Over het archief
Het OWA, het open archief van het Waterbouwkundig Laboratorium heeft tot doel alle vrij toegankelijke onderzoeksresultaten van dit instituut in digitale vorm aan te bieden. Op die manier wil het de zichtbaarheid, verspreiding en gebruik van deze onderzoeksresultaten, alsook de wetenschappelijke communicatie maximaal bevorderen.
Dit archief wordt uitgebouwd en beheerd volgens de principes van de Open Access Movement, en het daaruit ontstane Open Archives Initiative.
Basisinformatie over ‘Open Access to scholarly information'.
Reconstructing Arctic Sea Ice in the Common Era
Brennan, M.K. (2019). Reconstructing Arctic Sea Ice in the Common Era. MSc Thesis. University of Washington: Seattle. 57 pp.
|  |
| Beschikbaar in | Auteur |
|
Documenttype: Doctoraat/Thesis/Eindwerk
|
| Trefwoorden |
Data assimilation
Marien/Kust |
| Author keywords |
paleoclimate, Arctic sea ice |
| Abstract |
Arctic sea ice concentrations have undergone rapid declines in recent decades. Many factors have been shown to contribute to this decline, and much of it has been attributed to greenhouse gas forcing and natural variability. In order to understand the relative roles of these factors on Arctic sea ice decline, a longer record of spatially complete data is needed. This project employs data assimilation to combine climate model output and proxy records to reconstruct past climate fields using the Last Millennium Reanalysis (LMR) framework, resulting in spatially complete gridded fields with annual resolution over the last two millennia. First the use of the LMR framework to reconstruct Arctic sea ice concentrations is tested through two methods: pseudo proxy experiments and comparing real proxy reconstructions to other records. Pseudo proxy results indicate strong performance in reconstructing Arctic sea ice. Correlation coefficients between the true and reconstructed values range between 0.63 and 0.77 depending on the climate model output used in the assimilation. The total Arctic sea ice extent reconstructed with the LMR using real proxy data compare well with satellite observations with correlation coefficients ranging between 0.54 and 0.84 depending on the climate model data used in the assimilation. These reconstructions were also compared to other records that precede satellite data and the LMR reconstructions show larger and longer lasting sea ice decline in response to early 20th century warming. The total sea ice extent minimum observed in these reconstructions between 1920-1960 is similar to the values observed in the 1990s. Next, two major questions are investigated using the 2000-year Arctic sea ice reconstruction: (1) Are the current sea ice changes unprecedented? and (2) Does sea ice respond to volcanic eruptions? The first is investigated through examining the distribution of total Arctic sea ice extent. The results indicate that both the trends and values of sea ice extent observed in the satellite era are unprecedented with respect to the LMR reconstruction of the last 1000 years. The second question is investigated through a composite average of sea ice extent before and after the 23 largest volcanic eruptions. These results show a statistically significant increase in total Arctic sea ice extent one year after an eruption. Finally, the sensitivity of these reconstructions to proxy location, model prior and the number of prior ensemble members used in the data assimilation scheme is examined. These experiments indicate that the proxies in the Arctic region (above 60N) explain most of the variance in the LMR sea ice reconstructions as expected. The role of the model prior is investigated by comparing the covariance between surface air temperature and Arctic sea ice across four different models. Overall the covariance is very similar except in some isolated regions in central Russia and China, in the North Atlantic and the central Pacific. In these regions the correlation coefficient is positive, but the coefficient of efficiency is negative indicating that there is a difference in the mean or variance in the covariance across models. 200 ensemble members are found to be sufficient in representing the variance in the full 1000 year last millennium model runs. |
IMIS is ontwikkeld en wordt gehost door het VLIZ.
