ESA / ECSAT CCIWV.REP.022 Water Vapour Climate Change Initiative (WV_cci) - CCI+ Phase 2 D2.2 VRWV biases between each satellite instrument and BBH observations are calculated using a quantile-mapping technique for each 1 km altitude bin within the -6 to +6 km range. For each altitude bin in each month, the biases are computed through the cumulative distribution function (CDF) for the WV measurements as: &,() = * ++, # [ * & ( & )] , (Equation 4) where the * ++, # is the inverse CDF for BBH observations and * & is the CDF for the input data from each satellite instrument. Note that the & here is taken as the logarithm of the mixing ratio.
As the quantile-mapping technique uses the quantile–quantile matching to converge the satellite WV distribution function to the BBH one, the biases between WV profiles from each satellite instrument and BBH observations are calculated for each quantile. These biases are applied back to the original satellite VRWV profiles on the original vertical levels. Figure 3-2 shows an example of the distribution of bias-corrected WV measurements from MLS in the range 2–3 km below the tropopause in the tropical region. It indicates that the bias-corrected WV from each satellite instrument has a much better agreement with the BBH observations.
As mentioned above, in the latitude range 90S–60S, the bias-correction is not available due to lack of BBH observations. Note, this will introduce a discontinuity in the dataset and an associated increase in measurement uncertainty. Figure 3-2: Example for the bias-correction method with the quantile-mapping technique with the MLS measurements. For creating monthly mean data for the individual instruments, the bias-corrected VRWV profiles are taken for a temporal and spatial level 3 aggregation. The aggregation is performed 16