The Ensemble Transform Kalman Filter (ETKF) adaptive sampling methodology has been used in NOAA/NWS Winter Storm Reconnaissance (WSR) operations since 2001 to identify locations in which adaptive observations should be deployed to improve 1-3 day forecasts of winter weather. Commonly, the target locations are associated with the landfalling weather system itself or the relevant upper-level dynamics in the storm track, and are usually intuitively sensible.

One of the primary goals of THORPEX is to extend predictive skill and improve our understanding of predictability in the medium-range. Central to this goal is the identification of 'sensitive' locations upstream in which the assimilation of observations is likely to influence medium-range forecasts of high-impact weather downstream. A recent evaluation of the ETKF demonstrated that the method is able to discriminate between good and poor observations over the central Pacific ocean for the improvement of 3-6 day forecasts of 200 hPa wind, particularly on synoptic scales and when the flow is predominantly zonal. These results will be summarized, and new results will be presented for 20 WSR 2007 cases in which the spatial continuity of upstream sensitive regions over the northern Pacific Ocean can be traced as far as 7 days back from the forecast verification region over North America. These new results suggest that the targeting of in-situ or remotely-sensed observations over a broad area to the east of Japan may often be beneficial for 1-week forecasts over North America. Sensitivity in other locations such as the poles or tropics will be discussed briefly.