AI Graph Cast ECM ENS Becoming No. 1 Performing Model across North America in Medium-range

Highlight: AI Graph Cast ECM ENS becoming No. 1 performing model of the past 60 days.

Fig. 1-2: Skill scores for 500 MB Anomaly forecasts across North America for the past 60 days identifies the impressive skill of the AI Graph Cast ECM ENS.  

Discussion: The 500 MB anomaly verification across North America for the past 60 days reveals a trend indicating the AI Graph Cast ECM ENS competitive with ECM ENS as the lead forecast process. In the 6-10-day period during meteorological winter 2024-25 so far, AI Graph Cast ECM ENS is consistently second-best trailing the usual top model, the ECM ENS (Fig. 1). Note that AI Graph Cast ECM ENS passes ECM ENS during the past 7 days. The other AI model evaluated is the AIFS which has produced middle of the road skill sore. The GFS lags well behind other operational and AI models.

In the 11-15-day period, AI Graph Cast ECM ENS is competitive with ECM ENS with a significant jump in skill score to defeat ECM ENS during the past 7 days (Fig. 2). Once again, AIFS skill scores are middle of the road and GFS ranks last.

Findings are operationally significant. Climate Impact Company favors ECM ENS forecasts for daily medium range outlooks. Based on the established trend, increased usage of AI Graph Cast ECM ENS is warranted. Anticipated are similar findings for Europe/Western Russia, South America, and Australia forecasts.