Kane et.al (1987) found that the composite precipitation patterns associated with various types of MCSs (based on the Maddox archetypes) were very similar but that the scale of the events differed.  They also found that synoptic and frontal type systems were typically larger than mesohigh type events.  Why are there differences in scale between these different MCC archtypes?  A study of MCSs over the Southeast by Geerts (1998) found that MCSs were more common in summer but were of larger scale and lasted longer during winter. 

This suggests that the scale and strength of the forcing may be one factor that determines the scale of an MCS.  Augustine and Caracena (1994) have shown that the very largest scale, longest lived MCSs are associated with stronger frontogenetical forcing than smaller scale shorter lived MCSs

Scale Characteristics of MCSs.

Their work supports the idea that one of the factors that governs the size of the MCS is the scale of the forcing and the strength of the low-level frontogenesis.  

 

However, the scale of the event also may be dependent on the available moisture (the mean relative humidity and the precipitable water).  Junker et al. (1999) studied the rainfall associated with the MCSs during the 1993 warm season and found that the scale of the area covered by 3 inches or greater rainfall depended on the precipitable water and thickness at which the rain fell.  Almost all of the larger scale events occurred with mean 1000-500 relative humidity values of greater than 70 percent.  With lower relative humidity than 70 percent,  if a 3 inch area was observed, it was very small. 

 

 

 

 

 

Model guidance often does a pretty good job of predicting the scale of the event when strong forcing is present due to strong shortwave that is accompanied by a strong low level jet and warm advection pattern.  However, the models do poorer when the dynamics are weak and the scale is small. A graph showing the threat scores for four different types of convective systems (Watson 1999) illustrates this point. 

Higher threat scores were associated with synoptic and cool type low events than during frontal or mesohigh events which typically have a smaller scale than either the synoptic or cool low type event.  The small scale of the heavier amounts makes it difficult to provide meaningful deterministic or probabilistic forecasts for a specific point.  Note that the verification of 12-h  MOS PQPF for 2.00 inch or greater amounts shows only about a 3% improvement over climatology for the 24-36 hr period (from Shirey). 

 


Unfortunately the current generation of operational models often cannot handle propagation effects since the NMC models convective parameterization schemes and microphysical packages do not realistically simulate downdrafts.  Next, propagation will be discussed because of its importance in determining the mode of development (archtype) and movement of MCSs. 

 

The core of the heavy rainfall associated with MCSs and MCCs is also usually small making forecasting its location a major forecast problem (Kane et al. (1987), McAnelly and Cotton 1986), Junker et al, 1999).   Kane et al. studied 74 mesoscale convective systems that met the size and shape requirements to be defined as MCCs.  They noted that the maximum rainfall for the 74 cases averaged 104 mm and that the median was 89 mm.  However, they also noted that the size of the 77 mm or greater amounts was usually very small compared to the size of the MCC.   Therefore, when they developed a normalized composite of their cases, they found probability of rainfall amounts greater than 77 mm (3 inches) was limited to a little more than 10 percent.  This suggests that even if a forecaster were correct in predicting the development, evolution and track of an MCS, it would still be very difficult to predict the core of greater than 77 mm rainfall.    This argues for a probabilistic approach to predicting heavy rainfall.

 

A figure from McAnelly and Cotton (1989) shows that the area of precipitation thresholds associated with MCSs decay almost logarithmically.  

The figure below shows the cumulative rainfall that caused the Fort Collins flash flood.   The rainfall during the Fort Collins flood illustrates the small scale that extreme rainfall events can have. The 4 inch or greater rainfall  fit into a 10 by 10 km grid.  The small scale associated with many extreme convective events argues  that in the foreseeable future,  numerical models will show little, if any, skill in accurately forecasting them explicitly.    

HPC verification (figures above) show a similar decay as the thresholds increase but this time the scale on the left has been converted to a logarithmic one.  Note the almost log linear decay of precipitation thresholds and of the HPC forecast accuracy (Threat Scores).

A more rational approach to forecasting extreme precipitation thresholds that are associated with mesoscale convective systems might be to forecast and verify them within a pre-defined area.  For example a forecast 5 inch area that verified somewhere within 150 km of the forecast might be considered a hit. 

However,  HPC currently forecasts precipitation thresholds associated with MCSs including the  rare high end thresholds deterministically.  The following pages offer suggestions on predicting the evolution and movement of MCS in hopes of improving precipitation forecasts of convective systems. However,  HPC excessive rainfall potential outlooks are now probabilistic. 

However,  high resolution, non-hydrostatic ensemble runs with explicit convection may offer potential in assessing the probability of a small scale heavy rainfall event occurring within a general area providing the ensemble suite is unbiased. 

From McAnelly and Cotton 1989