ESA / ECSAT CCIWV.REP.022 Water Vapour Climate Change Initiative (WV_cci) - CCI+ Phase 2 D2.2 For each instrument ( i ), latitude grid, and pressure level, a mean offset between the observed and modelled water vapour time series is calculated as an average over the full mission (i.e. without time-dependency) according to [ ] = 〈[ ] ] (Equation 1) where i denotes a given instrument and CCM a chemistry–climate model nudged to observed meteorology. In a second step, each time series is bias-corrected with respect to a reference (here calculated as the average of Aura-MLS, MIPAS, and ACE-FTS) using the biases calculated according to Equation 1 for both instrument i and reference Ref.
The resulting bias-corrected water vapour concentration for each instrument ( [ ] _ ) can thereby be expressed as: [ ] _ = [ ] − d[ ] + 嬠 ] (Equation 2) In a third step, the bias-corrected water vapour time series are being scanned for unphysical values (that is negative values), detected outliers are filtered out, and the remaining data points are merged to a bias-corrected monthly mean using an optimal estimation approach to merge the multi-instrument data points ( merged ): [ ] [ ] ) # (Equation 3) where N denotes the number of instruments available, and is the weighting factor for each instrument.
Currently, is calculated as the moment of each instrument’s bias- corrected differences (seen in Figure 2-2, panel c). As shown in Figure 2-2, the biases between the individual instrument time series have been removed after step 2. Each step in the procedure has gone through visual inspection so to ensure correct functioning of the merging algorithm, which has been continuously automated. The implementation of an optimal estimation approach for the merging is the most innovative improvement of the merging methodology put forward by Hegglin et al. (2014) [RD-1] and provides a first essential issues that may affect anomalies.
However, these points are being further investigated during step towards dealing with inconsistent seasonal cycle amplitudes, outliers, and sampling WV_cci Phase 2. 10