Book Description

Precipitation from atmospheric fronts accounts for a significant portion of the total precipitation in the mid-latitudes, with some locations receiving the majority of their precipitation from atmospheric fronts. In addition, a significant proportion of extreme precipitation events coincide with a frontal passage in the mid-latitudes, and some of these events lead to extreme flooding which can have important and costly socio-economic consequences. Climatological studies regarding both atmospheric fronts and precipitation frequently use global reanalysis datasets due to their cohesive record of many atmospheric variables over a temporal range of generally 40 years or longer. Differences among these reanalyses regarding observations assimilated, atmospheric model used, and grid size contribute to differences in regional precipitation accumulations and the structure and frequency of identified atmospheric fronts. It is therefore important to understand how frontal precipitation is represented in global reanalysis datasets. As much of the literature on atmospheric fronts and frontal precipitation is limited to the use of a single global reanalysis or regional model, this thesis seeks to investigate the consistency among frontal identification and frontal precipitation within multiple reanalyses. The following reanalyses were used based on data availability, spatial and temporal resolution, and use within the literature: ERA-20C, ERA-40, ERA-Interim, ERA5, JRA-55, MERRA-2, NCEP-CFSR, and NOAA20C V2C. Two satellite precipitation products, CMORPH and TRMM, were also used for comparison of frontal precipitation with ERA5. There are numerous methods to identify atmospheric fronts that rely on different parameters involving temperature and/or wind. While different front diagnostics give similar results geographically in terms of frequency and structure, each diagnostic has its own strengths and weaknesses. As the choice of front diagnostic has been shown to result in differences in frontal frequency when using the same reanalysis, two different front diagnostics are used. Results show that reanalyses with a finer grid spacing (id est less than 0.5o x 0.5o) contain a 200% increase in globally averaged total mean frontal frequency for one diagnostic and a 450% increase in globally averaged total mean frontal frequency for the other diagnostic compared to reanalyses with coarse grid spacing (id est 2.0o x 2.0o). Results also show that reanalyses with a finer grid spacing see a 150% increase in globally averaged total mean frontal precipitation proportion for one diagnostic and a 460% increase in globally averaged total mean frontal precipitation proportion compared to coarser grid-spaced reanalyses. The largest differences between reanalyses in both frontal frequency and frontal precipitation proportion exist in the tropics for both diagnostics, a region which is typically considered to have few fronts. Differences among reanalyses regarding both frontal frequency and frontal precipitation proportion are indicated to be strongly related to the differing grid spacings of each reanalysis. To account for differing grid spacings, the reanalyses and satellite precipitation products are regridded to the same coarser grid spacing. Then both diagnostics and their frontal precipitation are recalculated on this common grid to attribute differences in both frontal frequency and frontal precipitation to either differing grid spacings between reanalyses or differences inherent to reanalyses. The regridded reanalyses have much more consistency regarding both frontal frequency and frontal precipitation proportion. Globally, 30% of the difference in the means of total mean frontal frequency of all eight reanalyses is attributed to grid size for one diagnostic, while 59% of the difference is attributed to grid size for the other diagnostic. Allocation of frontal precipitation closely follows the frequency of identified fronts. Globally, 28% of the difference in the means of total mean frontal precipitation proportion of all eight reanalyses is attributed to grid size for one diagnostic while 61% of the difference is attributed to grid size for the other diagnostic. Both diagnostics show that the percent difference in frontal frequency and frontal precipitation proportion is highly dependent on geographical area. Objective frontal identification and frontal precipitation proportion is highly dependent on the choice of diagnostic, the region under consideration, the grid spacing of the reanalysis, and the reanalysis or reanalyses used. These results strongly suggest that research regarding both frontal identification and frontal precipitation should use more than one reanalysis and diagnostic.