The Christmas Storm 2016: Comparing Snow Observations and the Operational Forecast Model MEPS

Previous studies showed the importance to have information about the vertical distribution of precipitation to simulate snow and related cyclone development correctly in regional, mesoscale models. During Christmas 2016, an extreme storm affected the local infrastructure of Eastern, Southern, and Western Norway. In this thesis, the Christmas storm 2016 is investigated for snow observations and the operational forecast model at Haukeliseter (991 m above sea level), Norway. The WMO measurement site Haukeliseter is equipped with conventional meteorological instru- ments and a double fence snow gauge instrument to reduce wind effects and increase catch-ratios for frozen precipitation. In winter 2016/2017, three additional instruments were installed for a US National Science Foundation funded field campaign, to estimate snow water content in the column with the help of the optimal estimation retrieval. In November 2016, the AROME-MetCoOp ensemble prediction system (MEPS) became operational at the Norwegian Meteorological Institute. In this thesis, the extreme weather event is studied using European Centre for Medium-Range Weather Forecasts weather analysis, meteorological measurements, including double fence gauge and radar observations, optimal estimation retrieval, and MEPS. During 21 and 26 December 2016, two cyclones as well as frontal passages affected Norway. Ob- served frozen and liquid precipitation is associated with the cyclones and the fronts. The met- eorological analysis of surface properties from observations and MEPS forecasts agree on the passages of occlusions and warm sector. Wind speeds and surface precipitation amount are pre- dicted too high by MEPS (mean absolute error: up to 10 ms−1 and 15 mm) during the entire event with westerlies revealing a better agreement with observations than south-easterlies. A sensitivity study of the optimal estimation retrieval shows the advantage of using the Multi-Angular Snow- fall Camera to choose the correct particle habit. During the Christmas 2016 storm, the average difference between the double fence gauge observations and the retrieved surface amount for assumed rimed aggregates is less than −5 % for 12 h and 24 h surface snow accumulation. With longer lead time the average difference decreases between double fence gauge observations and forecasted precipitation amount for 12 h and 24 h accumulation (+135 % and +33 %). However, for 24 and 26 December 2016, the surface precipitation amount is predicted too high compared to double fence gauge observations (+60 %). Liquid precipitation was observed at Haukeliseter in the afternoon on 25 December 2016. MEPS initialisations 24 h and 48 h prior successfully simulate the thickness and duration of the liquid layer in the lower most atmosphere, but it predicts less snow water content (≤ 1.2 gm−3 ) than the profiles of retrieved snow profiles (≤ 1.5 gm−3 ). Local topography effects by the surrounding mountains lead to continuous snow patterns during strong westerlies and weak south-easterlies, show high amount of snow water content with a pulsing pattern. Finally, orography impacts on snow observations and model forecast are discussed.