Finding Differentiating Characteristics between Tornado-Producing and non-Tornado-Producing Mesoscale Convective Systems with Synergistic Usage of GOES Imagery and NMQ Data

Mesoscale convective systems (MCSs) can produce severe weather such as flooding, hail, lightning, wind gusts, and particularly tornados. The severe weather associated with MCSs is difficult to predict. Thus, our study, assisted with an automated software tool, focuses on MCSs in the continental US with the hope to find distinguishing characteristics between tornadic (i.e. tornado producing) and non-tornadic MCSs. Our recently developed program, Event Tracker, written in full object-oriented Java for cross-platform compatibility, enables us 1) to merge multiple sources of high-resolution meteorological data, 2) to automatically identify and track MCS events, 3) to extract and visualize physical and morphological features of MCS events throughout their life cycles, and 4) to catalog extracted features into an object database making further analyses convenient. Event Tracker is easily configurable with text-based configuration files and easily customizable with pluggable components.

Identification of MCSs, and particularly tornado-producing MCSs, is accomplished by specifying a brightness temperature threshold for the 10.8-micron Geostationary Operational Environmental Satellite (GOES) imagery; areas with brightness temperature lower than the threshold are labeled as potentially MCS components. In association with the GOES brightness temperature field, the archived National Weather Service (NWS) Watches/Warnings database is used to determine whether a (potential) MCS area coincides with reported occurrence of tornado(es). The database compiled by the Tornado History Project is also used as a supplementary source of information for the same purpose. Once an MCS is identified, it is tracked through time and its structural/morphological, as well as spatiotemporal, parameters are extracted and catalogued in an object database. The precipitation intensities of the National Multisensor and Mosaic Quantitative Precipitation Estimates (NMQ) data are then co-located with the MCSs to extract their precipitation characteristics, which are, in turn, catalogued in the database.

To compare tornadic and non-tornadic MCSs, we have processed 5.5 years of 4-km, ~½ hourly GOES Rapid Scan Operations (RSO) satellite data with the current prototype of Event Tracker. Cloud top temperatures of tornadic MCSs during the 5.5 years have the highest frequency of occurrence in the 209 K range. We compare this to a subset of 5 days, May 20-24, 2011, during which an explosion of hundreds of tornados were produced, including the F5 tornado that devastated Joplin, MO. The temperature of the highest frequency of occurrence is again around 209 K in the tornadic MCSs, whereas that for the non-tornadic MCSs is substantially higher, in the ~224 K temperature range.

In the next phase of our investigation we plan to process the 1-km, 5-minute NMQ 2D precipitation intensity data associated with the MCSs. Once fully processed by Event Tracker, we will be able to identify and track MCS events covering a period of 5.5 years and over the entire continental US. We will then generate MCS life-time precipitation statistics based on the production of tornados. By comparing the statistics we hope to identify discriminating features in their structure/morphology and precipitation characteristics in their life cycles that can help improve our nowcasting capability for the occurrence of tornadoes.