Impact of atmospheric variability on validation of satellite-based temperature measurements

Satellite validation is often based on straight forward comparison of satellite-based data with non-satellite based measurements. For functional reasons satellite and reference measurements do usually not correspond exactly in time and space. Dynamical effects in the atmosphere lead to temporal and spatial variability of atmospheric parameters (e.g. temperature). This causes considerable differences that do not necessarily hint to an incorrect satellite measurement, so called mistime and misdistance errors. In this paper, the natural variability of the atmosphere is studied on scales effecting validation measurements. The approach is applied to temperature data from the ERA-40 reanalysis as well as to radiosonde (SIGMA-1) and satellite-based (SABER) measurements. Mistime and misdistance errors are quantified in dependence of geographic position, altitude, season and the temporal and spatial mismatch. The results allow a quantitative estimation of the impact of natural variability on validation analyses. In general, values lie in the range of a few Kelvin (e.g. up to 5 K for 500 km misdistance or 6 h mistime in the stratosphere), which indicates considerable effects on validation results. The determined results also point out regions in the atmosphere where the impact of natural variability is in general relatively high (e.g. the winter stratosphere in mid-latitudes) or rather low (e.g. the lower summer stratosphere). Altitudes, which are characterized systematically by only small mismatch errors, are indicated at about 10 and 25 km, respectively. These quiet layers are of special interest for validation activities. (C) 2013 Elsevier Ltd. All rights reserved.