Understanding the predictive capacity of climate change risks and impacts
QUESTION
The Climate Report Cards published on the Sea Change Australia website are fantastic — the way the information is presented is engaging and easy to understand.
One question that often comes up is about the predictive capacity (accuracy of the predictions) presented in these kind of reports. How well have past projections aligned with observed changes so far? And how likely are these predictions to occur? Especially, given that climate is becoming increasingly variable and we know that making accurate predictions is likely becoming more challenging.
Could you please explain the main challenges in predictive capacity? For example, what the key uncertainties or limitations are, how accuracy is evaluated over time, and what the main challenges are in improving climate predictions into the future? It would also be great to know whether there are any reviews or studies that explain this, and if the report cards could include some explanation or discussion about the predictive capacity behind these assessments.
ANSWER 1
Written response:
Great question! The Climate Report Cards are designed to make complex science accessible, and it’s fantastic to hear that you find them engaging and easy to understand.
The report cards present three types of information about our oceans: (1) observed data over the past year; (2) seasonal forecasts that predict temperature over the next few months; (3) climate forecasts that project conditions in the far future. These different timeframes – observations, predictions, and projections – support different types of decisions, from short-term planning to long-term policy and adaptation.
The important bit to note is that predictions and projections are two different things. A prediction is an estimate of what WILL happen in the future, based on current conditions. A projection is a conditional estimate of what COULD happen in the future, based on possible scenarios – such as how much carbon is in the atmosphere.
The predictive accuracy of seasonal temperature forecasts varies regionally but typically in Australia are reliable over \~1-3 months. This accuracy is assessed by comparing forecasts with what was observed. The Bureau of Meteorology (BOM) regularly produces these seasonal temperature forecasts to reflect current conditions, and to also allow ongoing access to end-users. BOM also produces long-range regional forecasts over land to indicate expected temperature and rainfall patterns over next few months, as well as forecasts of the El Niño Southern Oscillation (ENSO) for up to 6 months in advance. ENSO is one of Australia’s biggest climate drivers, and BOM has recently revised the index of the oceanic index of ENSO to account for the long-term changes in global warming that we have experienced. This is a good example of how we need to constantly re-evaluate our climate indices and forecasts to take into account our changing world.
Climate projections have been quite successful at predicting large-scale trends, such as global warming and sea level rise. However, regional and ecosystem-level changes – like ocean currents or marine productivity – are more complex and harder to model precisely due to complex interactions and variability in the natural system. Where we have considerable uncertainty, it is important to communicate this. For example, projections of biomass are highly uncertain and we show the range of the data to communicate this uncertainty in the Climate Report Cards.
Climate models are constantly evolving and improving. Improvements can come from better observational data, more powerful computers, and enhanced regional modelling that better captures local dynamics. There are many scientific reviews that assess how well climate models perform. The Intergovernmental Panel on Climate Change (IPCC) regularly publishes assessments, and in Australia, the State of the Climate reports and the National Climate Risk Assessment are also excellent resources for understanding climate related impacts and associated uncertainty.
Answered by:
