| Summary: | Thesis (Ph.D.)--University of Washington, 2022 Arctic sea ice has undergone rapid declines in recent decades. Given the short satellite record (1979–present), disentangling the relative role of natural variability and anthropogenic forcing on recent declines remains an important unresolved problem. In order to acquire a longer, reliable record of Arctic sea ice we employ data assimilation to combine temperature observations and climate model output. This technique results in fully gridded spatial fields of various climate variables throughout the Instrumental Era (1850-present). Specifically, the goal of this research is to reconstruct Arctic sea ice coverage and thickness on both annual and monthly timescales. We first reconstruct Arctic sea ice on annual timescales using an offline approach where each time step is independent from one another. This work reveals larger declines in total Arctic sea ice extent during the early 20th century (1900–1940) than previously estimated. Next, we build a Linear Inverse Model to forecast Arctic sea ice and other climate conditions on monthly timescales. We find that the Linear Inverse Model is able to skillfully forecast Arctic climate conditions during statistically stable time periods and is thus most useful when used as a model emulator. We then embed the Linear Inverse Model into a data assimilation scheme to produce monthly reconstructions of Arctic climate throughout the Instrumental Era.
|
|---|