Predictability of weather and climate

Abstract:

THE 1988 drought in the United States has been widely reported, not least as an indicator of the reality of the "greenhouse effect" - see for example ref. 1. On the other hand, drought is a naturally occurring phenomenon. Here we have studied 30-day forecasts of the atmospheric flow over the United States using a complex numerical weather-prediction model initialized with data from May 1987 and May 1988. In addition, an experiment with 1988 initial conditions and 1987 sea surface temperatures was made. The results indicate that much of the difference between the 1987 and 1988 forecasts was associated with interannual variability in sea surface temperature. These results provide complementary evidence to that in ref. 2, which suggested that the US drought was linked to anomalous oceanic conditions in the tropical Pacific. © 1989 Nature Publishing Group.

Abstract:

The observational studies focus on teleconnections between tropical outgoing longwave radiation and Northern Hemisphere extratropical height in wintertime, firstly using five-day mean fields, and secondly using monthly mean fields. On the month to seasonal time-scales, correlations between tropical sea surface temperature and extratropical geopotential height are discussed. The modelling studies include 10-day forecast experiments in which the tropical belt is relaxed towards the analysed state, seasonal time-scale GCM studies with imposed sea surface temperature anomalies, and results from simple nondivergent barotropic models. Results from the barotropic model indicate that the impact of the tropics on the extratropics is strongly dependent on the flow regime obtained in the extratropics. It is suggested that this may partially account for the variable impact of tropical forecast errors found in the forecast relaxation experiments. -from Author

Abstract:

It is shown from an assessment of a small set of extended range forecasts from two centres, and from a much larger set of medium range forecasts from one centre, that variability in predictive skill is strongly related to fluctuations in the Pacific/North American (PNA) mode of low frequency variability. A hypothesis is put forward that this is associated with the dependence of large‐scale instability of the forecast flow on the amplitude of the PNA mode. The hypothesis is tested in a barotropic model using as basic states, composite skilful and unskilful cases from the set of medium range forecasts and individual monthly mean fields. Results from the barotropic stability analysis suggests possible reasons for the asymmetric nature of the response of general circulation models to sea surface temperature anomalies of opposite sign, relevant to forecasting on monthly to seasonal timescales. Observational evidence for the stability hypothesis is also discussed. Copyright © 1988 Royal Meteorological Society