Investigating multi-decadal climate variations in seasonal forecasts

The ability to forecast weather and climate reliably on a seasonal timescale is of great societal and scientific interest. If floods, droughts, and rainfall onset can be reliably predicted a season or more in advance, the economic and social costs of these events can be significantly reduced. Examples include being able to better predict tropical droughts in northern South America, India, parts of southern Africa and Australie, and earlier warnings of tropical cyclones in the Philippines and other parts of southeast Asia. This is particularly topical as some extreme weather events increase in frequency and severity due to climate change.

Seasonal predictive skill has improved markedly over the past century, with successive generations of forecast systems improving on the accuracy and length of forecasts that are possible. A significant source of predictability, especially in the tropics, is provided by El Niño Southern Oscillation (ENSO), an irregular periodic variation in winds and sea surface temperatures (SSTs) over the tropical eastern Pacific Ocean. It is the strongest mode of multi-annual atmospheric variability and affects air pressure, precipitation, and temperatures all over the world through its impact on atmospheric and oceanic circulation patterns, known as teleconnections. There is evidence that these teleconnections are not stationary but can change over time. There is generally lower predictive skill in other areas of the world, notably the North Atlantic, where there is a complex, seasonally dependent relationship between ENSO, the North Atlantic Oscillation (NAO) and the strength and position of the jet stream. Issues, such as the signal-to-noise paradox (climate models have a low signal-to-noise ratio, but nevertheless are able to skilfully predict observed climate variability) and the ENSO spring predictability barrier (the correlation between predictions and observations dramatically decreases in boreal spring), have been identified and are the subject of open research.

This project aims to investigate and identify the large-scale atmospheric and oceanic dynamical reasons for the observed multi-decadal variations in forecast skill, helping to understand what factors influence such low-frequency fluctuations in the climate state and model forecast skill and, ultimately, how to improve seasonal forecasts.