Equivalent black carbon aerosol measured over the Southern Ocean in the austral summer of 2016/2017, during the Antarctic Circumnavigation Expedition.

***** Dataset abstract *****

The authors would highly appreciate to be contacted if the data is used for any purpose.

We measured equivalent black carbon (eBC) with an aethalometer (model AE33, Magee Scientific) at a time resolution of one second during the Antarctic Circumnavigation Expedition (ACE). We report five-minute averaged data, cleaned from exhaust gas influence. Temporal coverage is from December 20, 2016 to April 10, 2017. 

The mass concentration of eBC, reported in ng m^-3, reflects how far fossil fuel combustion or biomass burning contribute to the aerosol population over the Southern Ocean and between South Africa and Europe. Over the Southern Ocean there are no sources of eBC, except for ship emissions and (sub-)Antarctic station emissions, and hence an enhancement of eBC points towards long-range influence from Africa, Australia, New Zealand and South America. When plotted against latitude, eBC concentrations drop south of 60°S, indicating a more pristine environment. Elevated concentration around the equator are likely influenced by biomass burning in tropical Africa. 

***** Original data collection *****

We operated an aethalometer, model AE33 (Magee Scientific) behind a standard Global Atmosphere Watch whole air inlet (Schmale et al., 1027; 10.1038/sdata.2017.3) in an aerosol container constructed by the Paul Scherrer Institute situated on the foredeck of the R/V Akademik Tryoshnikov. More about this set-up can be found in the cruise report here: https://doi.org/10.5281/zenodo.1443511.

Data acquisition is performed by the computer in the instrument itself. The highest possible time resolution, one second, was stored directly. Original output was stored as text files on the instrument situated in the PSI laboratory container. No data processing was done during collection.

***** Data processing *****

The original txt files were loaded with the AETHALOMETERloader coded at the Laboratory of Atmospheric Chemistry at the Paul Scherrer Institute in April 2013 in Igor Pro v6.37 (WaveMetrics Inc.). Only mass concentrations derived for channel 6, i.e. wavelength 880 nm, were used. 

Exhaust influence was removed from the data. Exhaust influence on the data was identified as follows: The particle number concentration as measured by the CPC 3022 was used as the basis. First the absolute value of the ratio of the raw over the smoothed signal (running average over 12 points for five-minute time resolution data, which represents a curve similar to a baseline) was calculated. Ratio values that are larger than 1.24 and smaller 0.51 were flagged. The same was done for ratios of the raw and smoothed CO2 data with the thresholds being 1.2 and 0.8. In addition, since exhaust influence is characterized by abrupt concentration changes, subsequent values of the raw particle number concentration time series that were more than 50 cm^-3 apart were flagged. Finally, the flagged and original time series were compared by eye to identify any further values influenced by exhaust. The natural variability in number concentration is evident in timescales of hours rather than seconds, which are characteristics for exhaust plumes. New particle formation events can happen on timescales shorter than one hour: those were not eliminated. 

No further correction factors were applied.

Data were processed in Igor Pro v6.37. AETHALOMETERloader April 2013, Paul Scherrer Institute. 

***** Quality checking *****

The data have been cleaned from exhaust gas based on the spikiness of the total particle number concentration, the CO2 and equivalent black carbon concentrations at high time resolution (10.5281/zenodo.2636690, 10.5281/zenodo.2636779, 10.5281/zenodo.2636763, respectively). 

Due to the low concentrations of eBC over the Southern Ocean, the signal to noise ratio can be low, leading to five-minute average values varying by +/- 100 ng/m^3 around 0 ng/m^3. It is advisable to use longer integration periods in such cases. We have removed all data points that have a variability larger than +/- 200 ng/m^3 around 0 ng/m^3 because those are meaningless to determine the actual concentrations over the Southern Ocean that are < 50 ng/m^3. By plotting the time series the variation around the zero-value is apparent.

The influence of the ship exhaust can be very strong, producing a signal of several tens of µg/m^3. After such episodes, noise levels are increased until the filter tape advances to a clean spot.

***** Standards *****

We follow the best practice that we are aware of in the respective scientific community.

***** Further information for interpreting the data and using the dataset *****

Negative values: Negative values are an indication for a higher noise level. Each negative value should be embedded in a period where positive and negative values of similar intensity alternate. In this case a longer averaging time than 5 minutes is needed to estimate the ambient concentration.  

Timescales: The variation in the concentration is a function of the vicinity and variability of the eBC source. In pristine conditions where the temporal variation of the eBC concentration is very low, it is more likely to encounter a steady background value. In the vicinity of a source, e.g., the fire emissions near tropical Africa, the variability is much higher and depends on air mass dilution, mixing, and potentially wet and dry deposition processes.  

Interpolation: Interpolation of missing values is generally not meaningful. However, when in pristine conditions the variability of the concentration is very low and a general background value can be estimated and applied in higher time resolution. Such an approach would however discount the real ambient conditions encountered, because missing data represents mostly exhaust gas influence which is much higher than the background value. 

Aggregating to a lower temporal resolution: Whether or not this should be done depends on the research question. The data set features a number of environmental processes that happen on different timescales. Averaging will eliminate shorter term signals, but is useful to determine the eBC background concentration over pristine conditions. For more information, see Schmale et al. (2019, in review) “Overview of the Antarctic Circumnavigation Expedition Study of Preindustrial-like Aerosols and their Climate Effects (ACE-SPACE)”. 

***** Dataset contents *****

- ACESPACE_equivalent_black_carbon_aerosol.csv, data file, comma-separated values
- data_file_header.txt, metadata, text
- README.md, metadata, text

The data file listed above contains five-minute averaged values of equivalent black carbon (eBC) measured during the Antarctic Circumnavigation Expedition. Timestamps are the end of the five-minute period over which the eBC values were averaged. Latitude and longitude are average values of the position of the measurement during the five-minute interval. 

NaN values of eBC denote missing values because of e.g., ship exhaust contamination, maintenance, instrument failure or signal noise levels exceeding 200 ng/m^3. For latitude and longitude, NaN values are noted in cases where position data was not available for the given time period. 

***** Dataset contact *****

Julia Schmale, Paul Scherrer Institute, Switzerland. ORCID: 0000-0002-1048-7962 julia.schmale@psi.ch

***** Dataset license *****

This ACESPACE equivalent black carbon aerosol dataset is made available under the Open Data Commons Attribution License: https://opendatacommons.org/licenses/by/1.0/index.html

***** Dataset citation *****

Please cite this dataset as:
Schmale, J., Henning, S., Tummon, F., Hartmann, M., Baccarini, A., Welti, A., Lehtipalo, K., Tatzelt, C., Bukowiecki, N. and Gysel-Beer, M. (2019). Equivalent black carbon aerosol measured over the Southern Ocean in the austral summer of 2016/2017, during the Antarctic Circumnavigation Expedition. (Version 1.0) [Data set]. Zenodo. http://doi.org/10.5281/zenodo.2636763