Eugene W. McCaul, Jr., Charles Cohen and Donald J. Perkey
USRA, Huntsville, Alabama

Severe convective storms are among the most dramatic manifestations of Earth's weather, and have attracted the attention of numerous research meteorologists - and the general public - for a long time. The complexity and variety of these storms have, however, posed major challenges to those trying to understand and forecast them better. For instance, it is commonly observed that one synoptic-scale frontal cyclone will produce a concentrated outbreak of severe weather and tornadoes, while another, equally intense, cyclone might spawn storms that produce only heavy rainfall and scattered wind damage. There continues to be great difficulty in distinguishing which weather systems will produce which mix of the various types of severe convective storms. Indeed, achieving detailed and reliable forecasts of the convective storm component of every large weather disturbance is one of the Holy Grails of modern weather forecasting.

Part of the difficulty lies in the incompleteness and imperfection of weather observations. Even with the magnificent technological advances that have led to satellite and Doppler radar monitoring of weather systems, coverage is still often lacking in certain details crucial to the convective storm problem. Yet, perhaps an even larger difficulty lies in the inherent complexity of atmospheric convective motions, which can range in size and intensity from small cumulus clouds up through giant systems of cumulonimbus clouds organized into linear systems thousands of miles long. It is believed that the diversity of convective storm structures must ultimately derive from the sensitivities of convective storm dynamics to variations in either the vertical or horizontal structure of the local atmosphere, or both.

In the COMPASS project, we have devised a relatively simple way to classify the relevant vertical structure parameters of an idealized atmosphere, to construct distinct atmospheric profiles for different combinations of values of these parameters, and to build and conduct a parameter space numerical simulation study using a state-of-the-art cloud model to explore how convective storms look and behave in different parts of the parameter space. The number of key parameters in this study turns out to be eight. They are:

  1. Convective available potential energy (CAPE); this is a measure of bulk energy that clouds can turn into vertical kinetic energy; in this project, we use the standard definition of CAPE based on assumed pseudoadiabatic ascent of air parcels after they become saturated; thus we ignore the complexities of the latent of heat of fusion, which becomes a subject of study in itself within this project;

  2. Hodograph size parameter, here realized as a hodograph radius; this is a measure of the bulk magnitude of horizontal wind and its vertical shear in the storm environment;

  3. Altitude of maximum parcel buoyancy; this is implemented here using a buoyancy profile shape parameter acting on idealized analytical profiles of parcel buoyancy;

  4. Shape of the vertical shear profile; this is implemented here using a wind profile shape parameter analogous to that for buoyancy; by varying this parameter, we can generate significant additional vertical shear at low levels in the storm environment;

  5. Depth of the subcloud mixed layer; this corresponds roughly to the height of cloud base (the lifted condensation level, or LCL); in all simulations we use a boundary layer having a subcloud layer that is as close to well-mixed as possible, without encouraging spontaneous turbulence in the starting environment;

  6. Depth of the moist layer feeding the updraft; this corresponds roughly to the height of the level of free convection (LFC); here we specify a moist layer having constant equivalent potential temperature throughout the subcloud mixed layer, and sometimes above the mixed layer; in those cases where the moist layer is allowed to be deeper than the mixed layer, we specify a nearly-saturated moist adiabatic environmental profile up through the specified depth of the moist layer;

  7. Temperature at cloud base, or approximately equivalently, the amount of precipitable water, in the environment; cooler environments contain less water vapor and less precipitable water than warmer environments, all other things being equal;

  8. Free tropospheric relative humidity; for simplicity, we specify this humidity to be a constant everywhere in the troposphere above the moist layer; our default simulations use a value of 90%, but we are also examining other lower values.

Our buoyancy profiles are shaped with an analytical function of the form m2z e(-mz). This function, while not the only one that could have been used, has several convenient mathematical properties, especially the constancy of its full vertical integral under arbitrary changes to the shape parameter "m." We employ a similar function to shape the y-component of wind when building our environmental hodographs.

For carefully chosen limiting values of each of the eight parameters, we can do a series of numerical simulation experiments that is just tractable with current computer technology, but which should embrace most of the range of variability found in convective storms, resulting from variations in the vertical thermodynamic and kinematic structure of the storm's environment. Findings from the COMPASS project will thus help provide a new, comprehensive and logical framework for understanding and interpreting the behavior of deep convective storms in the real atmosphere.

Because COMPASS uses the already-existing dynamical and physical framework of a cloud model, no new terms in the dynamical equations are being looked for. Rather, the emphasis in COMPASS is on storm morphology, the combined effects of storm intensity and structure that are the manifestations of how a storm circulation responds to its environment. Preliminary findings indicate that simple changes to a few parameters from among the list of eight can yield very large differences in storm intensity and size. In weakly unstable environments having only 800 J/kg of convective available potential energy (CAPE), for example, storms fail to sustain themselves when parcel buoyancy is not concentrated to sufficient strength near cloud base. If, however, the same buoyancy is concentrated near cloud base, and the level of free convection is raised in conjunction with specification of a cooler cloud base temperature, updrafts approaching 40 m/s in strength become possible! These huge increases in updraft strength are the result of stronger accelerations from perturbation pressure gradient effects near cloud base, a deeper moist layer feeding the updraft when the LFC is high, the reduced water loading, and earlier release of the latent heat of fusion in the updraft afforded by the cooler storm environment. This same trend in storm intensity is found at larger CAPE values as well, except that storms are not so easily suppressed for unfavorable buoyancy profile shapes when CAPE is large. The strength of the storms in cooler environments seems surprising, mainly because of awareness that CAPE tends to be smaller in cool environments, a limitation that idealized numerical simulation studies can easily circumvent. The rigor of the conceptual framework of COMPASS encourages better general understanding of convective storm behavior by eliminating the possible sources of confusion resulting from the atmosphere's tendency to visit the various parts of the eight-dimensional parameter space very unevenly.

This website is intended to provide an overview of the COMPASS project and its scientific results. COMPASS is the largest and most comprehensive parameter space study of convective storm morphology ever attempted. Because it deals with eight independent parameters, the simulation experiment nomenclature, explained in the next section, is necessarily somewhat cumbersome. However, after a little practice, the simplicity and utility of the nomenclature will become apparent. The reader will find it beneficial to become acquainted with this nomenclature before proceeding to an examination of the simulation results. Results are presented in graphical form, in a series of four-panel maps of the structure of the mature storms, each from four separate but closely-related experiments. However, the reader is cautioned that storm structure is seldom time-independent, and the snapshots of storm structure shown in these maps cannot be taken as definitive. Additional graphics will be added as this research effort proceeds, and more results are obtained.

COMPASS is supported by a grant, ATM-0126408, from the National Science Foundation, under the supervision of Dr. Stephan Nelson.