High-spatial resolution (~40 km) dropwindsondes deployed during North Pacific adaptive observing programs have revealed large-amplitude variability in vertical profiles of wind velocity and temperature within GPS dropwindsonde soundings. Though these oscillations have spatial continuity between adjacent high-resolution soundings, they are uncorrelated over the typical distances between operationally targeted soundings (~200 km); less so over the ~400-km spacing of surfaced based radiosonde ascents. A critical unresolved issue is the extent to which operational forecast errors arise from uncertainty in observing system spatial/temporal representativeness. Under the current NWS/NCEP Winter Storm Reconnaissance Program (WRSP) and the previous NRL/Monterey-NCEP NOrth-Pacific predictability EXperiment (NORPEX, 1998), additional high-spatial resolution soundings were inserted between the 200-km resolution targeted soundings. These additional soundings were not transmitted for real-time ingest in the data assimilation systems of the international global prediction centers. It is our intent to use the high-spatial resolution (<40 km) soundings, taken in observationally sensitive regions of the North Pacific, to explore the impact of radiosonde and dropwindsonde representativeness error on the 2-10 day predictability. Ensemble forecasts will be made with the ECMWF model using differing subsets of the NOAA/G-4 soundings. For ~10 previous targeting cases, >40 soundings per flight were deployed, but only ~18 of these were assimilated into the operational global forecasts. Our study will address issues such as how sensitive is forecast error growth to variations in position and spatial resolution of targeted dropwindsondes.