Search

The research activity presented in this manuscript deals with the implementation of a methodology to merge in an optimal way atmospheric modelling and observations at different spatial scales. In particular, we approached the problem of assimilation of ground measurements and satellite columnar data and how the Data Assimilation (DA) could improve the chemical transport model (CTMs) and correct biases and errors in the chemical species forecast. The work focused on tropospheric ozone and the species linked to its formation, since they play a crucial role in chemical processes during photochemical pollution events. The study was carried out implementing and applying an Optimal Interpolation (OI) DA technique in the air quality model BOLCHEM and the CHIMERE CTM. The OI routine was chosen because it has given satisfactory results in air quality modelling and because it is relatively simple and computationally inexpensive. In the first part of the study we evaluated the improvement in the capability of regional model BOLCHEM to reproduce the distribution of tropospheric pollutants, using the assimilation of surface chemical observations. Among the many causes of uncertainties of CTMs simulations, a particular focus is given by uncertainties in emissions, that are known to be high. The scientific purpose was to analyse the efficacy of DA in correcting the biases due to perturbed emission. The work was performed using an Observing System Simulation Experiment (OSSE), which allowed the quantification of assimilation impact, through comparison with a reference state. Different sensitivity tests were carried out in order to identify how assimilation can correct perturbations on O3, induced by NOx emissions biased in flux intensity and time. Tests were performed assimilating different species, varying assimilation time window length and starting hour of assimilation. Emissions were biased quantitatively up to ± 50% and shifted temporally up to ± 2 hours. The analysis brought to the conclusions that NO2 assimilation significantly improves O3 maxima during the assimilation, making it almost independent on different emission scenarios. The assimilation impact lasts up to 36-40 hours after the end of the assimilation window. This is a considerable result, especially when it is taken into account that DA generally yields significantly better forecasts in the 6-12 hours range, but improvements vanish afterwards. The NO2 night-time chemistry has the role of maintaining the correction of O3 due to assimilation also in the following day. Assimilating NO2 and O3 simultaneously bring to rather better results, although the benefit lasts only a few hours after the end of the assimilation window. It was found that the best results are achieved assimilating observations during the photochemically active period (06-18 UTC). It was also found that temporally biased NOx emissions only slightly perturb O3 concentration during the photochemically active regime, while the perturbation is larger during night-time. Assimilation has a very low impact during the assimilation window and a negligible impact after its end. The second part of PhD research activity dealt with the evaluation of the impact of assimilation of satellite NO2 tropospheric columns on the distribution of pollutants at the ground level during photochemical pollution events at continental scale. In particular, we focused on the assimilation of observations from SCIAMACHY and from OMI, and its effect on ozone in the lowermost troposphere in Europe. For an effective improvement in assimilated fields it is particularly important the consistency between satellite and model resolution. SCIAMACHY and OMI have a considerable difference in spatial and temporal resolution, allowing to test the role of data resolution on the effectiveness of assimilation. The role of data resolution on the effectiveness of assimilation was investigated also changing the model resolutions. It was found the perturbation on NO2 field due to assimilation causes a modification on ozone field that appears more spatially variable and higher in some photochemical polluted areas. Similar effects are detected both for SCIAMACHY and OMI assimilation. Significative effects of assimilation on ozone can be appreciate in polluted areas at local scale. Focusing on specific subdomains, it was found that the effect of assimilation lasts, in general, 8 hours and in few cases until the reactivation of active photochemical period in the following day. This is a strong impact, considering that assimilation is performed at most once a day and it is probably linked to the model underestimate of ozone and its precursors in polluted areas with respect to those measured by SCIAMACHY and OMI. In wide and highly polluted areas assimilation achieves satisfactory results, comparing simulated ground ozone with independent ground measurements. In that region where OMI assimilation in the coarse and fine resolution simulations and SCIAMACHY assimilation were confronted, we could conclude that these different assimilation set-up are almost similar. Whereas, in more localised polluted areas (i.e. comparable to model and satellite resolution), OMI assimilation in the finer resolution simulation performs better with respect to OMI assimilation in the coarse resolution simulation and SCIAMACHY assimilation. As a general conclusive statement, assimilation can be an important tool to make the spatial and temporal distribution of pollutants more realistic and closer to the specific local differences with the caveat of horizontal resolution of the assimilated columns and model simulations.