The Bakerian Lecture, 1991 The predictability of weather and climate

There is large public and political interest in the predictability of weather and climate, in particular in the influence of human activities on the likely climate change during the next century. Numerical models are the main tools which enable the nonlinear processes involved in the dynamics and physics of the atmosphere and other components of the climate system to be integrated in an effective way. The performance of such models used for weather forecasting has continued to improve as more accurate data with better coverage has become available, as improved descriptions of the physics and dynamics have been incorporated and as computing capacity and speed have increased. Studies of the predictability with models suggest that with further improvements in data and models deterministic forecasting of detailed weather may ultimately have useful skill up to 2-3 weeks ahead. Beyond the limit of deterministic forecasting, some skill remains for the forecasting of general weather patterns which can be pursued by studying ensembles of model forecasts from slightly varying initial conditions. The largest difficulty with further improvements of numerical models lies in their inadequate treatment of the motions too small to be explicitly resolved. Interactions between the atmosphere and the ocean are responsible for substantial variations on seasonal, interannual and longer timescales. Forecasts are being provided of seasonal precipitation in the Sahel region of Africa based on a knowledge of global sea surface tem perature (SST) anomalies together with the assumption that such anomalies tend to persist from one season to the next. Attempts to forecast SST anomalies have centred on tropical regions in particular on the El Nino. Simple models show some skill in forecasting El Nino events 3-9 months in advance. Studies with more elaborate models which as yet only show partial success in simulating these events demonstrate the complex nature of the interactions involved. Turning to the likely changes in climate next century: if no changes occur in the atmosphere other than the increase in C0 2 and other greenhouse gases due to human activities, the increase in radiative forcing due to a doubling of atmospheric C0 2 concentration would lead to an increase of about 1.2 °C in global average temperature. Water vapour and ice-albedo feedbacks raise this to a figure of about 2.5 °C (with an uncertainty range of 1.5—4.5 °C) as estimated by the Intergovernmental Panel for Climate Change. Such a change would dominate over forcing likely to arise from other factors, and this estim ated rate of change next century is probably greater than any which has occurred on earth during the past 10000 years. The main uncertainties in climate change predictions arise from the inadequacies of the models in their descriptions of cloud-radiation and ocean circulation feedbacks. Until there is more confidence in the treatment of these feedbacks there are bound to be large uncertainties associated with any predictions of regional climate change. To reduce the uncertainties there need to be improvements in computer power, in model formulation and in our understanding of climate processes together with a large programme of observations of climate parameters to provide early detection of climate change and to provide validation of climate models and to provide data for initialization of model integrations. An important question is whether changes in climate due to changes in radiative forcing are predictable. It is pointed out that the response to climate over the past half million years to changes in forcing due to the variations in the Earth ’s orbit (Milankovitch cycles) is a regular one; some 60% of variations in the global temperature as established from the palaeontological record occur near frequencies of the Milankovitch cycles. We can, therefore, expect the changes in climate due to increasing greenhouse gases to be a largely predictable response. Large, but probably predictable, changes in the circulation of the deep ocean have modified climate change during past epochs and could have significant influence on future climate change.