Observing Network Design Applied to Antarctic Surface Weather Observations

The current surface observing network over the Antarctic continent is sparse, expensive to maintain, and not optimally configured for weather or climate monitoring. For regions not properly sampled by the surface observing network there is insufficient information to validate the accuracy of model analyses and forecasts. Moreover, the relatively rapid climatological changes that Antarctica has experienced in the past several decades motivate evaluating the need for additional observations to improve climatological sampling capabilities.

Here we use data from the Antarctic Mesoscale Prediction System (AMPS) to test the efficacy of the surface observing network through “data denial” experiments. The main focus is to 1) evaluate the impact of observations on the analysis of surface temperature over the Antarctic continent, 2) determine if surface observations are sufficiently sampling the climate signal of the continent, and 3) test the impact of observations compared to predictions from network design theory. Using the 00Z Global Forecast System (GFS) analyses as the prior, surface temperature data is assimilated using an ensemble square root filter, and the analyses are then compared to the 00Z AMPS analyses. By withholding stations of interest during the assimilation step, the impact of a given station may be gauged by determining if analyses degrade, improve, or remain the same if a given observation is not assimilated. The sampling abilities and coverage of the surface observing network is assessed through spatially and temporally averaged impacts of withholding observations.