• Data analysis and data assimilation of Arctic Ocean observations

      Stroh, Jacob Nathaniel; Panteleev, Gleb; Mölders, Nicole; Weingartner, Thomas; Rhodes, John (2019-05)
      Arctic-region observations are sparse and represent only a small portion of the physical state of nature. It is therefore essential to maximize the information content of observations and bservation-conditioned analyses whenever possible, including the quantification of their accuracy. The four largely disparate works presented here emphasize observation analysis and assimilation in the context of the Arctic Ocean (AO). These studies focus on the relationship between observational data/products, numerical models based on physical processes, and the use of such data to constrain and inform those products/models to di_erent ends. The first part comprises Chapters 1 and 2 which revolve around oceanographic observations collected during the International Polar Year (IPY) program of 2007-2009. Chapter 1 validates pan- Arctic satellite-based sea surface temperature and salinity products against these data to establish important estimates of product reliability in terms of bias and bias-adjusted standard errors. It establishes practical regional reliability for these products which are often used in modeling and climatological applications, and provides some guidance for improving them. Chapter 2 constructs a gridded full-depth snapshot of the AO during the IPY to visually outline recent, previouslydocumented AO watermass distribution changes by comparing it to a historical climatology of the latter 20th century derived from private Russian data. It provides an expository review of literature documenting major AO climate changes and augments them with additional changes in freshwater distribution and sea surface height in the Chukchi and Bering Seas. The last two chapters present work focused on the application of data assimilation (DA) methodologies, and constitute the second part of this thesis focused on the synthesis of numerical modeling and observational data. Chapter 3 presents a novel approach to sea ice model trajectory optimization whereby spatially-variable sea ice rheology parameter distributions provide the additional model flexibility needed to assimilate observable components of the sea ice state. The study employs a toy 1D model to demonstrate the practical benefits of the approach and serves as a proof-of-concept to justify the considerable effort needed to extend the approach to 2D. Chapter 4 combines an ice-free model of the Chukchi Sea with a modified ensemble filter to develop a DA system which would be suitable for operational forecasting and monitoring the region in support of oil spill mitigation. The method improves the assimilation of non-Gaussian asynchronous surface current observations beyond the traditional approach.