『Abstract
　The risks of exceeding EU limit values for NO2 concentrations have increased in many European cities, and compliance depends strongly on meteorological conditions. This study focuses on meteorological conditions and their influences on urban background NO2 concentrations in the city of Gothenburg for 1999-2008. The relations between observed NO2 concentrations and meteorological conditions are constructed using two modelling approaches: multiple linear regression and synoptic regression. Both approaches assume no trends in emissions over the study period. The multiple linear regression model is established on observed local meteorological variables. The synoptic-regression model first groups days according to synoptic conditions using Lamb Weather Types and then uses linear regression on each group separately. A model comparison shows that linear regression model and synoptic-regression model perform satisfactory. The synoptic-regression model gives higher explained variance (R²) against observations during the calibration years (1999-2007), in particular for the morning peak and afternoon-evening peak concentrations, but the improvement in the validation period is weak. The annual mean NO2 variations, and their trends during the study period, were assessed using the synoptic-regression model. The synoptic-regression model is able to explain 54%, 42% and 80% of the annual variability of daily mean, morning peak and afternoon-evening peak NO2 concentrations, respectively. The observed and modelled annual means of the daily mean and morning/afternoon-evening peak NO2 concentrations show decreasing trends from 1999 to 2008. All trends, except the trend in annual-average observed morning peak NO2 are statistically significant. The presence of trends in the modelled NO2 concentrations - even though emissions are assumed to be constant - leads us to conclude that weather and climate alone are responsible for a substantial fraction of the recent declines in observed NO2 concentrations in Gothenburg. Favourable meteorological conditions may have mitigated increases in local NO2 emissions during 1999 to 2008.

Keywords: NO2 concentrations; Dispersion conditions; Statistic downscaling; Linear regression model; Synoptic-regression model; Gothenburg』

1. Introduction
2. Data and methods
　2.1. Data
　　2.1.1. Air quality and local meteorological data
　　2.1.2. Lamb weather classification
　2.2. Methods
　　2.2.1. Multiple linear regression method
　　2.2.2. Synoptic-regression method
　　2.2.3. Predictor selection
　　2.2.4. Weekly emission index
　　2.2.5. Model evaluation measures
3. Results
　3.1. Seasonal, weekly and diurnal cycles
　3.2. Model comparison
　3.3. Assessment of annual mean NO2 variations and trends
4. Discussion and conclusions
Acknowledgements
References