Dan Collins & Emerson LaJoie, NOAA: Subseasonal-to-Seasonal Climate Prediction

Value of the Information & Strength of the Argument

The value of the information lies in its direct insight into the operational forecasting practices at NOAA’s CPC. The speakers explain complex statistical calibration techniques with a clear logical structure, demonstrating how model biases are corrected and probabilities are made reliable. The argumentation is solid: they show verification examples (e.g., Brier score improvements) to support the effectiveness of calibration. They honestly acknowledge limitations, such as state-dependent predictability and cases where uncalibrated models perform better. The presentation successfully conveys the need to combine multiple sources of predictability and tools. However, the talk assumes familiarity with meteorological concepts, and some slides are dense. Still, the reasoning is coherent and based on practical experience rather than speculation.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high: the speakers are practicing experts at a national forecasting center, and the methods described (e.g., multivariate linear regression, Brier score, reliability diagrams) are established in the literature. They reference at least one specific method proposed by ‘Bandel et al.’ (though details are not fully given), and the slides likely contain references, but these are not visible in the transcript. The title accurately reflects the content: the talk indeed focuses on accessing predictability for S2S prediction. No comments were provided to analyze social trends. The sources are primarily internal NOAA work and recognized statistical verification tools. The adéquation between title and content is strong, with only slight mismatches, such as not deeply covering AI methods despite mention, but overall the talk is well-aligned.

The title accurately reflects the content: the talk focuses on accessing sources of predictability for subseasonal-to-seasonal climate prediction, as presented by NOAA researchers.

8/10

The presentation is given by NOAA scientists with expertise in climate prediction, providing solid technical content based on operational practice and peer-reviewed methods. The information is reliable but not original research; it summarizes current forecasting approaches and challenges.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:30Introduction of speakers and context of the seminar.
• 05:00Explanation of S2S prediction challenges and sources of predictability (trends, ENSO, MJO).
• 10:00Discussion on model biases and the need for calibration with a comparison of winter temperature trends.
• 15:00Statistical and dynamical model examples, including weekly forecasts and the MLR model.
• 20:00Calibration techniques (Brier score, reliability diagrams) and verification results showing improvement.
• 30:00Seasonal forecasting and development of full probability distributions for official outlooks.