As series of evaluation studies have been conducted using a one year long archives of Wind Profiler and high-resolution radiosonde data from the Oklahoma ARM-CART site to determine: 1) whether the current quality control for 6-minute profiler data is adequate, 2) the temporal variability of wind reports, 3) the spatial variability of wind reports, 4) the accuracy of Wind Profiler observations relative to the Radiosonde data and 5) the portion of total Wind Profiler error attributable to instrument and error. This information is especially important if Wind Profiler data are to be used to maximum advantage in future mesoscale data assimilation systems and in otherwise ‘data sparse' regions of the world.

Major findings to be presented include:

A new, 2-sided Q/C scheme reduced undesirable temporal variability of the 6-minute profiler data by as much as 35%.

Tests of Profiler data alone quantified temporal variability increases both with time and elevation, ranging from about 1.5 ms-1 at mid levels to greater than 4.5 ms-1 at other elevation (>10 ms-1 at upper levels at 6 hours), along with significant diurnal variations and changes between “Low” and “High” mode reports.

Spatial variability test results, which compared contemporaneous rawinsonde and Profiler data in a series of 25 km thick ‘tubes' extending 125 km from the Profiler site, varied with altitude, ranging from 3 to 7 ms-1 at jet levels.

Rawinsonde/profiler comparisons made within 25 km showed mid-tropospheric RMS vector differences of ~2.3 m/s, with larger differences nearer the earth's surface and farther aloft.

For future data assimilation systems to take optimal advantage of highly accurate wind data (especially at the mesoscale), they must consider not only the accuracies of the separate wind data sources, but also these accuracies vary with altitude and time of day, as well as information about differences between the observational errors and the error of representativeness that can be obtained from results such as these.