Improvements to meteorological analysis over the Antarctic and Southern Ocean region from the inclusion of additional observing systems

The in situ meteorological observing network in the Antarctic and Southern Ocean region is sparsely populated relative to other continents. Numerical weather prediction skill in Antarctica is poor, and as a scientific community our understanding of Antarctic atmospheric processes is underdeveloped. Yet the atmospheric dynamics occurring in the Antarctic region impact greatly on the weather of populated continents further north. Here several meteorological analysis and forecasting problems are investigated with close attention to the Antarctic continent and the surrounding Southern Ocean. Scatterometer wind data, derived from satellite measurements of microwave backscatter by the ocean surface, are an important modern source of data for Southern Ocean Meteorology. Wind speed and direction estimates from the scatterometer instrument flown on-board the European Space Agency ERS-1 satellite are tested for accuracy against numerical model fields. Alternative algorithms have been developed by several meteorological agencies to deduce surface wind speed and direction from microwave backscatter measurements. Comparison is made between 10m wind estimates generated by two alternative algorithms and co-located data from the European Centre for Medium Range Weather Forecasts' numerical analyses. Some limitations in the derivation algorithms, unique to the Southern Ocean, are detected. A data impact study, conducted within the framework of the Australian Bureau of Meteorology global data assimilation and prediction system (GASP), was conducted to assess the influence of ERS-1 scatterometer winds on numerical analyses and forecasts. The study identified several cases of substantial analysis impact, predominantly over the Southern Ocean. Mean impact statistics from 120 six-hourly assimilation cycles indicated a small positive impact on 24 and 48 hour forecast skill. The number of positive impact events exceeded the number of negative impact events and the inclusion of scatterometer data reduced the magnitude of forecast errors. The majority of large impact events arose from simultaneous assimilation of two or more observing systems coincident in time and space. Large discrepancies between model air-pressure predictions and drifting buoy reports often resulted in the elimination of the latter on quality control grounds. On numerous occasions the superposition of scatterometer winds provided sufficient supporting evidence for the inclusion to proceed and subsequently generate improved analyses and forecasts. Another valuable form of remotely sensed atmospheric data comes from the TIROS Operational Vertical Sounder (TOVS). Estimates of thickness between pressure surfaces, variously from the surface to 250hPa, were used heavily for meteorological analysis during the Antarctic First Regional Observing Study of the Troposphere (1-ROST). Here TOVS estimates are compared with co-incident radiosonde data from Antarctic and sub-Antarctic meteorological stations, supporting their validation. The representation of the Antarctic surface temperature inversion in the GASP analyses is measured against results from observational studies. Improved boundary layer structure in later versions of the model is confirmed and surface wind forecasting skill has improved. Fluctuations in inversion depth and strength in successive model analyses and prognoses is identified as a wind strength forecasting aid for coastal locations. The forecast skill of numerical weather prediction systems has increased steadily over the past two decades. The GASP system, on which the bulk of this work focuses, has also seen substantial skill increases. Improvements in forecast skill of the Australian global system have come from a combination of incremental changes over many years. While horizontal and vertical resolution has increased substantially, skill improvements in the Antarctic and Southern Ocean region have also come from the marked increase in data supply and usage. Particularly important are remotely sensed data, automatic weather stations and drifting buoys. The potential for further dramatic skill improvement in the years ahead is real. Proposed future satellite missions will offer a great deal to Antarctic Atmospheric Science. Significant new satellite systems will include NPOESS (USA; programme beginning 2005), COSMIC (USA/Taiwan; programme beginning 2005), ADEOS-II (Japan; launch planned for 2002), ENVISAT (European Space Agency; launched March 2002), Coriolis (USA; launch planned for August 2002) and MetOp-1 (European Space Agency; launch planned for 2005). Together they will provide the Meteorological Community with advanced atmospheric profiling technology, improved quality and resolution surface wind data, higher resolution multi-channel imagery and more. These remote sensing advances will come alongside an expanding network of automatic weather stations on the Antarctic continent and ongoing improvements to numerical modelling systems. Antarctic Meteorology is set to continue to benefit from the incorporation of existing observing systems as well as from new systems as they become available.