Towards improved operational streamflow forecasts

                                                Abstract

A number of forecasting centers around the world produce streamflow predictions using methodologies 
that span a wide range of data requirements and complexity. From an operational perspective, 
understanding the marginal benefits of such approaches via hindcasting is necessary to improve the 
quality of real-time forecasts. In the first part of this talk, I will provide an overview of key 
lessons from the intercomparison of techniques for seasonal streamflow forecasting, conducted in five 
pilot basins of the U.S. Pacific Northwest Region. This intercomparison showed that hybrid techniques 
(i.e., those that combine statistical and dynamical approaches) that leverage predictability from 
initial hydrologic conditions and climate variables can lead to improved skill compared to current 
operational approaches. In the second part of the talk, I will present a summary of ongoing work on 
statistical post-processing of medium-range forecasts. These post-processing experiments are designed 
to better understand strengths and limitations of competing techniques over different seasons, and 
therefore to improve streamflow forecasts used for water resources management.