| Abstract |
In the current thesis we present the analysis of a long period data set of particle number size distributions, measured at the Environmental Chemical Processes Laboratory (ECPL) atmospheric monitoring station, situated at the remote site of Finokalia (35o 20'N, 25o 40Έ) on the island of Crete in Greece. The study uses data from a Scanning Mobility Particle Sizer (SMPS) for the time period spanning from May 2008 to December 2012.
Size distributions and the derived corresponding number concentrations of atmosphe¬ric particles in the submicron range, were found to demonstrate a clear seasonal variation. Number concentrations of four size ranges, Nucleation Mode particles (10 — 20nm) exhibited maxima during wintertime, while Aitken Mode particles (20 — 100nm), Accumulation Mode particles (100 — 800nm) and Total particles (10 — 800nm) exhibited maxima during sum¬mertime. A statistically significant anti-correlation was found between Accumulation and Nucleation Mode particles. From each Mode's diurnal pattern, from the seasonal mean size distributions and from analysis in terms of air mass origin, we can conclude that Nucleation events and consequential New Particle Formation (NPF) are bound to the existence of A¬ccumulation Mode particles, which are able, when in large concentrations, to suppress the NPF phenomenon. NPF events were mainly observed during springtime with a secondary maximum during October. The above indicate that two main antagonizing mechanisms are controlling the phenomenon. Firstly the existence of Accumulation mode particles, largely contributing to the Condensation Sink, is a removing mechanism, via condensation on their surface, of the necessary precursor gases, like H2SO4 and NH3. On the other hand solar radiation is a key feature of the production of the aforementioned chemical species in the atmosphere, favoring nucleation.
In order to parametrize the particle number size distribution and its behaviour we have developed an automating algorithm, implementing the Expectation Maximization Algorithm (EM Algo), so that we can fit the multi-lognormal size distribution function to ambient particle number size distributions. In order to control the algorithm's behaviour we have cross-checked the derived results for each fitted lognormal mode, with the results from an existing mode fitting algorithm based on the least-square approach (LSQ Algo). The EM Algo achieved significantly less computational time of about 30% relative to the LSQ Algo, it exhibited less uncertainties in predicting the correct number of lognormal modes to be fitted, but overestimated the particle mean diameter in the cases that the newly formed size distribution in the Nucleation mode diameters, was partly observed in the SMPS's measuring range.
Using the results obtained by the implementation on our data sets of both mode fitting algorithms, we were able to derive the particle Growth Rate (GR) during a NPF event. The particle Growth Rate at Finokalia was found to vary between 0.44nm • h-1 and 19.40nm • h-1 with a mean value of 5.56nm • h-1 in accordance to locations with similar characteristics. Growth rates were found to be smaller during wintertime and larger in the summer but in order to conclude to a clear trend, further analysis must be done.
|
Σωματιδιακές κατανομές στην ατμόσφαιρα της Ανατολικής Μεσογείου
