Abstract
Dissective analysis of predictive skill for medium range ensemble streamflow forecasting is carried out for 43 headwater locations in the US. To assess the component-specific contributions of the full HEFS, 20-yr hindcasting experiments were performed for the wettest 3-month period for each location. The results show that, when all ranges of observed flow are considered, hydrologic modeling, the use of NWP precipitation forecast and its post processing, streamflow assimilation and streamflow post processing contribute on average 6, 43, 31 and 20% to the collective predictive skill in reference to climatology, respectively. The contributions vary greatly from location to location and with the magnitude of the verifying observed flow in reflection of the predictability of precipitation and streamflow. When only high flows are considered, the use of NWP precipitation forecast and its post processing is by far the largest contributor. For several locations, including some with significant snowmelt, the predictive skill is noticeably poor due to deficient or very challenging hydrologic modeling relative to climatology. Overall, the results indicate that, for holistic improvement of medium range ensemble streamflow forecasting, efforts should be directed at improving hydrologic modeling particularly at skill-challenged locations in addition to improving future forcings particularly for large events.