Gillespie, Ian (2021) Investigating the potential use of sparse-input reanalyses to homogenise long-term land surface air temperature records. PhD thesis, National University of Ireland, Maynooth.

Preview

Download (10MB) | Preview

Abstract

The correction of meteorological observational records (homogenisation) for non climate artefacts is an important task. Very few, long-term meteorological station series are entirely free of non-climatic influences. Climate data homogenization aims to identify and remove these non climate factors. Numerous methods of homogenisation have been developed over the decades. Current state of the art approaches generally proceed using pairwise difference series between observations from a network of reference stations and the station under assessment. Such methods work well in well sampled regions such as Europe and North America, but are less successful in poorly sampled regions and epochs. Reanalyses are produced by assimilating available observations into a forecast model, producing complete fields that are consistent with: the input data, the model physics, and any external boundary conditions prescribed. Full-input reanalyses which assimilate data from all available sources have previously been used to homogenise radiosonde data records. This work sets out to investigate if sparse-input reanalysis products that only assimilate surface pressure and use prescribed sea-ice, sea surface temperatures and changes in atmospheric composition, can act as a suitable reference series for the homogenisation of land surface air temperatures and to compare the results to established methods. It is found that sparse-input reanalysis products have successively improved in their quality with each new generation. The most recent product from NOAA-CIRES – 20CRv3 – has comparable overall statistical properties when interpolated to station locations and differenced to pairwise differences. In well sample regions neighbour-based comparisons remain favourable, but in sparser regions and epochs –20CRv3 may be preferable. The 20CRv3 product is therefore used to identify breakpoints and then 4 distinct approaches are used to adjust the series. Two of these directly use the 20CRv3 fields to estimate adjustments, while the remaining pair use apparently homogeneous neighbour station series wherever possible. The resulting set of estimates show reasonable overall behaviour looking at station series behaviour, spatial anomalies and spatial trends. The thesis highlights the potential for sparse-input reanalysis to provide a substantial methodological degree of freedom in the homogenisation of global land surface air temperature estimates, but further work is required in developing an operational version.

Item Type:	Thesis (PhD)
Keywords:	sparse-input reanalyses; homogenise; long-term land surface; air temperature records;
Academic Unit:	Faculty of Social Sciences > Geography Faculty of Social Sciences > Research Institutes > Irish Climate Analysis and Research Units, ICARUS
Item ID:	14901
Depositing User:	IR eTheses
Date Deposited:	06 Oct 2021 09:55
URI:
Use Licence:	This item is available under a Creative Commons Attribution Non Commercial Share Alike Licence (CC BY-NC-SA). Details of this licence are available here

Repository Staff Only(login required)

Item control page

Downloads

Origin of downloads