Validation of Uncertainties of Ozone Profiles Retrieved from Envisat Limb Sensors
Sofieva, Viktoria1; Laeng, Alexandra2
1Finnish Meteorological Institute, FINLAND; 2Karlsruhe Institute of Technology, GERMANY
It is well appreciated that correct uncertainty estimates are nearly as important as data themselves. Validation of uncertainty estimates is important for remote sensing measurements, which retrieve the information about atmospheric parameters via solving an inverse problem: due to omissions in the atmospheric models used in retrievals, uncertainty estimates can be not perfect.
In this work, we present extensive validation of uncertainties (random error estimates) on ozone profiles for three sensors on board ENVISAT: GOMOS, MIPAS and SCIAMACHY. First, the uncertainty estimates were evaluated for each instrument separately, using closely collocated data. In this analysis, the sample variance of profile difference is compared with that predicted by uncertainty estimates. Then, in the regions of low ozone natural variability, inter-satellite comparisons of difference in sample variance with difference in uncertainty estimates were performed.
The developed method has allowed extensive validation of uncertainty estimates and identification of shortcomings. The inter-satellite comparisons have confirmed the conclusions made from separate-instrument analyses and thus enhanced their confidence.