Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Surrounded by mountains and several continents and affected by different types of pollution, the Mediterranean Sea is a natural laboratory for studying the variability of the chemical composition in the lower atmospheric layers and the interaction between pollutants from distant regions. The Chemistry-Aerosol Mediterranean Experiment (ChArMEx, http://charmex.lsce.ipsl.fr) is a French initiative which set up an international coordinated experimental and modelling effort based on the most updated tools, for an assessment of the present and future state of the atmospheric environment in the Mediterranean Basin, and of its impact on the regional climate, air quality, and marine biogeochemistry. The target is short-lived particulate and gaseous tropospheric trace species which are the cause of poor air quality events, have two-way interactions with climate, and impact the marine biogeochemistry, in a context of strong regional anthropogenic and climatic pressures.

Atmospheric aerosols in the north-western Mediterranean region are mainly composed by sea salt, anthropogenic pollution (from industrialized high populated coastal cities) and Saharan dust aerosols. The present study focuses on the analysis of the model capability to properly simulate long-range Saharan dust transport during summer 2012 in the Western Mediterranean. Moreover, the use of ensemble forecasts to improve event predictability is analysed.

2 Methods

The present study uses 9 regional dust models that are run at different centres in Europe which are presented in Table 13.1. All models produced 24 h forecasts beginning at 0UTC (except RegCM and Aladin-CCM, which are not forced by meteorological fields every 24 h) in order to investigate their performance and ability to predict the dust outbreaks over the Mediterranean from 1st June to 31st August 2012. Various model output fields at 3-hourly resolution were processed for comparative purposes. The research teams at the modelling centres configured their own model experiments. Therefore, the horizontal and vertical resolution, domain size, driving boundary conditions, dust field initial conditions, land surface conditions, dust emission and deposition parameterizations as well as dust physical and optical properties vary between the models. Some of these features are summarised in Table 13.1.

Table 13.1 Summary of the main features of the 9 regional models included in the present contribution
Full size table

The numerical weather prediction and European air quality communities have for some time been exploring the potential of multi-model ensembles to enhance understanding of forecast uncertainty. Similarly, a multi-model median product is also generated from the set of different predictions of the dust models interpolated to a common grid mesh.

The model outputs are compared against a variety of both ground-based and airborne in situ and remote sensing measurements performed during the pre-ChArMEx/TRAQA field campaign (Attié et al. 2014). The model outputs are also compared with satellite aerosol products, which provide a description of the spatial aerosol distribution over the basin. These observational datasets provide a complete set of exceptional quantitative constraints for model simulations of this period.

3 Results and Discussions

The aerosol optical depth (AOD) model intercomparison for summer 2012 (1st June–31st August) shows that the intensity and the spread of the Saharan dust intrusion over Europe reproduced by the models present large differences. Not only do dust emission schemes rely on various assumptions, but also their implementation within a regional model presents challenges. Significant biases are found between different regional models even when they have the same dust emission scheme implemented. Following (Basart et al. 2009), the dust AOD comparison against 4 AERONET sun photometers (Barcelona, Palma de Mallorca, Avignon and Lampedusa) shows that the set of models are able to reproduce the seasonal AOD variability with correlations >0.6 except for the climate model RegCM (not shown here). In this respect, the multi-model median product shows the best results (with correlation >0.8).

On late June 2012, a strong dust outbreak (with origin over Algeria) was observed over the Western Mediterranean coinciding within the intensive ChArMEx/TRAQA experimental campaign (Dulac et al. 2013; see Fig. 13.1). High aerosol concentrations were observed between 1 and 5 km by CALIPSO and LOAC flights from Martigues (France) meanwhile desert dust was predicted in altitudes <5 km (above sea level, a.s.l; see Fig. 13.1) with maximum dust concentrations at ~3.5 km (not shown here). Aerosols at high altitudes (between 3 and 7 km a.s.l.) were not predicted by the dust models. Observations and air-mass backward trajectories confirmed that these upper aerosols were associated to forest fires with origin in the south-eastern Iberian Peninsula as confirmed by.

Fig. 13.1
figure 1

Satellite and ground-based observations as well as backtrajectories at Martingues (at arrivals at 500, 1500 and 3000 m a.s.l.) collected during 29th June 2012

Full size image

These preliminary results demonstrate that the dust models are useful to complement dust-related observations over the total aerosol load, understand the dust processes and predict the impact of dust. Moreover, it is shown the potential use of multi-model ensembles to enhance understanding of forecast uncertainty.