|Titel:||Short-range sea ice forecast with a regional coupled sea-ice–atmosphere–ocean model||Sonstige Titel:||Kurzfristvorhersage von Meereis mit einem regionalen gekoppelten Meereis–Atmosphäre–Ozean-Modell||Sprache:||Englisch||Autor*in:||Gierisch, Andrea M. U.||Schlagwörter:||sea ice; modelling; forecast; shipping; Arctic||GND-Schlagwörter:||Meereis; Modell; Prognose; Schifffahrt; Arktis||Erscheinungsdatum:||2014||Tag der mündlichen Prüfung:||2015-01-30||Zusammenfassung:||
Connected with climate change, sea ice in the Arctic is reducing. This opens new possibilities for ship traffic, for instance along the Northern Sea Route. For navigators to find the best route through these partly ice-covered waters, forecasts of the sea ice conditions for a few days are required. For this purpose, a short-range sea ice forecast system, HAMMER, is set up and some relevant features are discussed in this thesis. In order to determine which physical processes have to be considered in a short-range model, the relevance of processes that affect sea ice is investigated. To do so, the impact-timescale is calculated which indicates how quickly a process can affect the target property of sea ice, which here are ice drift, ice concentration, and ice thickness. Furthermore, the variability of the process’ effect is evaluated by a newly developed measure: the update-timescale. This indicates how frequently the process has to be updated (i.e. recalculated) in a numerical model. The results reveal that some processes like the lateral melt of ice floes can be neglected for short-range forecasts. Moreover, most processes have to be updated only every 30 minutes or even less frequently. To correctly simulate the interaction processes of sea ice with its surrounding, HAMMER applies a coupling to an ocean model as well as to an atmosphere model. The setup of this system is presented, in combination with two numerical optimisations that reduce the computational costs. A) A time-split approach in the atmosphere model METRAS decouples the calculation of cloud-microphysical processes from the main model time step. Thus, the time step can be increased during precipitation events, which yields a speed-up of 10%. B) A new algorithm to solve the ice drift equation is developed. By enhanced coupling of the ice drift components u and v during the iterative procedure, numerical instabilities can be avoided. Because of the resultant decrease of required iterations the amount of computational time required by the sea ice model relative to the atmosphere model is reduced from 50% to 5%. The simulation results remain quasi-unchanged for A) and B). The optimised model system, HAMMER, was applied to operationally forecast sea ice conditions in the Barents Sea in March 2014. The simulated ice concentration is evaluated using hit rates. They reveal that HAMMER performs worse than a persistence forecast, especially in regions with high ice concentration. The benefit of HAMMER is shown by a new evaluation technique that addresses the navigability of randomly chosen ship routes. Even though HAMMER forecasts navigable routes to be non-navigable more often than persistence does, it can reduce danger for some types of ships because it can better forecast non-navigable routes.
|URL:||https://ediss.sub.uni-hamburg.de/handle/ediss/5919||URN:||urn:nbn:de:gbv:18-74778||Dokumenttyp:||Dissertation||Betreuer*in:||Schlünzen, K. Heinke (Prof. Dr.)|
|Enthalten in den Sammlungen:||Elektronische Dissertationen und Habilitationen|
Diese Publikation steht in elektronischer Form im Internet bereit und kann gelesen werden. Über den freien Zugang hinaus wurden durch die Urheberin / den Urheber keine weiteren Rechte eingeräumt. Nutzungshandlungen (wie zum Beispiel der Download, das Bearbeiten, das Weiterverbreiten) sind daher nur im Rahmen der gesetzlichen Erlaubnisse des Urheberrechtsgesetzes (UrhG) erlaubt. Dies gilt für die Publikation sowie für ihre einzelnen Bestandteile, soweit nichts Anderes ausgewiesen ist.