Characterizing Spatial and Temporal Sampling Uncertainty in the SWOOSH Database

The Stratospheric Water Vapor and OzOne Satellite Homogenized (SWOOSH) data set merges data from five different satellites to create a continuous record of stratospheric ozone and water vapor from 1984 to the present. This record can be used to understand how water vapor and ozone may have changed over time or to validate model results. Although SWOOSH provides rough uncertainty estimates, it does not currently provide comprehensive estimates of uncertainty resulting from non-uniform spatial and temporal sampling within monthly latitudinal grid boxes. These uncertainty estimates could be critical in interpreting the long-term SWOOSH record. To quantify such uncertainties, a climate model was sampled using the viewing geometries and overpass times of the SWOOSH instruments. The sampled climate model was compared to the true monthly zonal mean from the climate model, resulting in an uncertainty estimate. The results of the model sampling suggest that sampling uncertainties are substantial in the early record (1984-2004) and small in the later period, with the onset of the Aura MLS satellite (2004-on) providing significant improvements in coverage. In the early record, ozone sampling error is particularly high towards the poles at low pressure, while water vapor sampling error is higher around the equator and mid-latitudes at high pressure. Though sampling uncertainties decrease during Aura MLS era, they remain high towards the poles, reflecting poor satellite sampling at high latitudes. Different methods of correction, including interpolation to the center of each latitudinal and temporal bin, were used to correct the sampling bias. The effectiveness of these corrections was tested by comparing the corrected data to the model truth.