SEASONAL PREDICTION AND PREDICTABILITY STUDIES AT ECMWF
A report of a lecture given by David Anderson but based on work of the seasonal prediction group at ECMWF consisting of T Palmer, D Anderson, C Brankovic, T Stockdale, 0 Alves, M Balmaseda.
Abstract
At ECMWF, there are two projects concerned with seasonal prediction. The first is a programme of experimentation to quantify the degree of atmospheric predictability globally. This involves running a T63 version of the ECMWF atmospheric model with prescribed (observed) sea surface temperatures. The initial conditions for the atmospheric state and for land surface conditions are obtained from the re-analysis project at ECMWF. A nine-member ensemble of 120 day integrations is currently being performed for all four seasons for each year from 1979 to 1993. The second is the development of a comprehensive global coupled atmospheric/oceanic model complete with data assimilation system. A number of hindcasts have been performed for the 90s using as initial ocean conditions, a control run of the ocean forced by operational surface fluxes, or an assimilation run using all available XBT and TAO data from the tropical Pacific. The skill of the model predictions out to 6 month lead is encouraging for the cases tested, especially so since the 90s were a difficult time to predict as evidenced by the results from other models. However, only a few runs have been performed and the skill of the model has yet to be tested fully.
1. Introduction
The successful prediction of El Nino during the 1980s, and the realisation that much of the world's weather is influenced by El Nino on seasonal to interannual timescales, has lead to a dramatic increase of activity in the field of seasonal to interannual prediction. Two developments in particular have led to a realisation that operational seasonal prediction will be feasible: the development of realistic coupled oceanlatmosphere models, and an improving ability to measure the upper ocean, using both enhanced in situ observations, and remotely-sensed measurements.
The research programme on seasonal to interannual prediction at ECMWF is concentrating on two basic objectives. The first objective is to quantify the predictability of seasonal mean atmospheric variability, over Europe and regions where there may be a European interest, based on atmospheric model integrations with prescribed observed sea surface temperature (SST). The second objective is to develop a fully coupled ocean atmosphere model and ocean data assimilation system to the stage where real-time seasonal forecasting is possible.
2.Predictability of atmospheric fluctuations on seasonal timescales.
