Toohey, Matthew , Strong, K., Bernath, P. F., Boone, C. D., Walker, K. A., Jonsson, A. I. and Shepherd, T. G. (2010) Validating the reported random errors of ACE-FTS measurements. Journal of Geophysical Research: Atmospheres, 115 (D20). D20304. DOI 10.1029/2010JD014185.

Preview	Text Toohey_et_al-2010-Journal_of_Geophysical_Research__Atmospheres_(1984-2012).pdf - Published Version Download (685kB) | Preview
	Text jgrd16508-sup-0001-t01.txt - Supplemental Material Download (502B)
	Text jgrd16508-sup-0002-t02.txt - Supplemental Material Download (428B)

Abstract

In order to validate the reported precision of space-based atmospheric composition measurements, validation studies often focus on measurements in the tropical stratosphere, where natural variability is weak. The scatter in tropical measurements can then be used as an upper limit on single-profile measurement precision. Here we introduce a method of quantifying the scatter of tropical measurements which aims to minimize the effects of short-term atmospheric variability while maintaining large enough sample sizes that the results can be taken as representative of the full data set. We apply this technique to measurements of O(3), HNO(3), CO, H(2)O, NO, NO(2), N(2)O, CH(4), CCl(2)F(2), and CCl(3)F produced by the Atmospheric Chemistry Experiment-Fourier Transform Spectrometer (ACE-FTS). Tropical scatter in the ACE-FTS retrievals is found to be consistent with the reported random errors (RREs) for H(2)O and CO at altitudes above 20 km, validating the RREs for these measurements. Tropical scatter in measurements of NO, NO(2), CCl(2)F(2), and CCl(3)F is roughly consistent with the RREs as long as the effect of outliers in the data set is reduced through the use of robust statistics. The scatter in measurements of O(3), HNO(3), CH(4), and N(2)O in the stratosphere, while larger than the RREs, is shown to be consistent with the variability simulated in the Canadian Middle Atmosphere Model. This result implies that, for these species, stratospheric measurement scatter is dominated by natural variability, not random error, which provides added confidence in the scientific value of single-profile measurements.

Actions (login required)

View Item