New Report: The State of Climate Prediction

A new report recently released by the National Research Council called “Assessment of Intraseasonal to Interannual Climate Prediction and Predictability” examines the current state of climate forecasting over time periods of a few weeks to a few years, and makes suggestions on how these forecasts might be improved.

The International Research Institute for Climate and Society is among a number of institutions that produce climate predictions, such as seasonal or longer-term rainfall and temperature forecasts on a monthly basis. Such institutions, the report states, could increase the value of these products to decision makers by enhancing procedures for archiving and disseminating information. In addition, the report concludes that making advances in observational capabilities, statistical and dynamical models and data assimilation systems could improve our understanding of key climate processes as well as improve forecasts.

IRI research scientist Lisa Goddard was on the committee that wrote the report. In the brief Q&A below, she discusses the publication and some of its key recommendations.

Assessment of Intraseasonal to Interannual Climate Prediction and Predictability

First, why is such a report necessary?

LG: The report was primarily commissioned by the National Oceanic and Atmospheric Administration, which wanted an assessment of the current capabilities in seasonal prediction and what additional efforts might improve the quality of forecast information. We know we have some skill in predicting the climate on intraseasonal-to-interannual time scales. For the United States, much of this skill is realized during El Niño or La Niña events. In order to improve our skill, we would need not only better models, but more complete observing systems as well as better techniques for inserting those observations into the models’ initial conditions for prediction. There are other aspects of the climate system that may influence the climate on these time scales, such as the stratosphere or land-atmosphere interactions. These will require much more research, observations, and modeling before the operational community can quantify their impact on intraseasonal-to-interannual predictions.

However, we wrote the report with a broader audience in mind. We included sections on the history of prediction, on how forecasts are made, and the extensive observations, scientific research and operational efforts required to develop, improve and communicate these forecasts.

The report recommends some “best practices” for improving the utility and accessibility of forecasts to researchers and decision makers. What are the major impediments that prevent the uptake of this information by these groups currently? Is there one best practice that stands out from the rest in your opinion?

LG: In my opinion, creating publicly-available archives of information associated with forecasts is paramount. IRI’s experience is that the needs of researchers, decision makers and others who would use climate forecasts, or the model predictions on which they are based, are too diverse and difficult for any operational center to address thoroughly. So making available the data from the models and the observations, as well as what considerations went into the issued forecasts is very important. It allows different communities to tailor or assess the information in ways that are more consistent with their decision processes or risk thresholds.

The report also lays out some key research questions that need addressing if we are to improve our forecasts. Which of these intersect directly with your work and why are they important to answer?

LG: The focus of my research is on how to make the best use of available prediction information, especially to those who might be able to act on that information. This is related to the report’s recommendations on improving the development and understanding of multi-model ensemble prediction and merging statistical and dynamical techniques

I think this is an important issue because models are still deficient when it comes to representing some of the characteristics of the climate and its variability. These deficiencies aren’t necessarily the same from one model to the next. The better the models and their use of observations become, the more robust the data I have to plug into my own research. So the key research questions that others throughout the climate community are addressing to improve forecasts also intersect directly with my work.