The blog post below was a combined effort of Ryan Ellis, Barrett Smith and Katie Dedeaux.
The National Weather Service describes wind shear as “…a change in horizontal wind speed and/or direction, and/or vertical speed with distance, measured in a horizontal and/or vertical direction. Wind shear is a vector difference, composed of wind direction and wind speed, between two wind velocities. A sufficient difference in wind speed, wind direction, or both, can severely impact airplanes, especially within 2,000 feet of ground level because of limited vertical airspace for recovery.” NWS forecasters must include wind shear within the aviation forecast if non-convective vertical wind shear of 10 knots or more per 100 feet in a layer more than 200 feet thick is expected within 2,000 feet of the surface at, or in the vicinity of the airport. This stringent criterion is difficult to attain and therefore, going explicitly by this criteria, very few, if any forecasts would include non-convective low level wind shear. Feedback obtained from many users has indicated that they would like to be alerted to possible low level wind shear more frequently, even if conditions do not explicitly meet the criteria.
Using a definition of low level wind shear (LLWS) of 10 kts or more per 100 feet in a layer more than 200 feet thick within 2000 feet of the surface is a forecasting problem in the sense that it is very hard to obtain data at this resolution whether in an observational platform or a numerical model. Radiosondes have good vertical resolution (still not enough to resolve 100 foot layers) but vary widely in space and have very poor temporal resolution. High resolution numerical models such as the Rapid Refresh Model (RAP) and the High Resolution Rapid Refresh model (HRRR) use sigma levels that pack more datapoints into the boundary layer but even in doing so, there are only five levels below 2000 feet, the lowest two of which are roughly 90 feet apart and the second and third levels being 160 feet apart at Greensboro,NC (KGSO). This is a start for forecasting low level wind shear, but if we can’t observe it, how can we verify it? Aircraft soundings are helpful but once again few and far between, as are pilot reports, especially specific to LLWS.
During the day on 7 February, 2015, an upper level trough was just beginning to move into the Northern Plains while the associated surface moved into the Midwest (Fig. 1). In response, the tightening pressure between the low and a high off the Southeast coast (Fig. 1), led to an increase in winds across central North Carolina during the afternoon hours. The strongest winds occurred over the eastern half of North Carolina, where winds were sustained 15-20kts and gusting to 25kts at times. A 22Z RAP forecast sounding (18Z run) shows these winds in a relatively shallow mixed layer depth around 3000 ft with a momentum transport value (a good approximation of gust potential) of 21kts (Fig 2). After sunset, forecast soundings showed a rapidly developing surface based inversion despite the relatively unchanged pressure gradient. Most model guidance indicated a 5 to 10kt wind at times overnight. Above the surface, the RAP showed the already modest flow of 20 to 25kts at 2000 feet strengthening with an intense nocturnal low-level jet (LLJ) of around 40kt at the top of the inversion. By 07Z, RAP soundings showed a LLJ sharply increasing to 39kts and 43kts at 1100 feet at KFAY (Fayetteville) and KRWI (Rocky Mount), respectively (Figure 3). Surface winds throughout the night were mostly 6 to 9kt and slowly increased. Due to the previously mentioned limitations in RAP forecast data made it difficult to determine if the technical definition of LLWS would be met since the two lowest data points (below 1000 ft) showed 21kt and 9kt, or 12kt of shear over roughly 500 ft. However, based on the magnitude of the LLJ and sharpness of the inversion, forecasters felt confident enough to add LLWS to the 00 UTC TAF issuance,
KFAY 072338Z 0800/0824 21006KT P6SM SCT250 WS012/22035KT
FM081300 23008KT P6SM SCT250
FM081600 23014G28KT P6SM SCT250=
While a PIREP report from near KFAY did not explicitly mention LLWS, it did described severe turbulence within 1 to 3 miles of the terminal.
FAY UUA /OV FAY240001-FAY240003 /TM 1455 /FL020 /TP SR22 /TB SEV 020-BLO /RM SMOOTHED OUT AT 022-023 DURC
One observational platform that can help tremendously to observe and verify LLWS is the wind profiler. The Earth System Research Laboratory (ESRL) Hydrometeorology Testbed – Southeast project operates wind profilers in the state of North Carolina in Charlotte, Clayton, Research Triangle Park, and New Bern. The website with data for these sites and others around the country can be found at http://www.esrl.noaa.gov/psd/data/obs/datadisplay/. The data for this case was provided by the Environmental Protection Agency.
On February 8th there were a number of pilot reports for moderate to severe turbulence. Profiler data from Research Triangle Park shows winds near 40 kts on both February 8th and February 9th as low as 0.5km (~1600 ft) with lighter winds below (Fig. 4). A great advantage of this data is that depending on the scanning mode the profiler is in, there can be as many as 11 levels below the 2000 foot threshold to use, which gives us increments every 160 feet or so. The temporal resolution is also good because data is collected every hour. There are some disadvantages to using the data for LLWS because the presentation as wind barbs still leaves the forecaster the task of calculating both the speed shear and the directional shear on their own.
In order to better visualize this data, we have written a python script to calculate the total wind shear between each layer of the profiler data, which then outputs a total wind shear value on a graph with time on the x-axis and height on the y-axis. Data is color coded for better visualization. Shear over 5 kts is colored yellow and shear values over 10 kts are colored red. Looking first at the data collected in the 2800 scanning mode (which directly corresponds to the wind barb plot shown) on February 8th, we see increased shear values in the lower levels before 1200 UTC and then relaxing after that time (Fig. 5). What may be surprising is that despite having values of 40 kts in the lower levels of the wind barb plot, the change in speed and direction from layer to layer is relatively small. As a result we can see that despite the obvious presence of a low level jet resulting in pilot reports of moderate to severe turbulence we still barely observe low level wind shear criteria per the definition . Looking quickly at the data from the 400 scanning mode (more data below 2000 feet) for the same days, we see similar trends and numbers that are not all that different from those obtained in the 2800 scanning mode. What we see here is that the numbers may be a little higher in the 2800 scanning mode because the vertical distance between data points is greater. The numbers are a little smaller in the 400 scanning mode because the vertical distance is less, but because there are more data points, the signal seems to be more consistent than in the 2800 mode.
This project is in its infancy and questions that we hope to address include an evaluation of the definition of LLWS as it currently stands and how that definition corresponds to observed impacts we are seeing in aviation. We hope to be able to provide better operational guidance for forecasters regarding LLWS by applying the same technology to high resolution numerical model output. Finally, this example did not fit our traditional conceptual model for LLWS, which typically involves a cold air damming airmass. Thus we hope to better understand the synoptic and mesoscale situations that lend themselves to observed LLWS in North Carolina and across the mid-Atlantic.