The central goal of this project is to serve as a strong, interdisciplinary center of forecasting research and science. We will achieve this through innovation in development of forecasting methodologies and systematic research into optimal the communication and visualization of forecasts. By working with the US Centers for Disease Control and Prevention in this cooperative agreement, we aim to extend existing methodologies to incorporate new data sources and model structures, thereby improving influenza forecast accuracy in the US. We will review and revise existing FluSight forecasting guidance, targets, and accuracy evaluation at the national, regional, and state levels. This will include conducting stakeholder interviews with federal, state, and local epidemiologists to track uses of forecasts and outcomes and identifying unmet needs from current forecasts. We will develop and refine methods to create forecast ensembles, with a specific focus on developing ensemble weighting schemes using robust, penalized methods to estimate model weights. We will identify methodologies and data sources that increase forecast accuracy for start and peak week forecasts, peak intensity, and short-term forecasts at the national, regional, and state level. Our work here will focus on developing multi-scale spatial models that leverage state and zip-code level data on influenza infections from public and private sources. We will develop communication products and methods to describe forecast results and uncertainty for federal and state public health officials and the public. We will achieve this by incorporating new visualizations into our existing interactive data visualization product for influenza forecasts in the US and studying systematically the end-user perception of various visualization and data presentation layouts. Finally, we will develop and adapt successful seasonal methodologies, data sources, and communication approaches for forecasting the timing, intensity, and short-term trajectory of an emerging influenza pandemic. Specifically, we will create, test, and disseminate weekly data summaries and visualizations from the most up-to-date sources of reported influenza cases in the US (including data from our real-time point-of-care data sources), and validate our new spatial models against simulated pandemic data.