We have evaluated the ability of NOAA Wind Profiler Network (NPN) radar profiler observations to diagnose the low-level jet (LLJ) over the central United States. The importance of the LLJ in the hydrologic cycle of the central United States has been established through previous studies which have found a close a relationship between warm-season heavy precipitation episodes and strong LLJs. Hourly NPN data has the potential to provide insight into the structure and dynamics of the LLJ by providing continuous vertical wind profile observations, and in particular observations during the time of peak LLJ occurrence between the standard 00 and 12 UTC rawinsonde launches.
Hourly observations from the Lamont, Oklahoma radar profiler were time-matched with research rawinsonde data taken over the period of 7 April 1994 thorough 30 March 1996 at the Cloud and Radiation Testbed (CART) site yielding 2614 pairs of independent, co-located observations. These observations were used to determine the impact of two suspected sources of error on the ability of the profiler to accurately diagnose LLJ events. These are the limitations imposed by the height of the lowest range gate, and the possibility of contamination by spurious returns from migrating birds.
First, the effect of the height of the lowest range gate (500 m) on profiler detection of LLJs was investigated. Studies have shown that more than 50% of LLJ events have a wind speed maximum below 500 m calling into question the ability of the profiler to accurately depict these events. Our research indicates that even though LLJ events having a wind speed maximum below 500 m are less accurately represented than those above 500 m, about 50% of the events below 500 m are detected.
Second, the impact of signal returns from migrating birds on the accuracy of LLJ diagnosis by the profiler was evaluated. We instituted a variable second moment (velocity variance) thresholding routine to evaluate the effect on LLJ detection accuracy. Verification statistics for profiler accuracy in LLJ detection (e.g. probability of detection, threat score, and false alarm rate) were calculated over a velocity variance range of 1 to 5 m2s-2 for full year, seasonal, and LLJ criterion dependence. It was found that a second moment threshold value around 2 to 2.5 m2s-2 provided an effective compromise between maximizing threat score and probability of detection while maintaining a reasonable false alarm rate and data availability.