Comparing remote sounding measurements of a variable stratosphere

The measurement of trace gases through remote sounding techniques has led to a better understanding of the processes controlling the structure and variability of the stratosphere. Differences between measurements over space and time are due to atmospheric variability and instrument errors: thus, comparison of measurements can be used to test our knowledge of both.\n\nComparisons of measurements over long time periods are used to identify trends. Balloon-borne infrared emission radiometer instruments have been used to make measurements of midlatitude stratospheric HNO3 spanning a period of twelve years. The timing of the measurements is notable, since they occur before and well after the eruption of Mt. Pinatubo, which significantly perturbed HNO3 levels, complicating prior trend analyses. No significant differences are found between the HNO3 retrievals, although large measurement uncertainties preclude any conclusion concerning trends.\n\nComparisons of measurements that are closely spaced in space and time are useful for satellite validation, where one aims to reduce the effect of atmospheric variability on the estimation of systematic and random errors. A novel technique for the estimation of systematic error, which differentiates between additive and multiplicative bias, is introduced. In a comparison of measurements by the ACE-FTS and Aura MLS instruments, significant multiplicative biases are identified and described.\n\nIn order to validate the reported random errors (RREs) of measurements, satellite validation studies often focus on measurements in the tropical stratosphere, where variability is weak. The scatter in tropical measurements can then be used as an upper limit on instrument precision. In an analysis of tropical measurements by the ACE-FTS, scatter is found to be roughly consistent with the RREs for H2O and CO. The scatter in measurements of O3, HNO3, and N2O, while larger than the reported random errors, is roughly consistent with the variability simulated in the Canadian Middle Atmosphere Model. This work implies that the random error of the ACE-FTS measurements is smaller than the weak natural variability of the tropical stratosphere.