Summary:
Mesoscale convective vortices (MCVs) are diabatically induced, mid-tropospheric circulations that form within mesoscale convective systems, that often subsist beyond the “parent” convective system, and that can even lead to the formation of new mesoscale storm systems (Clark et al 2010). Typically forming during late spring and early summer, MCVs can cause widespread, potentially heavy rainfall that can produce localized flash flooding. They are also capable of producing localized severe weather, such as strong winds, hail, and even tornadoes (Snook 2011).
In spite of these hazardous phenomena, little research has been done to investigate the effectiveness of model ensembles in forecasting MCV formation (Davis et al 2003) and intensity (Zhang et al 2006). While the models can usually predict an MCV’s track, forecasters must rely on other means, such as radar and past cases, to determine if or when they will produce severe weather, should the MCV form at all. The purpose of this project is to determine model effectiveness for past cases using the Storm Scale Ensemble Forecasting (SSEF) system, singling out models and parameters as needed. This will allow for the use of multiple models and multiple data sets, which will in turn allow for a greater range of comparisons to examine.
To accomplish this, we will use national radar and satellite data from an archive kept by the University Corporation for Atmospheric Research (UCAR) to confirm visually the presence of MCVs. Upon finding several cases of MCVs, data from the SSEF will be used and compared to judge the model’s effectiveness. Of course, it should be noted that there are limitations to this method. The SSEF has been operational since 2007, but only during the months of April, May, and June, and with several days missing during those months. The number of cases found during this time will be fewer than preferred, and there may be some exceptional case that must be missed due to the lack of data; however, there are still plenty of cases to be had and there should be no other major problems in this regard.
A reasonable amount of error is expected, but we hope to find that this project will help to identify the source of the errors and allow for more accurate prediction of MCVs in the future. By increasing the accuracy of MCV forecasts, lives may be saved due to improved warnings and better preparation in the event of MCV-related severe weather.
Project Narrative:
Introduction
Many studies have been conducted about the characteristics of mesoscale convective vortices (MCVs), including their general structure and maintenance mechanisms. MCVs are cyclonic circulations in the lower troposphere, found within mesoscale convective systems (MCSs) in mesoscale areas of condensational heating. They are formed out of deep convection due to diabatic heating combined with local vorticity maximas, as found by Hertenstein and Schubert (1999). These vortices have a distinct tendency to exist well beyond the death of the MCS that births them, sometimes going on to form new mesoscale storm systems, and have been found to last multiple days in some cases (e.g., Clark et al 2010). While it was initially thought that these long-lived MCVs could only form in environments of weak shear, Davis et al (2003) showed that in general, strongly sheared environments can support long-lived MCVs as well.
On radar, MCVs are generally visible as tight circulations, in extreme cases resembling miniature land-based hurricanes, an aspect that has led some historical cases [such as the 2009 (Coniglio et al. 2011) and 1980 MCVs that affected Missouri and Illinois, respectively] to be informally referred to by the public as “inland hurricanes.”
Size and location is another important aspect of MCVs, as the vortices can affect surrounding weather (on a synoptic scale) for multi-day periods (Davis et al 2003).
MCVs are not to be confused with so-called “dry vortices.” These are generally undetectable on radar and satellite, and yet can move into an area with favorable conditions for convection and develop into something that can resemble an MCV. Once convection occurs, dry vortices are actually capable of inducing MCVs, though they cannot considered MCVs themselves (Davis et al 2003).
Image 1: June 7, 2010 MCV enclosed in red box. Example of a typical MCV.
These storms typically form in the warm season, most often during May, June, and July. Observational studies have shown that MCVs do not decay after sunset, at times strengthening despite the lack of diabatic heating (Davis et al 2003).
MCVs are an integral source of rainfall for many of the drier regions of the Great Plains. This very feature, however, can also lead to widespread, potentially heavy precipitation that causes localized flash flooding. Other potential hazards can come of MCVs as well, such as localized severe weather conditions (strong winds, hail, and even tornadoes).
Davis et al (2003) found that the Rapid Update Cycle (RUC) was unable to predict formation of MCVs while others (e.g. Zhang et al 2006; Kong et al 2007) found increased error growth at convective scales in models that allow for convection. These errors are compounded by the generally poor quantitative precipitation forecasting during the warm months, when MCVs are more frequent. This deficiency in the prediction of MCV formation and intensity highlights a dire need for further documentation of ways to measure and forecast these storm systems.
Despite the potential hazards related to MCVs, little research has been done on the effectiveness of forecasting for MCVs using model ensembles. While the general movement of pre-existing MCVs can be accurately forecast, predicting their formation and the extent of their effects is much more difficult. This allows for the troubling possibility of potentially deadly situations caused by the heavy rainfall, as well as by more isolated severe weather events (e.g. heavy winds, hail, etc.). Since MCVs are such an important source of water in the Great Plains, being able to predict where and when they might occur may also help with water resource management.
Objectives
We plan to investigate the forecasting abilities of the model ensembles and in doing so, uncover weak points and/or errors that may be addressed by future researchers.
Uncovering the extent to which ensemble forecasts can accurately predict MCVs will, in the immediate future, help forecasters determine how much trust they can place in the models. Hopefully, our study will help reveal weaknesses in the forecasts that later researchers and model developers can then overcome to improve model performance.
Scope, Methods, Tools (How)
Using the satellite and radar image archive set up by NCAR, we have isolated seven cases of MCVs in the months of April, May, and June, from 2007 through 2011, to be analyzed. In each case, the MCV is evident in the image, and thus, the aforementioned “dry vortices” will be avoided for more textbook examples. We will request data from the SSEF and feed that through a script given to us by our advisor, Dr. Adam Clark, with modifications wherever necessary.
We have no way of knowing how well the SSEF will forecast the MCVs because this specific topic has yet to be fully researched. As for the script itself, it will be specified once all the model ensemble data has been received, as it is dependent on the input. As such, the procedure is still quite tentative.
Broader Impacts
As earlier mentioned, MCVs can beget hazardous conditions such as heavy, localized rain, strong winds, large hail, and tornadoes. We have chosen two specific cases in order to showcase these deadly effects.
On June 10, 2011, an MCV moved through Texas and Arkansas, dumping up to 7 inches of rain in the Albert Pike Recreation Area, located in the Ouachita National Forest in southwestern Arkansas. The localized heavy rain led to deadly flash flooding along the Little Missouri River, which swept through the campgrounds in the Albert Pike Area during the early morning hours, killing around 20 people.
The MCV in question is pictured below in image 2. The radar image is from the early morning, when rain was falling heaviest on the Recreation Area. Figure 1
Image 2: June 11, 2010 MCV enclosed in red box. Related to deadly flash flooding in AR.
Figure 1: SPC Storm Reports from the MCV event on June 11, 2011
Another high-impact MCV took place on May 8, 2009, affecting a large swath of the central Mississippi Valley. Originating as a derecho in Kansas on May 7, the system developed into a long-lived MCV. Sustained winds were observed at speeds over 60 mph in Kansas, Missouri, Kentucky, and Illinois, where the strongest gusts (over 100 mph) were measured.
Forty-seven tornadoes were reported across the states of Missouri, Illinois, Kentucky, and Tennessee, 39 of which were verified. The strongest of these were two EF-3s, one near Kirksville, Kentucky, and the other near Ponoma, Missouri. In southeast Missouri, hail up to the size of baseballs was reported by a storm spotter, with golf ball sized hail reported in other states. Flash flooding was also reported in Kentucky and Tennessee.
Four people were killed in this MCV-related event, with an estimated $500 thousand in damages across the effected region.
Image 3: May 9, 2009 MCV enclosed in red box. Related to strong winds and tornadic conditions.
These two cases illustrate potential MCV impacts on the public, as well as the importance of being able to forecast their initiation and progression. If forecasters have greater confidence in their models, then they can make better forecasts. In order to have such confidence, there needs to be certainty in the effectiveness of the model's ability to predict MCVs.
It is the hoped that the results of this study will allow for greater accuracy in these predictions, through the verification of the SSEF's MCV capability.
Statement of Work
As mentioned before, this study is meant to further investigate the MCV forecasting capabilities of the SSEF. This will be done through a verification process that will compare model data to the observations for selected cases.
A tentative timeline for the project:
2011, December 13: Submit proposal to advisor for revisions
2011, December 16: Submit final proposal
2011, Mid-to-Late December: Request SSEF data
2012, January: Receive data, adjust script, begin analysis
2012, February: Finish analysis
2012, March: Begin write-up
2012, April: Begin final draft, complete poster and other media
2012, May: Submit and present
Throughout the process, we will meet with our mentor Dr. Clark for help and to confer over our progress.
Team member contribution will be as follows:
Preliminary research was undertaken under the guidance of Dr. Clark, with particular emphasis on the structure and formation of MCVs. Autumn Losey conducted much of the initial research.
Further instruction led us to search for specific cases of MCVs. The initial searching was done by Dan Stewart, with verification by both Andrew Ryan and Autumn Losey. The final selection of MCVs are as follows, organized by date:
2007, May 8-9: Oklahoma, tornadoes reported
2009, May 8: Central Mississippi Valley, strong winds, hail, flooding, and tornadoes reported
2010, June 2: Iowa, Illinois, Michigan, hail reported
2010, June 3: Southern Texas, flooding reported
2010, June 7: Kansas, Missouri, hail, and wind reported
2011, May 9: North Dakota, Minnesota
2011, June 1: Florida, hail reported
Image 4: Selected MCV cases, (from top left, clockwise) May 8-9, 2007, Oklahoma; May 8, 2009, Upper Mississippi Valley; June 2, 2010, Iowa, Illinois, Michigan; June 3, 2010, Texas; June 7, 2010, Kansas, Missouri; May 9, 2011, North Dakota, Minnesota; June 1, 2011, Florida
Andrew Ryan designed and uploaded the website, taking charge of much of the technological aspects of the project.
Dan Stewart completed the project summary, while Autumn Losey and Andrew Ryan performed proofreading. Both Autumn Losey and Dan Stewart shared the responsibility of writing the proposal.
We will submit the cases to Dr. Clark, who will request the SSEF data.
Upon receiving the SSEF data, all team members will share the duty of analyzing and comparing them to the real life cases.
Autumn Losey will write the actual paper and press release, while Dan Stewart will create the poster.
Andrew Ryan will be in charge of the main presentation, with some accompaniment by the other team members.
References:
Clark, A. J., W. A. Gallus, M. Xue, and F. Kong, 2010: Convection-allowing and convection-parameterizing ensemble forecasts of a mesoscale convective vortex and associated severe weather environment. Wea. Forecasting, 25, 1052-1081.
Coniglio, M. C., S. F. Corfidi, and J. S. Kain, 2011: Environment and early evolution of the 8 May 2009 derecho-producing convective system. Mon. Wea. Rev., 139, 1083-1102.
Davis, Christopher A., D. A. Ahijevych, S. B. Trier, 2003: Detection and Prediction of Warm Season Midtropospheric Vortices by the Rapid Update Cycle. Mon. Wea. Rev., 130, 24–42.
Hertenstein, R. F. A., and W. H. Schubert, 1991: Potential vorticity anomalies associated with squall lines. Mon. Wea. Rev, 119, 1663–1672
Kong, F., K. K. Droegemeier, and N. L. Hickmon, 2007a: Multi-resolution ensemble forecasts of an observed tornadic thunderstorm system. Part II. Storm- scale experiments. Mon. Wea. Rev., 135, 759–782.
Snook, N., Z. Ming, and Y. Jung, 2011: Analysis of a Tornadic Mesoscale Convective Vortex Based on Ensemble Kalman Filter Assimilation of CASA X-Band and WSR-88D Radar Data. Mon. Wea. Rev., 139, 3446-3468.
Zhang, F., A. M. Odins, and J. W. Nielsen-Gammon, 2006: Meso-scale predictability of an extreme warm-season precipitation event. Wea. Forecasting, 21, 149–166.
FOR FUTURE PAPERS:
∗ Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. Wily-Blackwell, 265-273.