Diurnal temperature range and cloud cover in the Nordic countries: observed trends and estimates for the future

A simple regression model relating local monthly mean diurnal temperature range (DTR) and cloudiness at 10 synoptic stations in the Nordic region to monthly mean large-scale tropospheric flow has been constructed. The tropospheric flow data, which is input to the model, includes both 500 hPa height and 5001000 hPa thickness fields which are compressed via an EOF-technique. The regression coefficients are calculated from a 27 year record of analysed flow and locally observed DTR and cloudiness. gression model has been used to down-scale the large-scale flow produced in a CO2 sensitivity experiment with a coupled ocean-atmosphere climate model (ECHAM-1). The down-scaled “high CO2” DTR is generally smaller than found when down-scaling the corresponding climate model control experiment. The opposite statement applies to cloudiness. DTR-anomalies of the order 2 Kelvin (K) are obtained in the central/northern parts of Fenno-Scandia in the middle of the 21st century while anomalies much closer to zero are found over the Iceland/Southern Greenland region. Consistent with this picture the cloudiness anomalies in the high CO2 case are predicted highest near the central/northern parts of Fenno-Scandia. sults presented here indicate that periods observed to be anomalously warm on a large-scale — corresponding to high CO2 concentrations — are also observed to be more cloudy and it is argued that clouds may have a thermostatic effect. e calculated DTR and cloud cover trends at the 10 stations used in the regression model in order to assess if they are consistent with the predicted estimates and as such reflect a forcing of the climate system. We find that none of the stations show DTR/cloud cover trends significantly different from zero in the period 1961–1987. We do, however, see a negative DTR trend and a positive cloud trend when we consider a slightly longer time series (1950–1992) at Reykjavik. All statements about trends are, however, very uncertain because the observed records in addition to any trends reflect a large natural variability on decadal time scales as well as possible observational inhomogeneities.