Estimate of uncertainties in the Aquarius salinity retrievals

We present a method for formally assessing random and systematic uncertainties in the Aquarius salinity retrievals. The method is based on performing multiple retrievals by perturbing the various inputs to the retrieval algorithm. This results in calculating the sensitivity of the Aquarius salinity to these inputs. Together with an error model for the uncertainties in the input parameters it is possible to calculate the uncertainty in the retrieved SSS. It is important to distinguish between random uncertainties, which get suppressed when computing weekly or monthly averages and systematic uncertainties, which do not get suppressed by taking averages. We compare the results of the formal uncertainty estimates with uncertainty estimates based on comparing the Aquarius salinities with those from external validation sources finding excellent agreement.