Theodore W. Funk
 James T. Moore

National Weather Service Forecast Office        6201 Theiler Lane
Louisville, Kentucky 40229


Formally of Earth and Atmospheric Sciences Saint Louis University
St. Louis, Missouri 63103


Heavy precipitation within winter extratropical storm systems often exhibits mesoscale organization, evolution, and banding. Therefore, accurate prediction of the location, duration, and precipitation amounts associated with mesoscale banded structure is difficult. Numerical model data often indicate the potential for heavy precipitation, but usually lack the resolution to accurately forecast mesoscale banded structure. Several studies (e.g., Shields et al. 1991; Moore and Kaster 1993) have addressed the relationship of various forcing mechanisms to the development and evolution of mesoscale bands of heavy snow. These mechanisms include:

1) Isentropic Lift 5) Elevated Upright Convection
2) Jet Streak Interaction 6) Melting-Induced Circulation
3) Conditional Symmetric Instability 7) Gravity Wave Propagation
4) Frontogenesis  

Except for gravity waves, the above mechanisms are investigated in this study to determine their role and interaction in producing a zone of record-breaking snowfall across the northern half of Kentucky on 16-17 January 1994.

Precipitation began as freezing rain across northern Kentucky during the evening of the 16th, changed to heavy snow near midnight (in Louisville), then continued through the morning of the 17th. In less than 12-h, snowfall amounts of 1-2 feet occurred within a mesoscale east-west band across northern Kentucky (Fig. 1). Louisville officially received 15.9 inches (a record at that time), with isolated amounts up to 23 inches east of the city. Hourly snowfall rates within the band ranged from 1-3 inches during much of the storm, enhanced at times by embedded thunderstorms.

Storm Total Snowfall Map Fig. 1: Storm total snowfall (in inches) for the period 1200 UTC (700 am EST) 16 January 1994 to 0000 UTC (700 pm EST) UTC 18 January 1994.

The case is somewhat unusual in that a deep low pressure center at the surface and aloft was not present. Instead, southerly surface flow existed on the backside of an arctic high pressure center, with a broad moist southwest flow from 850 to 300 mb ahead of an upstream "open" trough axis (not shown).


Synoptic-scale upward vertical motion of moist air produced precipitation across the entire Ohio and Tennessee River Valleys. Much of this lift was accomplished through strong warm air advection and isentropic lift. The observed 296 K isentropic surface at 0000 UTC 17 January (Fig. 2) indicated a tight gradient of pressure and mixing ratio across western Tennessee and Kentucky with strong cross-isobaric and cross-isohume flow from higher-to-lower values of pressure and mixing ratio. Thus, pronounced adiabatic upward motion and moisture advection were present in Kentucky along the warm conveyer belt. Gridded data from the NGM revealed that significant lift and moisture advection continued across central Kentucky between 0000 and 1200 UTC on several low-level isentropic surfaces. Latent heat release likely resulted in a diabatic contribution to vertical motion with respect to the isentropic surfaces (Moore 1993). Garcia (1994) has determined empirically that average isentropic mixing ratios of 4-5 g/kg, located across western Kentucky on the 292 K surface at 0000 UTC (not shown), are capable of generating 12-h snowfall amounts of 8-10 inches. While strong isentropic lift and moisture advection caused significant precipitation across the Ohio Valley, other mechanisms apparently concentrated very heavy snowfall amounts within a narrow mesoscale zone across northern Kentucky (Fig. 1).

296 K Isentropic Surface at 0000 UTC 17 January 1994 Fig. 2: Wind vectors (arrows in m/s), constant pressure lines (isobars; solid lines in mb), and constant mixing ratio lines (isohumes; dashed lines in g/kg) on the 296 K isentropic surface at 0000 UTC 17 January.


Several studies have documented the importance of jet streak interaction to the production of enhanced vertical motion and heavy snowfall. Lifting occurs in the ascending branch of the thermally direct transverse circulation within the entrance region, and the thermally indirect circulation within the exit region of a jet streak. Uccellini and Kocin (1987) showed that upward motion is enhanced by the merger (coupling) of the ascending branches of two separate jet streaks, which often accompany heavy snowfall along the East Coast. Hakim and Uccellini (1992) documented jet streak coupling for a snow band across the northern Plains, while Shea and Przybylinski (1993) did likewise for a snowstorm in Missouri.

In this case, the 6-h NGM forecast valid 0600 UTC 17 January revealed a well-defined 300 mb jet streak across the Great Lakes and another across Texas and Arkansas (Fig. 3). The entrance region of the northern jet and the exit region of the southern jet appeared coupled over the Lower Ohio Valley, with both exhibiting substantial along-stream variation in the wind field. As a result, the coupled ageostrophic wind response to the geostrophic deformation was significant, as evidenced by divergent ageostrophic wind vectors (Fig. 3) and NGM 300 mb divergence fields (not shown). The exit region of the southern upper-level jet also may have intensified the southerly low-level jet (e.g., 70 kts at 0000 UTC at Little Rock at 850 mb), resulting in enhanced low-level convergence and ascent across Kentucky.

NGM 6-Hour Forecast of 300 mb Isotachs at 0600 UTC 17 January 1994 Fig. 3: Nested Grid Model (NGM) 6-h forecast valid 0600 UTC 17 January of 300 mb constant wind speed lines (isotachs; solid lines in m/s) and ageostrophic wind vectors (arrows; m/s). Bold straight line shows path of the cross-section used to depict ageostrophic circulation shown in Fig. 4.

A spatial-height cross-section, derived from the 6-h NGM forecast and defined by the line in Fig. 3, depicts the transverse circulation of tangential ageostrophic winds valid at 0600 UTC (Fig. 4). The direct ("D") and indirect ("I") thermal circulations, as well as a pronounced deep-tropospheric upward motion area clearly are defined between the two jet streaks across the western half of Kentucky. The location and timing of this enhanced ascent are coincident with an increase in precipitation intensity across western and northern Kentucky.

Spatial-Height Vertical Cross-Section Showing an Ageostrophic Circulation at 0600 UTC 17 January 1994 Fig. 4: Vertical ageostrophic circulation along cross-section line shown in Fig. 3 based on NGM 6-h forecast valid 0600 UTC 17 January. Horizontal component of arrows is proportional to the ageostrophic wind component in the plane of cross-section; vertical component of arrows is proportional to the vertical motion in the plane of cross-section. Dashed (solid) lines are isopleths of upward (downward) vertical motion. "D" ("I") indicates direct (indirect) thermal circulation.


CSI has been studied by numerous researchers (e.g., Emanuel 1983; Bennetts and Hoskins 1979). Given moisture and lift, CSI is an instability that can result in slanted mesoscale circulations of saturated air parcels. CSI results from the combined effect of vertical (gravitational) and horizontal (inertial) forces. When CSI exists, the atmosphere often is weakly stable to both vertical (upright convection) and horizontal (inertial instability) displacements, but unstable to slanted movement. In other words, a parcel displaced vertically or horizontally eventually would come back to its original position; however, a parcel displaced slantwise would result in a titled upward acceleration. (Inertial instability and, therefore, CSI are possible near a significant anticyclonically-curved jet entrance region).

Several (qualitative) atmospheric conditions must be met to produce CSI. These include 1) near saturated atmosphere, 2) near neutral stability (but slightly stable or else upright convection could develop), 3) strong vertical speed shear (baroclinic environment), and 4) large scale forcing for upward motion to produce parcel displacements. The Paducah 0000 UTC 17 January sounding (Fig. 5), isentropic ascent (e.g., Fig. 2), and NGM forecast relative humidity and wind fields (not shown) indicated that these four conditions were met in this case.

Paducah, Kentucky Sounding at 0000 UTC 17 January 1994 Fig. 5: Environmental sounding from Paducah, KY (PAH) at 0000 UTC 17 January 1994. The right (left) solid line show the environmental temperature (mixing ratio/moisture) vertical profile. Wind direction and speed (in kts) are shown at right. The sounding shows a low-level subfreezing layer, an elevated warm/melting layer (above 0 deg C) and moist subfreezing air aloft.

An assessment of instability is possible by evaluating spatial-height cross-sections of absolute momentum (M) (Emanuel 1988) and equivalent potential temperature (theta-e) surfaces. CSI apparently is present in saturated areas where the slope of the theta-e surfaces are steeper than the M surfaces (i.e., near neutral stability and significant vertical speed shear exist). In the 17 January case, CSI criteria was met in the stippled area from northern and western Kentucky to eastern Louisiana at 0600 UTC (Fig. 6). Note, however, that within the northern (left) portion of the shaded CSI region, theta-e and M lines actually are nearly parallel.

Spatial-Height Vertical Cross-Section of Equivalent Potential Temperature and Geostrophic Momentum Fig. 6: Cross-section of equivalent potential temperature (theta-e; dashed lines; K) and absolute geostrophic momentum (solid lines; m/s) using NGM 6-h forecast gridded data valid 0600 17 January. Cross-section was constructed from Michigan to Louisiana, normal to the 1000-500 mb thickness field. Stippled region denotes apparent area of CSI. Right portion of stippled area actually shows convective instability where theta-e lines bend back, i.e., decrease with height, from about 850 to 700 mb).

The release of CSI can result in the development of multiple bands of heavier precipitation within a general area of lighter precipitation. The heavier precipitation bands apparently are caused by the enhanced slanted upward motion, whereby areas of lighter precipitation between bands may be due to a downward component in the mesoscale CSI circulation superimposed on the larger scale ascent field. CSI bands normally are about 100-300 miles in length and oriented parallel to the thermal wind (i.e., thickness field). Cross-sections to evaluate CSI should be made normal to 1000-500 (or 850-300) mb thickness lines.


Frontogenetical forcing has been associated with mesoscale bands of enhanced precipitation within larger synoptic-scale storm systems (e.g., Locatelli et al. 1995; Shields et al. 1991). Frontogenesis (frontolysis) refers to an intensification (weakening) of a thermal gradient at the surface or aloft. Frontogenesis acts to destroy thermal wind balance, so an atmospheric adjustment must take place to restore the balance. The adjustment is achieved through patterns of rising and sinking motion (vertical circulation) associated with the ageostrophic wind. The upward component of the frontogenetical circulation in low levels often is located near the warm side of the intensifying thermal gradient, i.e., displaced slightly away from the zone of maximum low-level frontogenesis. The ascent produces adiabatic cooling of air parcels. Meanwhile, parcels sink and warm adiabatically near the cool side of the thermal ribbon. Typically, frontogenetical zones are sloped with height toward cold air within baroclinic winter systems. Therefore, the responsive circulation may include sloped vertical motion, with ascent aloft often superimposed with the low-level maximum frontogenesis region. The horizontal ageostrophic components of this circulation consist of an acceleration of air parcels from cold to warm air in low levels and from warm to cold air at upper levels. This circulation is "thermally direct," and acts to relax the temperature gradient that frontogenesis attempts to strengthen. The opposite effect occurs for frontolysis.

In the 17 January snowstorm, frontogenesis apparently played an important role in forcing mesoscale vertical motion. For example, frontogenesis was evident from Indiana to Arkansas at 850 mb at 0000 UTC 17 January (Fig. 7), coincident with an area of enhanced precipitation (not shown). Gridded model forecast data (not shown) also indicated a zone of frontogenetical forcing in the 1000-700 mb layer across Kentucky during the period of heavy precipitation between 0600 and 1200 UTC. In addition, low-level convergence over Kentucky may have contributed to deformation of the low-level thermal gradient, resulting in a frontogenetical contribution to the overall vertical motion field. However, while a strong thermal gradient existed at 850 mb across the Ohio Valley, the exact role of deformation in enhancing frontogenesis and subsequent vertical motion near the heavy snow band over northern Kentucky is still being examined. Maximum frontogenesis values during the 0600-1200 UTC period (not shown) were predicted south of Kentucky and the main snow band, which likely contributed to the development of a convective squall line across Tennessee and the Lower Mississippi Valley.

Frontogenesis at 850 mb at 0000 UTC 17 January 1994 Fig. 7: Frontogenesis function at 850 mb computed from observed data at 0000 UTC 17 January. Units are 10 (to the -10 power) K/m-s.


Upright convection can occur within warm advection/overrunning situations in the winter season. Colman (1990) studied this phenomena in detail. He suggested that these thunderstorms can occur in an environment characterized by stable surface air, if strong baroclinicity, warm advection, and lift are present above a shallow, but significant inversion. Upstream source air may be even more conducive for thunderstorms and exhibit convective instability (theta-e decreasing with height). The result is the development of "elevated" upright convection from the top of the inversion.

In the 17 January snowstorm case, embedded thunderstorms (with thunder and lightning) occurred periodically for a few hours over north-central Kentucky within the widespread precipitation, which enhanced snowfall rates. The Paducah sounding (Fig. 5) showed that very stable surface air existed below a significant temperature inversion. Strong isentropic lift/warm advection also were present, along with convective instability upstream from Kentucky (right portion of stippled area at 0600 UTC in Fig. 6). Finally, the Total Totals index at Little Rock at 0000 UTC was 51 indicating the potential for elevated convection. The CSI circulation also may have contributed to thunderstorm development but convective instability likely was more important (see Bennetts et al. 1988). Thunderstorms dissipated later in the event as the temperature inversion was eroded by cooling aloft.


Diabatic processes can have substantial dynamic and thermodynamic effects in the atmosphere. In particular, significant melting of snow can cause atmospheric cooling as ice crystals aloft fall into a warm layer above 0 C. This cooling then can result in a relatively deep (up to 1 km) isothermal layer at or below 0 C (Szeto et al. 1988). A saturated isothermal layer is important for heavy precipitation production since a larger absolute moisture content will exist in the layer than one in which temperature, and thus mixing ratio, decrease with height.

The melting process also creates a mesoscale circulation driven by a melting-induced temperature gradient near a distinct rain-snow boundary (Szeto et. al 1988; Lin and Stewart 1986). This circulation (Fig. 8), analogous to an elevated sea breeze, consists of descent on the warm (rain) side of the boundary and ascent on the cool (snow) side, resulting in a localized band of enhanced upward motion and snowfall. The circulation will persist as long as significant melting and the rain-snow boundary persist (Stewart and Macpherson 1989). The small-scale circulation, like CSI, is superimposed on the larger scale vertical motion regime.

Schematic of Indirect Thermal Circulation Due to the Melting Process Near the Zero Degree Celsius Isotherm Fig. 8: Schematic of indirect thermal circulation due to melting processes near the 0 deg C isotherm. Dashed lines are constant temperature (isotherms; deg C) with colder air to the right, while solid lines are perturbation stream functions for the rain/snow boundary at 1 hr into the simulation (adapted from Szeto et al. 1988).

The melting process may have played an important role in the 17 January 1994 snowstorm, as the Paducah 0000 UTC sounding (Fig. 5) indicated an above freezing layer from 900-750 mb while, at the same time, colder air existed north and east of Paducah. This created a distinct rain-snow boundary across northern Kentucky late on the 16th and early on the 17th, which contributed to a narrow zone of heavy precipitation (reflectivity values greater than 35 dBZ) oriented east-west on the Louisville-Ft. Knox (KLVX) WSR-88D (not shown). Sleet was occurring within the band, with heavy snow on the band's northern (back) edge, and only light freezing rain immediately south (ahead) of the band. By 1200 UTC, snow was occurring over most of Kentucky with embedded heavy snow bands (see WSR-88D example).

The enhanced lift and snowfall rates just north of the transition zone and the lighter freezing rain just south of the band are consistent with and appear to be at least partly due to this mesoscale melting-induced circulation. Along with strong adiabatic cooling from strong ascent, the melting may have helped eventually cool the warm layer aloft resulting in a subfreezing isothermal layer and a changeover to heavy snow across most of Kentucky during the morning of the 17th. This cooling acted to diminish and terminate the melting-induced circulation later in the snowstorm event.


Heavy snow rates are associated with large snowflakes consisting of dendritic crystals (Auer and White 1982). It is suggested that heavy snow can occur in events when the level of maximum vertical motion (i.e., the level of non-divergence) is near the level of maximum growth rate of dendritic crystals. This growth rate level typically occurs at temperatures of -13 to -17 C. A NGM-derived time-height cross-section for Louisville (Fig. 9) showed that these two levels were coincident between 0600 and 1200 UTC 17 January (the 6-12-h model forecast), i.e., the time period in which very heavy snow and large snowflakes occurred in Louisville.

Time-Height Vertical Cross-Section of Temperature and Vertical Motion at Louisville at 0000 UTC 17 January 1994 Fig. 9: Time-height cross-section for Louisville, KY from the 0000 UTC 17 January NGM. Isotherms (solid; deg C) and vertical motion (dashed; microbars/s) are shown. Forecast hour (time) is along the horizontal axis (x-axis) and increases to the left (e.g., 600 is the 6-h forecast valid 0600 UTC). Pressure (in mb) is along the vertical axis.


Several mechanisms combined to produce very strong ascent and record-breaking snowfall across the northern half of Kentucky on 16-17 January 1994. Strong isentropic lift/warm air advection produced significant precipitation over a large area, although other processes apparently concentrated the heaviest snowfall into a mesoscale band over northern Kentucky.

The banding mechanisms in this event likely did not act independently from one another in producing strong ascent and heavy snow, although the exact mechanism interaction is not completely clear. However, meteorologists must be aware that banding mechanisms exist within winter storm systems, and be able to recognize when these phenomena may act to significantly alter the ultimate accumulation of rainfall and snowfall. Most of these phenomena can be evaluated qualitatively from observed and gridded model data. However, somewhat limited model resolution often prevents accurate model depiction of mesoscale organization and evolution, despite better prediction of general precipitation areas within winter storms. The WSR-88D can help greatly from a short-term forecast standpoint. It clearly can reveal 1) banded reflectivity structure in winter storms, 2) the presence, location, and trend of the bright band/melting layer, and 3) inversions aloft using the VAD Wind Profile (VWP) and radial velocity data. Training and experience are crucial in an effort to understand mesoscale atmospheric structure and the implications on storm evolution and precipitation.


Auer, A.H. Jr. and J.M. White, 1982: The combined role of kinematics, thermodynamics, and cloud physics associated with heavy snowfall episodes. J. Meteor. Soc. Japan, 60, 500-507.

Bennetts, D.A., J.R. Grant, and E. McCallum, 1988: An introductory review of fronts. Part I: Theory and observations. Meteor. Magazine, 117, 357-371.

Bennetts, D.A. and B.J. Hoskins, 1979: Conditional symmetric instability-a possible explanation for frontal rainbands. Quart. J. Roy. Meteor. Soc., 105, 945-962.

Colman, B.R., 1990: Thunderstorms above frontal surfaces in environments without positive CAPE. Part I: A climatology. Mon. Wea. Rev., 118, 1103-1121.

Emanuel, K.A., 1988: Observational evidence of slantwise convective adjustment. Mon. Wea. Rev., 116, 1805-1816.

Emanuel, K.A., 1983: On assessing local conditional symmetric instability from atmospheric soundings. Mon. Wea. Rev., 111, 2016-2033.

Garcia, C. Jr., 1994: Forecasting snowfall using mixing ratios on an isentropic surface. NOAA Technical Memorandum NWS CR-105, U.S. Dept of Commerce/NOAA/NWS, 31 pp.

Hakim, G.J and L.W. Uccellini, 1992: Diagnosing coupled jet-streak circulations for a northern Plains snow band from the operational Nested Grid Model. Wea. Forecasting, 7, 26-48.

Lin, C.A. and R.E. Stewart, 1986: Mesoscale circulations induced by melting snow. J. Geophys. Res., 91, 13299-13302.

Locatelli, J.D., J.E. Martin, J.A. Castle, and P.V. Hobbs, 1995: Structure and evolution of winter cyclones in the central United States and their effects on the distribution of precipitation; Part III: The development of a squall line associated with weak cold frontogenesis aloft. Mon. Wea. Rev., 123, 2641-2662.

Moore, J.T., 1993: Isentropic analysis and interpretation: Operational applications to synoptic and mesoscale forecast problems. National Weather Service Training Center Training Manual, Kansas City, MO, 99 pp.

Moore, J.T. and M.A. Kaster, 1993: Physical processes organizing wintertime frozen precipitation events in the Midwest. Preprints, 13th Conf. on Weather Analysis and Forecasting, Vienna, VA, AMS, 512-515.

Shea, T.J. and R.W. Przybylinski, 1993: Assessing vertical motion fields in a winter storm using PCGRIDDS. Preprints, 13th Conf. on Weather Analysis and Forecasting, Vienna, VA, AMS, 10-14.

Shields, M.T., R.M. Rauber, and M.K. Ramamurthy, 1991: Dynamical forcing and mesoscale organization of precipitation bands in a Midwest winter cyclonic storm. Mon. Wea. Rev., 119, 936- 964.

Stewart, R.E. and S.R. Macpherson, 1989: Winter storm structure and melting-induced circulations. Atmos.-Ocean, 27 (1), 5-23.

Szeto, K.K., R.E. Stewart, and C.A. Lin, 1988: Mesoscale circulations forced by the melting of snow in the atmosphere. Part II: Application to meteorological features. J. Atmos. Sci., 45, 1642-1650.

Uccellini, L.W. and P.J. Kocin, 1987: The interaction of jet streak circulations during heavy snow events along the east coast of the United States. Wea. Forecasting, 2, 289-308.

Back to Research and Papers is the U.S. government's official web portal to all federal, state and local government web resources and services.