Subseasonal-to-seasonal (S2S) precipitation outlooks are crucial to support the management of agriculture, water resources, and natural disasters (FLOODS/DROUGHTS)

However, IT is still challenging to produce reliable S2S precipitation outlooks for the contiguous united states. While weather regimes defined as typical large-scale circulation patterns are proven useful to characterize precipitation in observations, less is known about the prediction skill of the weather regimes by forecast models and whether and to what extent these regimes can be used to improve S2S precipitation outlooks. This proposal directly responds to the opportunity provided by the climate testbed (CTB-3) that supports method improvements used to make S2S outlooks, including model post-processing via innovative statistical techniques and applications of existing statistical techniques. To aid NOAA?S CTB-3, this project will comprehensively test and verify a newly developed method that directly applies dynamically forecasted weather regimes for developing S2S precipitation outlooks. Specifically, this method will combine weather regimes in forecast models and precipitation in observations through finite mixture models. The S2S precipitation outlooks are a weighted SUM of precipitation maps that will blend forecasted weather regimes and observed precipitation. To test this unique method developed by the PI?S group, this project will accomplish three tasks: 1) develop S2S precipitation outlooks, i.e., week 3-4 outlooks, monthly outlooks, and seasonal outlooks by applying the weather regime-gaussian mixture model (WR-GMM) framework to the dynamical forecast outputs of those available to NOAA?S climate prediction center (CPC) in real-time, 2) develop a web-based interface for disseminating the precipitation outlooks and their verifications, and 3) transit these precipitation outlooks into experimental mode for evaluation and verification during the duration of this project. This project is expected to advance NOAA?S S2S precipitation outlook through the unique method we have developed by machine learning methods including K-means clustering for obtaining weather regimes and innovative statistical techniques - finite mixture models. The web-based interface maintained by utah state university will not only provide useful tools for forecasters, but IT can also be used as an education and outreach platform. NOAA CPC will be our primary partner and recipient of the proposed products. Diagnostics and products created from this two-year CTB project will be ready for rapid transition to operations, pending a positive review by ncep-WPO. The WR-GMM framework will also be immediately applicable to the operational seasonal forecast system (SFS).