Et dynamisk studium av stormen Narve - et kaldluftsutbrudd i Finnmark - ved hjelp av observasjoner og numeriske simuleringer

Title: "A dynamical study of the storm Narve - a cold air outbreak in Finnmark - with the use of observations and numerical simulations."During the storm Narve in January 2006 strong winds from land were blowing during six full days in Northern Norway. In addition to strong and cold winds out the fjords and on the lee-side of the mountains, the storm is characterized by the great spatial variations in wind speed and direction. In this Master’s thesis the dynamical phenomena that initialized these unusual wind situations during Narve in Finnmark are studied in detail. This is accomplished by comparing wind observations from the ground stations of the Norwegian Meteorological Institute and wind speed derived from satellite measurements named SAR (Synthetic-aperture radar) with numerical simulations from a mesoscale numerical weather prediction model entitled «Fifth-Generation Penn State/NCAR Mesoscale Model» (MM5). MM5 is run with different horizontal resolution with 9 km, 3 km, and 1 km, respectively, different vertical resolution with 36 and 23 vertical levels, respectively, and two different boundary layer schemes. In addition, sensitivity tests are performed in which the sea-surface temperature is set constant and equal to + 5 degrees C, and -33 degrees C in order to study the effect that a change in the temperature difference between the sea and the inland has on the wind speed in the fjords during Narve. The winds in Rognsundet and Sørøysundet are studied in more detail, e.g. by making the island Seiland totally flat (z = 0 m) in order to find the effect that such a change in topography has on the winds in these sounds. The study as a whole shows that wind phenomena like mountain waves with downslope windstorms, gap winds out the fjords, and wakes upstream of the mountains or in connection with hydraulic jumps, are explaining the great spatial variation in wind in a weather situation in which the large-scale geostrophic wind is more or less constant. The fact that the 1 km model has lower verification scores than the 3 and 9 km model when the simulated winds are verified against the wind observations on the regular stations, is first and foremost caused by the erroneous representation in the MM5 model of the downslope windstorms at the places where the stations are located. On the other hand, the verification against the winds from the SAR instrument and vertical crossections from the 1 km model with 36 vertical layers demonstrate that the gap winds in