Close Window

Iowa State University (Joseph Gale, COMET graduate student fellow): "Improved prediction of mesoscale convective system dissipation"

Final Report


The project proposed to study mesoscale convective system (MCS) dissipation, with emphasis on MCSs affecting the county warning area of the DMX-NWS. MCS dissipation is an important forecasting challenge for the DMX office because MCSs commonly pass through Central Iowa while progressing through their mature to decaying stages of development. Sometimes, MCSs approach Central Iowa with ongoing strong to severe lines of thunderstorms, only to weaken and dissipate while moving through the area. Other times, the MCSs continue to exhibit strong to severe convection while moving through the area. Accurate MCS dissipation forecasts are imperative not only in regards to severe weather potential, but also for rainfall, cloud cover, and temperature forecasting.

Observations and model output were analyzed from 47 cases that occurred during May-August 1998-99 in Iowa and surrounding states. Originally, the MCS cases were to be limited to those that affected Central Iowa directly, but in order to attain a larger dataset, the area of study was expanded to states surrounding Iowa. Cases were limited to nocturnal MCSs since it is the late night through afternoon dissipation of these MCSs that challenges forecasters. Cases were also limited to MCSs that did not form in strongly forced environments (e.g. a squall line ahead of a strong cold front would be excluded). The DMX-NWS does not typically experience difficulties with MCSs formed and maintained through strong dynamic forcing. It is a greater challenge to the forecasters when the MCSs form and progress in weakly forced environments.

The original observational parameters thought to possibly be important for MCS dissipation were the cold pool-shear balance described by Weisman (Journal of the Atmospheric Sciences, 1992) and Rotunno et al. (Journal of the Atmospheric Sciences, 1988), surface storm-relative inflow (SSRI), maximum 0-2 km winds, maximum 0-2 km theta-e, and lower to middle level lapse rates. All of these parameters were computed, but during the course of the project, two additional parameters were added—elevated storm-relative inflow (ESRI) and the low-level jet (LLJ). ESRI was added because as the project progressed, it became apparent that the majority of the MCSs studied occurred north of a boundary (stationary front, warm front, or outflow boundary), and thus, much of their inflow was likely elevated. In addition, low-level nocturnal temperature inversions may have developed in the vicinity of many of the MCSs for a significant portion of their life cycles, and these inversions may have decoupled the MCSs from the surface so they would have to rely on warm, moist air from above the surface as inflow. Adding the ESRI calculation was perhaps the best decision made throughout the project, as its results are perhaps the most significant of the parameters studied.

From the beginning, we recognized that the LLJ may play a significant role in MCS dissipation, but we felt the LLJ was indirectly included in the calculation of the maximum 0-2 km winds and consequently ESRI. However, it became increasingly important to us to analyze LLJ influences directly since it is a parameter DMX-NWS employees already utilize to aid with MCS dissipation prediction.

There was still one other addition to the observational parameters. Within the cold pool-shear balance calculation, at first we were just computing low-level vertical wind shear to compare to the cold pool strength. However, once we realized that almost all our cases exhibited linear convection for some if not most of their life cycles, we determined that it is just the line-normal vertical wind shear that is involved in the cold pool-shear balance since the cold pool advances normal to the line. This was just a minor adjustment, as all it demanded was the calculation of the line orientation angle and subsequent modifications to the already-calculated vertical wind shear.

Parameters originally examined from Eta model output included winds, convergence, moisture convergence, frontogenesis, and maximum theta-e all at 850 mb, vorticity advection at 500 mb, divergence at 250 mb, and jet streak influences. Here, too, there was an additional parameter added-850 mb theta-e advection. Members of the DMX-NWS suggested the addition of this parameter to the study, and it may be the parameter that has the most significant influence on MCS dissipation in the Eta model among the parameters studied.


The results are very intriguing, as some are expected results, while others are surprising. The cold pool-shear balance included both expected and surprising results. We expected this parameter to be difficult to measure with observations, but we did not expect the results to show an imbalance so overwhelmingly in favor of the cold pool. An optimal cold pool-shear balance is 1, but our cold-pool shear balances average greater than 2 at each time sampled, which would suggest the cold pool is overpowering the shear over long time periods. Analysis shows our shear values are roughly one-third Weisman's (1992) shear values for his moderate-shear simulation and one-fifth the shear values for his strong-shear simulation. Curiously, our cold pool-shear balances are roughly three times as large as those for his moderate-shear simulation. Thus, our large cold-pool shear balance values are due almost entirely to low shear values, as opposed to stronger cold pools.

At first, we were concerned about the lack of agreement between our results and the cold pool-shear balance theory; however, there appear to be several reasonable explanations for the differences, and a few other studies support our findings. It appears that wind shear above the 0-2.5 km layer considered in the theory is not negligible as assumed in Weisman and Rotunno's studies and can aid in system longevity. In addition, an imbalance in the shear balance may more directly affect the intensity of individual convective elements, and not necessarily the larger-scale convective system.

Of the remaining observational results, ESRI appears to have the most significant influence on MCS dissipation. A significant decrease in ESRI occurs, on average, as the MCSs approach dissipation. The decrease is due to a combination of a decrease in MCS speed of movement and a decrease in the maximum 0-2 km winds. SSRI also experiences a decrease as the MCSs approach dissipation, though it is not as great a decrease as with ESRI. The decrease in SSRI is primarily due to a decrease in MCS speed of movement since surface winds generally change little and are small in magnitude compared to the MCS movement. It is interesting to note that MCSs that occurred north of boundaries did not experience a large average decrease in SSRI but did in ESRI, while warm sector MCSs experienced large average decreases in both SSRI and ESRI. These results support the notion that MCSs north of boundaries depend more on elevated inflow than surface-based inflow, and warm sector MCSs may depend on inflow from both layers.

There is also evidence that the LLJ is a factor in MCS dissipation, and MCSs in general. Over 90% of June-August cases (LLJ data was unavailable for the May months) were affected by an LLJ at some time during their life cycles. In addition, there was a tendency for MCSs previously affected by an LLJ to weaken and dissipate once no longer affected by an LLJ.

Lapse rates generally changed very little with time and do not appear to be factors in MCS dissipation. However, there were a handful of MCSs that experienced larger decreases in lapse rates than most, and those MCSs tended to dissipate at or soon after 12 UTC. Maximum 0-2 km theta-e decreased on average by nearly 10 K from 00 UTC to 12 UTC. However, in general, the theta-e values at 12 UTC from 2 to 5 km AGL are lower than the 0-2 km maximum, implying the MCSs are still moving into conditionally unstable environments at 12 UTC. The sounding used by Rotunno et al. (1988) and Weisman (1992) to represent the initial environments of their squall line simulations had a maximum low-level theta-e value of 338.2 K. Most MCSs in this study moved through environments of maximum 0-2 km theta-e values greater than 338 K at both 00 UTC and 12 UTC. There is also a lot of variability in these values as the standard deviation is also around 10 K.

One additional informal observation was made during the summer of 2000. It was observed in a few ongoing MCSs in the late evening to early morning hours that if an outflow boundary or gust front could be discerned moving ahead of the convective line, within two hours, the convective line would weaken in intensity. After another couple hours, the convection would be almost gone, and then soon after the precipitation area would shrink and bring about the dissipation of the MCS. Since this is just a casual observation of a small number of MCSs, it cannot be stated that the outflow boundaries moving ahead of the convective lines are signs of upcoming MCS dissipation. However, we believe DMX-NWS employees should study this and see how well it seems to work and how quickly the MCSs dissipate after their outflow races ahead of their convection.

The Eta model output provided mixed results with no one parameter standing out as a strong predictor of MCS dissipation. There are a few parameters that may be moderate predictors of MCS dissipation: winds, frontogenesis, and theta-e advection all at 850 mb. Each of these parameters shows a moderate decrease as the MCS progresses toward dissipation in the Eta model. The 850 mb convergence, moisture convergence, and theta-e 500 mb vorticity advection, and 250 mb jet streaks do not appear to be factors in MCS dissipation in the Eta model.

The 250 mb divergence is more complicated. The original intent of examining upper-level divergence was to gauge whether or not decreases in divergence suppressed upward motion and led to dissipation of MCSs in the Eta model. Unfortunately, the Eta model generates divergence from its MCS convection that masks environmental divergence. Thus, it became impossible to accurately examine the ambient divergence, and the focus instead turned to the convectively generated divergence. The most useful aspect of the convectively generated divergence is that it shows the instantaneous location of the Eta model MCS. However, evidence does not suggest that a decrease in 250 mb divergence is a strong predictor of MCS dissipation.


When the project was just beginning, there was a fair amount of DMX-NWS interaction, mostly involving trips to the DMX-NWS to discuss how to organize the project and to ask DMX-NWS employees for input. Much time during DMX-NWS visits was also spent learning about MCSs via COMET training modules. Once the project procedure was organized and work began, there were less DMX-NWS visits since we determined most of the work could be done faster and more efficiently at Iowa State. Thus, for a while, trips to the DMX-NWS were made primarily to update them on progress and get more input. As the project went on, the DMX-NWS was able to help us fill some data gaps in model output and profiler data. Also, Karl Jungbluth would make trips to Ames to meet with Dr. Gallus and Joe Gale to discuss how the project was coming along.

In addition, Joe Gale was able to spend one morning and afternoon at the DMX-NWS to witness a severe weather event firsthand. A strong MCS moved into Central Iowa from the south, and all along its long convective line, the weather was borderline severe, which meant the radar had to be monitored constantly in case a warning needed to be issued. Later, as the system moved through the Des Moines area, the office was bombarded by calls from the phones and the HAM radio. This experience showed that if any MCS dissipation guidelines were developed for the DMX-NWS to follow, they would have to be simple and easy to execute because when MCSs affect the area, it gets very hectic at the office.

Other valuable interactions with the DMX-NWS included giving presentations. Two "official" presentations were given, deemed seminars by the DMX-NWS, one early on to introduce Joe Gale and explain what the project was all about and what role the DMX-NWS employees could play in the process. The second presentation was about a year into the project when Joe Gale presented results completed with the observational data. At that time, he was beginning the model analysis and discussed the parameters he was going to compute and asked for input. That is when DMX-NWS employees suggested theta-e advection as a parameter to examine. There were other "unofficial" presentations as well. These were just informal meetings in the middle of the office that were not planned in advance. These were quite helpful because Joe Gale could talk one-on-one at length with DMX-NWS employees and gain valuable insight as to what they already consider important in regards to MCS dissipation.

Since the project has been completed, we have now been working with the DMX-NWS to develop ways the results can be used operationally to aid MCS dissipation forecasting. Formulas for parameters such as SSRI and ESRI are being programmed into AWIPS. In addition, an MCS dissipation guideline sheet or checklist is being written for DMX-NWS employees to use during the upcoming convective season. The checklist reminds the forecaster of which parameters to examine while the MCSs are in progress. This will help the forecaster examine the parameters quickly in the heat of the moment. This checklist may be formulated into a flow chart as well. We hope our results will benefit the DMX-NWS' prediction of MCS dissipation, and perhaps if our guidelines work well for the DMX-NWS, other NWS offices in the region may want to use it also.


Throughout the course of the project, we took advantage of several great opportunities to present our results at conferences. There were NWS personnel at each conference, from the DMX-NWS and other NWS offices spanning the country. At both the conferences in Des Moines and Orlando, Joe Gale presented the results of the Fellowship in front of audiences of over 200 people. We engaged in valuable interaction with members of the meteorological community, including NWS personnel. One NWS member from the Houston office discussed in detail Texas MCSs and how they challenge the Houston office's forecasters. He hoped in the future to undertake an MCS dissipation study for his region and contact us for ideas and input on procedure and parameters to examine.

Gallus, W. A., Jr., and J. J. Gale, 1998: Improved prediction of mesoscale convective system dissipation. Low Level Jet Science Sharing Session-National Weather Service, Sept. 2, Ames, IA.

Gale, J. J., and W. A. Gallus, Jr., 2000: Toward improved prediction of mesoscale convective system dissipation. Severe Storms and Doppler Radar Conference-Central Iowa NWA, March 31-April 2, Des Moines, IA.

Gale, J. J., and W. A. Gallus, Jr., 2000: Toward improved prediction of MCS dissipation. 20th Conference on Severe Local Storms-American Meteorological Society, 11-15 Sept, Orlando, FL.

Gale, J. J., W. A. Gallus, Jr., and K. A. Jungbluth 2000: Toward improved prediction of MCS dissipation. Preprints, 20th Conf. on Severe Local Storms, 11-15 Sept., Orlando, FL, 343-344.

Gale, J. J., W. A. Gallus, Jr., and K. A. Jungbluth 2000: Toward improved prediction of mesoscale convective system dissipation. Submitted to Weather and Forecasting, Dec. 2000.