On the seasonal predictability of Arctic sea ice

• Main Author: Chevallier, Matthieu
• Other Authors: Centre d'Enseignement et de Recherche en Environnement Atmosphérique (CEREA), École des Ponts ParisTech (ENPC)-EDF R&D (EDF R&D), EDF (EDF)-EDF (EDF), Université Paris-Est, Marc Bocquet
• Format: Doctoral or Postdoctoral Thesis
• Language: French
• Published: HAL CCSD 2012
• Subjects: Sea ice; Arctic; Seasonal forecasting; Climate predictability; Glace de mer; Arctique; Prévisions saisonnières; Prévisibilité; Océan; [SDU.STU]Sciences of the Universe [physics]/Earth Sciences; Arctic Ocean; Arctique*
• Online Access: https://pastel.hal.science/pastel-00806125; https://pastel.hal.science/pastel-00806125/document; https://pastel.hal.science/pastel-00806125/file/TH2012PEST1117_complete.pdf

Sea ice experiences some major changes in the early 21st century. The recent decline of the summer Arctic sea ice extent, reaching an all-time record low in September 2012, has woken renewed interest in this remote marine area. Sea ice seasonal forecasting is a challenge of operational oceanography that could benefit to several stakeholders : fishing, energy, research, tourism. Moreover, sea ice is a boundary condition of the atmosphere. As such, as tropical sea surface temperature, it may drive some atmosphere seasonal predictability. The goal of this PhD work was to set up a dedicated Arctic sea ice seasonal forecasting system, using CNRM-CM5.1 coupled climate model. We address the initialization strategy, the creation and the evaluation of the hindcasts (or re-forecasts). In contrast to sea ice concentration, very few thickness data are available over the whole Arctic ocean. In order to initialize sea ice and the ocean dynamically and thermodynamically, we used the ocean-sea ice component of CNRM-CM5.1, named NEMO-GELATO, in forced mode. The initialization run is a forced simulation driven by ERA-Interim forcing over the period 1990-2010. Corrections based on satellite data and in-situ measurements leads to skilful simulation of the ocean and sea ice mean state and interannual variability. Sea ice thickness seems overall underestimated, based on the most recent estimates. Some characteristics of sea ice inherent predictability are then addressed. A diagnostic potential predictability study allowed us to identify two regimes of predictability using sea ice volume and the ice thickness distribution. The first one is the 'persistence regime', for winter sea ice area. March sea ice area is potentially predictable up to 3 months in advance using simple persistence, and surface covered by thin ice to a lesser extent. The second one is the 'memory regime', for summer sea ice area. September sea ice area is potentially predictable up to 6 months in advance using volume and to a greater extent the area covered by ...