March 1, 2012 in Chicago: Benji & Jack

Part 1: Precipitation Event in Chicago

From the hours of 0 - 2 Z, a clustered storm system traveled East across much of the Midwest bringing with it precipitation like snow, rain, and mixed phase wetting the tracks of it path. With variety in precipitation type and amount throughout the region, forecasts in Chicago were difficult to pin down due to an extremely low melting level. Rain ensued from the precipitation yet the low melting level was hardly detectable even using a variety of radar instrumentation. To try and better understand the event a case study of Chicago and surrounding areas was conducted primarily using data from the Romeoville, Illinois radar.

Figure 2: 0 Z Sounding, Davenport Iowa

While there is no sounding data from Chicago specifically, the University of Wyoming's sounding website offers historical data from two equidistant stations from Chicago, Davenport, Iowa and Lincoln, Illinois. To study atmospheric conditions near Chicago, we can analyze the two soundings in tandem.

From the Davenport Sounding, at 0 Z we have a significantly moist sounding as profiles of temperature and dew point temperature nearly overlap from the surface to 675 hPa. This significant moisture is indicative of precipitation or at the very least highly precipitable conditions as the saturation finally breaks off near 675 hPa and there is a very low lifting condensation level (926.8 hPa) showing low moisture laden clouds. In Davenport, the surface temperature just above freezing allows for wet snowfall, rainfall, or both in mixed phase precipitation. If raining, the sounding suggests an extremely low melting level consistent with the note given in Chicago. Lastly, constant westerly winds are present with height throughout the sounding as the system later seen in our radar travels eastward with time.

Figure 2: 0 Z Sounding, Lincoln Illinois

In Lincoln, we have a much larger boundary layer extending from the surface to 725 hPa. In stark contrast to the previous sounding, Lincoln has a fairly dry boundary layer only with conditions only becoming precipitable at the top of the boundary layer with a 725 hPa inversion. While the planetary boundary layer and low level temperatures of 12 Celsius in Lincoln differ greatly from Davenport, the soundings are similar in their Westerly winds. Finally, the Lincoln lifting condensation level presents itself at 796 hPa and surface conditions dictate that any precipitation falling must be rain.

Figure 3: 0.5 Degree Base Reflectivity Loop, 2012-03-01 0-2 Z

To begin a primary investigation into the March 1 Chicago event, an analysis of base reflectivity was done. Radar data from Romeoville, Illinois illustrates a a two hour loop from 0 - 2 Z following the storm system. Over the course of the event reflectivity values range from near 10 to 40 DbZ indicative of light to moderate rainfall in a bullseye pattern moving East across Illinois into Indiana. Base reflectivity can be used to get a good initial glimpse at our targets and non-targets to identify them. As the storm moves west, there are stationary blips of reflectivity within the 50 km range ring; the size and constant presence of these reflectivity values tell us that we are observing ground clutter that will also likely be present in other radar instruments.

Figure 4: 0.5 Degree Differential Reflectivity Loop, 2012-03-01 0-2 Z

 

Accompanying base reflectivity, dual-polarization capabilities allow us to extend our radar methods into new instruments like differential reflectivity. Differential reflectivity, ZDR, is mathematically calculated as 10 times the logarithm of the ratio of horizontal and vertical reflectivity values. Depending on the two values, our ratio can be either greater than or less than 1 and consequently our ZDR can range be positive, negative or zero.

The physical interpretations of these ZDR measurements are as follows: for positive measurements, our horizontal reflectivity is larger than our vertical and we have a target with greater horizontal axis than vertical. This is often seen in heavy rainfall with large drops as they become oblate due to drag forces altering a spherical geometry. When equal to zero, this is indicative of a spherical geometry or a tumbling particle, this is because in both cases the captured hydrometeor is non-preferentially reflecting with equal values in the horizontal and vertical. Lastly, when the value is negative we have targets with larger vertical axes and reflectivity, this may be seen in certain types of precipitation like needles, columns, and prisms. 

From our measurements, the values we have range from small positive values from 0.25 - 0.75 DbZ to rare but observable values of 2-4 DbZ. This indicates that we are likely experiencing large spread light precipitation with certain locations like the central of the border line between Indiana and Illinois and parts of Michigan receiving larger, heavy precipitation.

Figure 5: 0.5 Degree Correlation Coefficient Loop, 2012-03-01 0-2 Z

Another extension of dual polarization radar is correlation coefficient, in our 2 hour correlation coefficient (CC) loop from 0 - 2 Z we see a fairly constant reading of values as they transition from West to East. Colloquially, CC measures how uniform radar targets or hydrometeors are compared to others in their sweep. Possible values range from zero to one; zero meaning absolutely no correlation between horizontal and vertical reflectivity and greatly dissimilar targets. A value of one translates to absolutely perfect correlation between horizontal and vertical reflectivity and purely identical targets.

While both are theoretically possible yields for measuring CC, most observed values will fall between the two ends. Studying our storm over Chicago, CC values are near 1 everywhere outside of the 50 km range ring. This suggests significant uniformity among radar targets, thought to be nearly identical precipitating hydrometeors in light rain or snowfall. The values in blue and green with smaller correlation coefficients are indicative of ground clutter near the radar.

One specific application of CC is identification of a melting layer where stratiform rainfall is created from melting snowfall. True to the event's low freezing level, under all elevation angles from 0.5 - 4.5 degrees, a melting layer does not present itself as the temperatures are close but not sufficient enough to produce stratiform rainfall.

Figure 6: 0.5 Degree Differential Phase Loop, 2012-03-01 0-2 Z

Finally, differential phase is another application of dual-polarization radar to analyze weather systems. This instrument measures the slowed phase shift from EM waves propagating through different mediums like rain and clear air. The larger, more oblate, and more prevalent the rainfall amount, the larger the returned phase shift. By subtracting the vertical degree of phase shift from the horizontal, a differential phase shift is calculated.

Diagnosing the Chicago event, at 1:48 Z on the 30 degree line 200 km from the radar, we read a differential phase value of 47 degrees while 150 km away on the same line we have a value of 25 degrees. A more useful application of differential phase is specific differential phase shift, KDP. Applying the formula for specific differential phase shift we get [(47 - 25)/(2(200 - 150)] = 0.22 deg/km. This is indicative of light precipitation averaged over the area. 

Figure 7: NWS KDP Values

PART II: HAIL IDENTIFICATION USING DUAL-POL

August 7, 2012 - Milwaukee

On August 7th of 2012, a few convective storm-cells north of Milwaukee swept southeastward across the region before passing over Lake Michigan. In the process, these storms likely produced hail. Dual-polarization variables in radar analysis strongly point to this conclusion.

Base reflectivity values in southeast Wisconsin taken from the 0.5-degree elevation angle on August 7, 2012. 

Base reflectivity data clearly displays the presence of NW–SE-oriented horizontal convective rolls and shallow convective cells across southeast Wisconsin. These signatures have such low reflectivity that they are likely not precipitating to the ground if they are precipitating at all. They stand in distinction with the cells to the northeast of the radar which are much more spatially extensive in their continuous regions of very high reflectivity (upwards of 60 dBz). These large and intense cells likely contain heavy rain. In addition, regions of high base reflectivity values are a prerequisite for the potential presence of hail.

Correlation coefficient values in southeast Wisconsin taken from the 0.5-degree elevation angle on August 7, 2012.

Note that the convective rolls mentioned above have extremely low CC values—far too low to be directly attributed to meteorological phenomena. What is likely occurring is the upward and downward flux of dust, bugs, and other particulate matter into these shallow convective rolls. Any of these things are possible for an early August day in Wisconsin. The cells with high reflectivity, however, have CC values of nearly 1. This implies that the radar targets with the highest reflectivity are nearly congruent with each other (CC is reflectivity-biased, so the largest targets are preferred). Hailstones are relatively similar to each other as they fall, so this CC data provides more strength to the hypothesis that hail is falling within these storm-cells.

Differential reflectivity values in southeast Wisconsin taken from the 0.5-degree elevation angle on August 7, 2012.

ZDR imagery from August 7th shows how unlikely the convective rolls mentioned earlier are precipitating. This is since ZDR is reflectivity-biased—like CC—and the ZDR values in the regions with convective rolls are extremely high—above 4 dBz. This means that the radar targets the reflectivity imagery indicated in these regions are probably dust, debris, or bugs, since they are so irregularly shaped. However, in the stronger and larger storm cells where hail is likely occurring, ZDR values are near zero. ZDR values of around zero strongly point to hail since hail tumbles as it falls through the sky making it seem like a round object. Round objects have ZDR values of zero. 

Taking into account the various data that dual-polarization techniques in remote sensing provides us, we can confidently conclude that hail precipitated from the cluster of storm cells northeast of Milwaukee on August 7th, 2012. High reflectivities, high correlation coefficients, and near-zero ZDR values within these storm cells are all strong indications of hail.