of the particle size distribution. The size distributions taken from the literature are displayed as lines. While dashed lines in the same color indicate the range of the measured concentration from a study, solid lines represent data, already averaged by the authors. The general shape of the CARIBIC size distribution and its variation within the regions will be discussed in detail within the statistical analysis in Sec. 6.3. Within this section the focus lays on the comparison to other in situ measurements.

23 The particle wet diameter (dp,wet) describes the size of a certain aerosol particle at ambient humidity.

The corresponding dry diameter (dp,dry) gives the residual particle size if all H2O evaporates from the particle.

6.1 Regular data analysis and evaluation

The CARIBIC mid-latitude tropospheric data in Fig. 6.6a are mostly within the concentration range published by Schröder et al., [2002]. The Schröder minimum concentration curve is similar to the CARIBIC 10% percentiles. Only for particles 450 nm < dp < 900 nm the CARIBIC mean concentration is somewhat larger than the maximum concentration line of Schröder et al. [2002]. The slope of the two curves of Schöder et al. [2002] is similar to the CARIBIC data. The slope of the size distribution published by Young et al., [2007] is similar, too. However, the total particle concentration is much lower than the CARIBIC mean, reaching only the CARIBIC 10 % percentile and the lower limit curve of Schröder et al., respectively. As stated above, the atmospheric life time of UT accumulation mode particles is relatively long.

Consequently, the different measurement locations (Young et al.: central USA, Schröder et al.: Germany, CARIBIC 120°W < θ < 140°E) can be not the only reason for this huge difference (up to one order of magnitude). Because all data were corrected for aerosol inlet efficiency, the data from different research aircraft with different aerosol inlets should be comparable. The normalization of concentrations to STP would explain

Figure 6.6: Comparison of particle size distributions measured in the latitude UT (a), the mid-latitude LMS (b), and the tropical and subtropical MUT(c). The borders of the regions are given in Fig.

6.1 and Tab. 6.2. Gray areas indicate the CARIBIC 10- and 90% percentile (15 min average). The CARIBIC mean with the same averaging time is indicated with red squares. If not given, published data (solid and dashed lines) were normalized to standard conditions (p = 1013.25 hPa, T = 273.15 K) using the US standard atmosphere.

CARIBIC 10 - 90% percentile CARIBIC mean

de Reus et al., 2001 (8-12.5km; min-average-max) de Reus et al., 2000 (9-11km)

Krejci et al., 2003 (10-12.6km) Zaizen et al., 2004 (11km; min-max)

dN/dlog(dp) [1/cm3 STP]

d_p [nm]

tropical and subtropical mid- and upper troposphere

100 1000

CARIBIC 10 - 90% percentile CARIBIC mean

Schröder et al., 2002 (10-12km) Young et al., 2007 (10km LMS)

10³ CARIBIC 10 - 90% percentile

CARIBIC mean

Schröder et al., 2002 (10-12km) Young et al., 2007 (10km UT)

dN/dlog(dp) [1/cm3 STP]

d_p [nm]

mid-latitude upper troposphere

(a)

89  

a difference of 20%, but not 500% to 1000%. Possible reasons for the much lower concentrations from Young et al. might be the measurement statistics and the season at which the measurements were carried out. The CARIBIC dataset is based on 39 measurement flights carried out over one year and Schröder et al. [2002] report averages over ten measurement flights in July and August. On the contrary, Young et al.

reported a singular measurement on one measurement flight in December 2005 (region 1 in Young et al., [2007]). Thus the low concentration of accumulation mode particles measured by Young et al. might occur locally but might be not representative for the yearly average in the mid-latitude UT.

Schröder et al. [2002] attributed their measurement to the tropopause region and thus the data are shown in the mid-latitude LMS comparison, too. Young et al. [2007]

differed between UT and LMS measurements by using the PV and the thermal tropopause definition. The UT measurement at 10 km altitude (PV < 2 PVU) was used for Fig. 6.6a, while the measurement above the thermal tropopause (07.12.2005 region 2, altitude also 10 km) is shown in Fig. 6.6b. In the LMS for particles smaller than ~300 nm, the CARIBIC concentrations are within the range given by Schöder et al.

[2002]. Above 300 nm the slope of the maximum curve of Schröder et al. [2002]

changes and the CARIBIC data are significantly higher. The stratospheric particle size distribution published by Young et al. [2007] is mostly within the range of data by Schröder et al. [2002] but lower than the CARIBIC data. The slope of all three data sets is similar. The higher CARIBIC concentration might be caused by the PV limit of 3.5 PVU for separating stratospheric air when compared to 2 PVU of Young et al.

[2007] and tropopause of Schröder et al. [2002]. Therefore the latter two data sets might be more influenced by tropospheric air than the CARIBIC measurements. The occurrence of the peak (mode) in the CARIBIC size distribution at ~700 nm will be discussed in detail in Sec. 6.1.4

Most of the data were published for the tropics and subtropics (Fig. 6.6c). The particle size distributions measured west of Tenerife [de Reus et al., 2000] and the data from Surinam [Krejci et al., 2003] are comparable with the CARIBIC means. While the average concentration, given by de Reus et al. [2001] is higher for dp < 600 nm, the max-min range covers the CARIBIC 10- and 90% percentile. The data published by Zaizen et al., [2004] are generally somewhat lower than the CARIBIC data but higher than the CARIBIC the 10% percentile.

Considering that the measurements were obtained in different locations (longitude), on different times scales (number of flights, measurement season), and with different research aircrafts (aerosol inlets, measurement instruments), the CARIBIC data are generally in reasonable agreement with the published data. For all compared regions the published data are within the same range of the CARIBIC OPC and the size distribution slopes are similar. The best agreement is observed for the tropical and subtropical MUT. Consequently, it is concluded that the CARIBIC OPC data are comparable with other data sets and representative for the observation locations.

6.1 Regular data analysis and evaluation