Evaluation of NWP results for wintertime nocturnal boundary-layer temperatures over Europe and Finland

Four operational numerical weather prediction (NWP) models were evaluated in winter conditions against (a) synoptic observations in Europe, (b) observations at a 48 m high micrometeorological mast in Sodankyla, northern Finland, and (c) observations at the Helsinki Testbed stations: (i) to evaluate the skills of the models to compute nocturnal 2 m air temperature (T2m) and the temperature inversion; and (ii) to distinguish between the T2m bias and the subgrid-scale spatial variability of T2m. The models were (1) the Integrated Forecast System (IFS) of the European Centre for Medium-Range Weather Forecasts (ECMWF), (2) the High Resolution Limited Area Model (HIRLAM), (3) the Applications of Research to Operations at Mesoscale (AROME) developed by Meteo-France, and (4) the Global Forecasting System (GFS) of the US National Center for Environmental Predictions (NCEP). The results demonstrated a T2m bias increasing with decreasing temperature and strengthening temperature inversion. When a strong temperature inversion was observed in Sodankyla, the models underestimated it, whereas in near-neutral conditions the stratification was overestimated. Comparison of observed and modelled 3 h temperature tendencies showed that the T2m tendency in the models was on average only 1720% of the observed one. The warm bias in T2m forecast in Sodankyla during periods of observed temperature inversion partly resulted from a warm bias in the initial conditions. This was due to problems in data assimilation in IFS and HIRLAM, in initialization in AROME, and in either or both procedures in GFS. In particular, the IFS data assimilation increased the T2m bias. Evaluation of modelled T2m against grid-averaged T2m observed at Helsinki Testbed demonstrated that the T2m model error dominated over the spatial variability of observed T2m. This suggests that over an almost flat terrain horizontal resolution is not a major factor for the accuracy of T2m forecast at low T2m typically associated with temperature inversions. Copyright (c) 2012 Royal Meteorological Society