Journal cover Journal topic
Atmospheric Chemistry and Physics An interactive open-access journal of the European Geosciences Union
Atmos. Chem. Phys., 16, 15629-15652, 2016
https://doi.org/10.5194/acp-16-15629-2016
© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.
Research article
20 Dec 2016
Insights into the deterministic skill of air quality ensembles from the analysis of AQMEII data
Ioannis Kioutsioukis1,2, Ulas Im3, Efisio Solazzo2, Roberto Bianconi4, Alba Badia5, Alessandra Balzarini6, Rocío Baró12, Roberto Bellasio4, Dominik Brunner7, Charles Chemel8, Gabriele Curci9,10, Hugo Denier van der Gon11, Johannes Flemming13, Renate Forkel14, Lea Giordano7, Pedro Jiménez-Guerrero12, Marcus Hirtl15, Oriol Jorba5, Astrid Manders-Groot11, Lucy Neal16, Juan L. Pérez17, Guidio Pirovano6, Roberto San Jose16, Nicholas Savage15, Wolfram Schroder18, Ranjeet S. Sokhi8, Dimiter Syrakov19, Paolo Tuccella9,10, Johannes Werhahn14, Ralf Wolke18, Christian Hogrefe20, and Stefano Galmarini2 1University of Patras, Department of Physics, University Campus 26504 Rio, Patras, Greece
2European Commission, Joint Research Centre, Directorate for Energy, Transport and Climate, Air and Climate Unit, Ispra (VA), Italy
3Aarhus University, Department of Environmental Science, Roskilde, Denmark
4Enviroware srl, Concorezzo (MB), Italy
5Earth Sciences Department, Barcelona Supercomputing Center (BSC-CNS), Barcelona, Spain
6Ricerca sul Sistema Energetico (RSE) SpA, Milan, Italy
7Laboratory for Air Pollution and Environmental Technology, Empa, Dubendorf, Switzerland
8Centre for Atmospheric & Instrumentation Research, University of Hertfordshire, College Lane, Hatfield, AL10 9AB, UK
9Department of Physical and Chemical Sciences, University of L'Aquila, L'Aquila, Italy
10Center of Excellence for the forecast of Severe Weather (CETEMPS), University of L'Aquila, L'Aquila, Italy
11Netherlands Organization for Applied Scientific Research (TNO), Utrecht, the Netherlands
12University of Murcia, Department of Physics, Physics of the Earth, Campus de Espinardo, Ed. CIOyN, 30100 Murcia, Spain
13ECMWF, Shinfield Park, Reading, RG2 9AX, UK
14Karlsruher Institut für Technologie (KIT), IMK-IFU, Kreuzeckbahnstr. 19, 82467 Garmisch-Partenkirchen, Germany
15Zentralanstalt für Meteorologie und Geodynamik, ZAMG, 1190 Vienna, Austria
16Met Office, FitzRoy Road, Exeter, EX1 3PB, UK
17Environmental Software and Modelling Group, Computer Science School – Technical University of Madrid, Campus de Montegancedo – Boadilla del Monte, 28660 Madrid, Spain
18Leibniz Institute for Tropospheric Research, Permoserstr. 15, 04318 Leipzig, Germany
19National Institute of Meteorology and Hydrology, Bulgarian Academy of Sciences, 66 Tzarigradsko shaussee Blvd., 1784 Sofia, Bulgaria
20Atmospheric Modelling and Analysis Division, Environmental Protection Agency, Research, Triangle Park, USA
Abstract. Simulations from chemical weather models are subject to uncertainties in the input data (e.g. emission inventory, initial and boundary conditions) as well as those intrinsic to the model (e.g. physical parameterization, chemical mechanism). Multi-model ensembles can improve the forecast skill, provided that certain mathematical conditions are fulfilled. In this work, four ensemble methods were applied to two different datasets, and their performance was compared for ozone (O3), nitrogen dioxide (NO2) and particulate matter (PM10). Apart from the unconditional ensemble average, the approach behind the other three methods relies on adding optimum weights to members or constraining the ensemble to those members that meet certain conditions in time or frequency domain. The two different datasets were created for the first and second phase of the Air Quality Model Evaluation International Initiative (AQMEII). The methods are evaluated against ground level observations collected from the EMEP (European Monitoring and Evaluation Programme) and AirBase databases. The goal of the study is to quantify to what extent we can extract predictable signals from an ensemble with superior skill over the single models and the ensemble mean. Verification statistics show that the deterministic models simulate better O3 than NO2 and PM10, linked to different levels of complexity in the represented processes. The unconditional ensemble mean achieves higher skill compared to each station's best deterministic model at no more than 60 % of the sites, indicating a combination of members with unbalanced skill difference and error dependence for the rest. The promotion of the right amount of accuracy and diversity within the ensemble results in an average additional skill of up to 31 % compared to using the full ensemble in an unconditional way. The skill improvements were higher for O3 and lower for PM10, associated with the extent of potential changes in the joint distribution of accuracy and diversity in the ensembles. The skill enhancement was superior using the weighting scheme, but the training period required to acquire representative weights was longer compared to the sub-selecting schemes. Further development of the method is discussed in the conclusion.

Citation: Kioutsioukis, I., Im, U., Solazzo, E., Bianconi, R., Badia, A., Balzarini, A., Baró, R., Bellasio, R., Brunner, D., Chemel, C., Curci, G., van der Gon, H. D., Flemming, J., Forkel, R., Giordano, L., Jiménez-Guerrero, P., Hirtl, M., Jorba, O., Manders-Groot, A., Neal, L., Pérez, J. L., Pirovano, G., San Jose, R., Savage, N., Schroder, W., Sokhi, R. S., Syrakov, D., Tuccella, P., Werhahn, J., Wolke, R., Hogrefe, C., and Galmarini, S.: Insights into the deterministic skill of air quality ensembles from the analysis of AQMEII data, Atmos. Chem. Phys., 16, 15629-15652, https://doi.org/10.5194/acp-16-15629-2016, 2016.
Publications Copernicus
Download
Short summary
Four ensemble methods are applied to two annual AQMEII datasets and their performance is compared for O3, NO2 and PM10. The goal of the study is to quantify to what extent we can extract predictable signals from an ensemble with superior skill at each station over the single models and the ensemble mean. The promotion of the right amount of accuracy and diversity within the ensemble results in an average additional skill of up to 31 % compared to using the full ensemble in an unconditional way.
Four ensemble methods are applied to two annual AQMEII datasets and their performance is...
Share