Mesoscale Gravity Waves and Their Environment in the Central United States during STORM-FEST

Steven E. Koch Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina

Search for other papers by Steven E. Koch in
Current site
Google Scholar
PubMed
Close
and
Leanne M. Siedlarz Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina

Search for other papers by Leanne M. Siedlarz in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

In an effort to better understand mesoscale gravity waves in winter storms in the central United States—their frequency of occurrence, wave characteristics, the general conditions under which they occur, and their effects upon the weather—mesoscale surface and rawinsonde data as well as radar and satellite imagery collected during the Storm-scale Operational and Research Meteorology–Fronts and Experimental System Test are analyzed. In addition, factors affecting the ability of objective surface map analysis to properly represent the waves are investigated.

Thirteen coherent pressure pulse events with amplitudes of 0.2–4.0 mb and periods of 1–6 h were identified in the surface pressure data during the 6 weeks of the project, involving 34% of the total hours investigated. A variety of wave types occurred, including wavelets, wave trains, and singular waves. The three largest amplitude events were analyzed in detail using autospectral analysis and a Barnes time-to-space conversion objective analysis of bandpass-filtered mesonet data. All three events displayed high perturbation pressure–wind covariances (pu*′), consistent with a gravity wave explanation for the disturbances (u* is the wind component in the direction of wave propagation). The pu*′ values were closely related to the strength of the wave amplitudes. The waves found in these events displayed mean phase velocities of 19.9–27.9 m s−1, wavelengths of 200–260 km, and periods of 2.3–3.5 h.

Wave crests appeared to be closely aligned with associated rainbands throughout their lifetimes, suggesting that a codependency existed. Some of the waves were evident before the rainbands formed, indicating that the precipitation developed in response to the waves, though this was not true for all of the waves. Values of pu*′ decreased during the development stage of deep convection, but high covariance between the pressure and wind fields redeveloped as the thunderstorms and incipient gravity wave matured into a stable, coupled mesoscale convective system.

Three of the four wave events displaying the largest amplitudes occurred primarily on the cool side of a stationary front in an environment in which a jet streak was approaching an inflection axis in a diffluent height field downstream from an upper-level trough. The waves also extended some distance into the warm sector in the presence of a statically stable lower troposphere, suggesting wave ducting was operative. The results indicate that this conceptual model for the wave environment should prove useful as a tool for forecasting the most significant mesoscale gravity wave events.

Corresponding author address: Dr. Steven E. Koch, Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Campus Box 8208, Raleigh, NC 27695-8208.

Email: Steve_Koch@ncsu.edu

Abstract

In an effort to better understand mesoscale gravity waves in winter storms in the central United States—their frequency of occurrence, wave characteristics, the general conditions under which they occur, and their effects upon the weather—mesoscale surface and rawinsonde data as well as radar and satellite imagery collected during the Storm-scale Operational and Research Meteorology–Fronts and Experimental System Test are analyzed. In addition, factors affecting the ability of objective surface map analysis to properly represent the waves are investigated.

Thirteen coherent pressure pulse events with amplitudes of 0.2–4.0 mb and periods of 1–6 h were identified in the surface pressure data during the 6 weeks of the project, involving 34% of the total hours investigated. A variety of wave types occurred, including wavelets, wave trains, and singular waves. The three largest amplitude events were analyzed in detail using autospectral analysis and a Barnes time-to-space conversion objective analysis of bandpass-filtered mesonet data. All three events displayed high perturbation pressure–wind covariances (pu*′), consistent with a gravity wave explanation for the disturbances (u* is the wind component in the direction of wave propagation). The pu*′ values were closely related to the strength of the wave amplitudes. The waves found in these events displayed mean phase velocities of 19.9–27.9 m s−1, wavelengths of 200–260 km, and periods of 2.3–3.5 h.

Wave crests appeared to be closely aligned with associated rainbands throughout their lifetimes, suggesting that a codependency existed. Some of the waves were evident before the rainbands formed, indicating that the precipitation developed in response to the waves, though this was not true for all of the waves. Values of pu*′ decreased during the development stage of deep convection, but high covariance between the pressure and wind fields redeveloped as the thunderstorms and incipient gravity wave matured into a stable, coupled mesoscale convective system.

Three of the four wave events displaying the largest amplitudes occurred primarily on the cool side of a stationary front in an environment in which a jet streak was approaching an inflection axis in a diffluent height field downstream from an upper-level trough. The waves also extended some distance into the warm sector in the presence of a statically stable lower troposphere, suggesting wave ducting was operative. The results indicate that this conceptual model for the wave environment should prove useful as a tool for forecasting the most significant mesoscale gravity wave events.

Corresponding author address: Dr. Steven E. Koch, Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Campus Box 8208, Raleigh, NC 27695-8208.

Email: Steve_Koch@ncsu.edu

1. Introduction

How important are gravity waves insofar as they might affect the weather, and under what conditions do they occur? As observational technologies and numerical weather prediction models improve, there is a growing awareness of the importance of mesoscale gravity waves. Mesoscale models are increasingly being used to study such waves and their interactions with precipitation bands (Powers and Reed 1993; Pokrandt et al. 1997; Powers 1997; Kaplan et al. 1997; Koch et al. 1998). The Meso-Eta Model is fully capable of producing significant gravity waves (Barnes et al. 1996). New observing systems such as the Automated Surface Observing System (ASOS), wind profilers, WSR-88D Doppler radar, and the new series of Geostationary Operational Environmental Satellites (GOES) are providing increased information on a more frequent basis and at resolutions not previously available nor economically feasible (Friday 1993). Wind profiler data have been used recently to study the vertical structure of mesoscale gravity waves (Ralph et al. 1993; Trexler et al. 1998), so it would seem feasible to use the wind profiles obtainable from the WSR-88D radar in a similar manner. The development of the ASOS system has led to a significant decrease in average station spacing across the United States, and more importantly for gravity wave analysis, the ability to obtain digital, high temporal resolution (1–5 min) data. With the addition of such data and development of objective analysis methods that exploit the temporal information, it is now conceivable that mesoscale gravity waves can be analyzed in real time (Koch and O’Handley 1997, hereafter KO97).

Most case studies of mesoscale gravity waves (wavelengths of 50–500 km) have considered only large-amplitude waves (>2.0 mb) despite the fact that these events are not that common (Uccellini and Koch 1987). Pecnick and Young (1984) claim that such events occur in the Midwest perhaps only once or twice per year, thus suggesting this type of gravity wave to be extremely rare. On the other hand, a gravity wave train in Montana studied by Koch et al. (1988) displayed amplitudes of only 0.6 mb, yet it was instrumental in organizing thunderstorms into a destructive mesoscale convective complex. Relatively weak gravity waves with amplitudes <1 mb also played a significant role in the 1994 Palm Sunday tornado outbreak (Koch et al. 1998).

Einaudi et al. (1989) state that the frequency of occurrence, amplitude, period, horizontal wavelength, and vertical structure of gravity waves together determine the significance of these waves in tropospheric dynamics and their importance as a forecasting consideration. Unfortunately, our knowledge of the climatology of mesoscale gravity waves in the troposphere is practically nonexistent and restricted to information obtained from very-small-scale networks of pressure sensors (Einaudi et al. 1989; Hauf et al. 1996). This has prevented understanding of the wave environment and the responsible wave generation mechanisms. A recent climatological study by Grivet-Talocia et al. (1999) conducted in the central United States found that coherent pressure perturbations occur 21% of the time during the fall and winter months, and that wave amplitude and period are correlated.

The assumption underlying these gravity wave detection approaches is that, since ground-based pressure records detect changes in the total mass of the overlying atmospheric column (and nonhydrostatic vertical accelerations), microbarographs should readily detect gravity waves. However, this belief has not been subjected to rigorous testing. Indeed, the present study casts doubt on this tacit assumption, which in its essence is that wave phase surfaces do not display a strong shift over the column depth.

Of particular note is that no climatological investigation has yet addressed the mesoscale gravity wave environment in a comprehensive and quantitative manner. Uccellini and Koch (1987, hereafter UK87) conducted a 13-case sample of mesoscale gravity waves reported as of that time. They proposed that such events occur systematically in an environment characterized by the approach of an upper-level jet toward the inflection axis in a highly diffluent flow pattern, and north of a warm front or stationary front at the surface. They argue that this conceptual model (see Fig. 1 in KO97) suggests an important role played by geostrophic adjustment in mesoscale gravity wave formation and wave ducting in gravity wave maintenance. Unfortunately, quantitative dynamical analysis is still largely lacking in observational case studies of mesoscale gravity waves.

The current study uses special observations collected during the Storm-scale Operational and Research Meteorology–Fronts and Experimental Systems Test (STORM-FEST) in a climatological investigation of mesoscale gravity waves, their synoptic-scale environment, and their general importance to clouds and precipitation. This field experiment was held from 1 February to 15 March 1992 for the purpose of investigating the structure and evolution of fronts, embedded precipitation, and mesoscale phenomena in winter storms over the central United States (Szoke et al. 1994). An outgrowth of the STORM-FEST studies has been the development of a new conceptual model for lee cyclones and attendant mesoscale precipitation systems in the central United States (Hobbs et al. 1996). One of the main components of this model is the “pre–dry trough rainband.” Martin et al. (1995) argue that this band is the result of synoptic-scale lifting associated with warm advection and a frontogenetic circulation at the leading edge of the warm frontal zone acting upon a broad region of convective instability ahead of the surface dryline. However, since the pre–dry trough rainband is quite narrow (∼75 km) and may in reality be composed of multiple precipitation bands, these mechanisms do not provide a fully satisfactory explanation of the observations. The present study examines the possible roles played by gravity waves in this regard.

A second component of their model is the precipitation band associated with the cold front aloft (CFA), a midlevel frontal zone that becomes decoupled from the surface front and dryline. As the CFA advances ahead of the dryline, it tips forward over an upward-sloping region of high equivalent potential temperature to form a warm occluded-like structure. According to Locatelli et al. (1995), two processes then act to generate the rainband in the vicinity of the leading edge of the CFA. The first hypothesized process is quasigeostrophic frontogenesis. It is questionable how this mechanism can explain a narrow band of deep convection along the CFA, such as the squall line in Oklahoma and Texas on 9 March 1992 during STORM-FEST that they argued was forced by lifting along the CFA. The second process is low-level ageostrophic convergence resulting from a pressure gradient created by the advection of cold, dense air aloft. Locatelli et al. (1997) discuss how the passage of CFAs is typically associated with rapid pressure changes at the surface. Their application of the linear divergence equation to representative barograms indicates that the strongest convergence should occur at the time when surface pressure is increasing most rapidly, a prediction, which is, for all practical purposes, identical to that for linear gravity waves. However, a closer inspection of the barograms presented by them suggests multiple pressure surges in some cases (e.g., Fig. 4b of Locatelli et al. 1997) and the pressure–wind covariance during these surges was not examined. Gravity waves have yet to be considered within the context of the CFA rainband phenomenon.

The present study seeks to attain these objectives:

  1. determine mesoscale gravity wave characteristics over the central United States during the STORM-FEST period, using the surface mesonet data;

  2. assess the importance of diagnosed mesoscale gravity waves in the formation and organization of precipitation bands, and possible feedback effects of convection on the incipient waves; and

  3. determine whether the features contained in the UK87 conceptual model of the wave environment were actually present during such events in STORM-FEST. Depending on its applicability, this model would allow a forecaster to easily identify ideal conditions for mesoscale gravity wave development in advance.

A brief review of mesoscale gravity wave structure, genesis mechanisms, and typical analysis methods are presented in section 2. Data and methodology used in this study are explained in section 3. The test results of analysis sensitivity to bandpass filter interval selection and wave-tracking techniques are presented in section 4. Detailed analyses of the three cases showing the strongest wave amplitudes, and the applicability of the UK87 conceptual model of the environment of mesoscale gravity waves to the STORM-FEST dataset, are presented in section 5.

2. Background

Pertinent gravity wave concepts and analysis techniques are briefly discussed. In particular, the identification of gravity wave signals in surface mesoanalyses and wave generation and maintenance mechanisms are reviewed.

a. Gravity wave structure

Eom (1975) developed a simple model for a trapped, linear, plane gravity wave with no basic current. This model (Fig. 1a) has been widely applied in case studies of mesoscale gravity waves (Uccellini 1975; Stobie et al. 1983; Pecnick and Young 1984; Bosart and Sanders 1986; Koch et al. 1988; Ramamurthy et al. 1993). The lack of any tilt in the vertical plane in this model is the result of wave trapping, in which upwardly and downwardly propagating waves are reflected and constructively interfere. In general, decreased cloudiness and stronger winds at the surface, from the direction in which the wave is propagating, are observed in the pressure trough (L). Conversely, enhanced cloudiness and winds that have veered relative to those during the passage of the trough typically accompany the pressure ridge (H). Convergence occurs at the wave nodal point between the trough and the ensuing ridge, while divergence is observed at the nodal point ahead of the trough. Hence, in the absence of tilt, deep cloudiness and the greatest precipitation rate associated with this idealized wave most often occur between the upward motion maximum and the trailing wave ridge axis.

Koch and Golus (1988) note that if the wave tilts upstream with height, the maximum precipitation rate can occur during or even slightly after ridge axis passage (Fig. 1b). Schneider (1990) remarked that passage of a large-amplitude wave crest was followed by heavy snow, ice pellets, thunder, and lightning. Pecnick and Young (1984) also observed an upstream tilt in the wave they analyzed (opposite to the propagation direction). The presence of such structures (i.e., multiple vertical wavelengths) could drastically reduce the usefulness of surface pressure sensors in detecting gravity waves.

Another complication of the simple Eom model occurs when a single negative pressure perturbation (p′) is not accompanied by a subsequent positive p′. The rapid dissipation of midlevel clouds or end to an extended period of precipitation in such situations has been attributed to subsidence ahead of the pressure minimum (Tepper 1951; Bosart and Cussen 1973; Bosart and Seimon 1988; Ramamurthy et al. 1993). These events bear similarity to large-amplitude, form-preserving solitary waves, in which there exists a balance between the steepening effect of nonlinearity and the flattening effects of frequency dispersion (Lin and Goff 1988; Ramamurthy et al. 1993; Rottman and Einaudi 1993).

b. Wave detection techniques

Analysis techniques typically used to detect gravity wave signatures at the surface consist of applying a bandpass filter to the time series to reveal the presence of the waves as pressure perturbations (p′). The filter may be chosen based on the results from power spectrum analysis of the pressure data (Stobie et al. 1983; Koch and Golus 1988), or the frequency band to be passed can be chosen based on a priori knowledge of the wave period (KO97). An alternative to bandpass filtering is the wavelet technique, which finds its greatest use in unsteady, nonmonochromatic wave cases (Grivet-Talocia et al. 1999). Either objective map analysis of the p′ fields or cross-spectral analysis (Stobie et al. 1983; Koch and Golus 1988) is then used to find the propagation velocity of the wave in the selected frequency band.

Changes in the cloud and precipitation fields are also frequently used to aid in the precise determination of gravity wave location and shape. If the descent induced by the passage of a wave of depression extends throughout a deep layer, cloud depressions (seen as warmer areas in infrared satellite imagery) can be expected (Tepper 1951; Pecnick and Young 1984). Thus, cloudbands observed in satellite imagery can be used in instances in which the waves propagate in tandem with the bands to assist in the identification of gravity waves as they propagate between widely spaced barograph stations (Koch et al. 1988).

c. Gravity wave generation mechanisms

Forcing mechanisms for mesoscale gravity waves have been attributed to a wide range of processes, including geostrophic adjustment, shearing instability, and convective and orographic forcing mechanisms. The UK87 conceptual model, which is depicted schematically in KO97, argues that geostrophic adjustment is a viable candidate for the wave generation mechanism in many instances, according to the following logic. When a jet streak propagates from the base of a negatively tilted trough to a diffluent region near the axis of inflection in the height field, it separates from the geostrophic (gradient) wind maximum. The cross-stream ageostrophic flow then becomes leftward directed in the exit region of the geostrophic wind maximum. Such flow is unbalanced in the sense that air parcels would accelerate in the exit region when quasi- and semigeostrophic theory dictates that the opposite kind of behavior is needed to maintain thermal wind balance1 (Koch and Dorian 1988). UK87 hypothesized that mesoscale inertia–gravity waves are generated downstream of this region as the result of geostrophic adjustment. More recent case studies (e.g., Schneider 1990; Ramamurthy et al. 1993) offer qualitative support for the UK87 model. A more general qualitative assessment of the applicability of the UK87 conceptual model to the local wave environment is made in the present study using the STORM-FEST data.

The focus of the current investigation is on surface analysis of gravity waves and on whether they tend to occur in the kind of environment described by the UK87 model. The roles of shearing instability, topographic forcing, and geostrophic adjustment as possible gravity wave generation mechanisms are not examined. However, the interactions between gravity waves and deep convection are examined to some extent in this study. Jin (1997) has used a mesoscale model to perform a quantitative analysis of all of these mechanisms for the three strongest gravity wave events in this study.

Gravity wave–convection interactions are poorly understood. Uccellini (1975) concluded that mesoscale gravity waves are capable of releasing convective instability and initiating thunderstorm development in areas that have the necessary amount and proper vertical distribution of moisture. Stobie et al. (1983) describe a case in which rapid amplification of gravity waves and the growth of convection occurred simultaneously several hours after the waves were first detected, suggesting that convection can play a role in strengthening gravity waves over a long period of time.

Convection can lead to gravity wave generation either when latent heat is released in a shear flow (Lin 1994) or by wave–conditional instability of the second kind (CISK; e.g., Raymond 1983). According to wave-CISK theory, precipitating convection and the gravity wave form a coupled system, whereby the wave supplies the divergence and convergence patterns necessary for maintenance of the convection, while the gravity waves are, in turn, supplied energy by the divergence of convective mass fluxes. Wave-CISK was suggested as the agent responsible for generating gravity waves in studies by Powers and Reed (1993), Powers (1997), and Koch et al. (1998).

Convection can also alter an incipient gravity wave. Koch et al. (1988) found that the typical pressure–wind relationship for a linear gravity wave was altered in the local vicinity of a severe thunderstorm. As a gust front developed, the horizontal wind perturbation, which earlier had been in phase with the pressure perturbation (as in Fig. 1a), advanced ahead of the pressure field and became associated with the pressure-nodal line (as in Fig. 2b). This is 90° out of phase from the typical gravity wave relationship. In addition, the wave propagation velocity became slower and deviated to the right of its former direction. This large reduction in pu*′ correlation and change in phase velocity were also accompanied by an increase in wave amplitude, reduction in wavelength, and the appearance of nonplanar (arc-shaped) wave fronts. The original gravity wave signal, however, remained intact beyond the local influence of the thunderstorm. Convection did not appear to significantly affect the incipient gravity wave unless a gust front was present.

Johnson and Hamilton (1988) present a conceptual model of the surface fields surrounding a squall line with a trailing stratiform precipitation region, based upon their detailed analysis of surface mesonetwork and radar data (Fig. 2a). Southeasterly winds gradually increase in speed, reaching a peak during the passage of the wake low. The maximum convergence occurs behind the wake low in association with an abrupt increase in pressure. It is highly interesting that this phase relationship is consistent with gravity wave theory, though their station spacing of ∼50 km may be too large to be able to draw a definitive conclusion. By contrast, the classic Fujita (1955) model of a thunderstorm shows convergence of air into the center of the wake low (Fig. 2b). Our study addresses the applicability of these conceptual models to those STORM-FEST events in which strong convection developed from an incipient gravity wave.

d. Gravity wave maintenance mechanisms

The three processes that can maintain gravity wave coherence are a source for wave energy, a mechanism for preventing energy loss, and solitary wave processes (the latter process maintains wave shape as nonlinearity counteracts the effects of wave dispersion). Wave-CISK and shearing instability are the most common sources for wave energy. Energy-loss prevention occurs due to either wave ducting or wave overreflection in the presence of a critical level. Mesoscale gravity waves quickly lose most of their energy before traveling even one horizontal wavelength due to upward energy propagation (Lindzen and Tung 1976; Pecnick and Young 1984). Yet, these kinds of waves are often observed and tracked for a significant period of time, in excess of 12 h in some cases (Uccellini and Koch 1987). Requirements for a highly reflective wave duct have been outlined in the linear theory by Lindzen and Tung (1976); key features include a stably stratified lower layer of sufficient depth beneath a conditionally unstable layer containing a critical level. However, Lin and Goff (1988) note that these requirements (notably, the critical level) are not always met by observed solitary-like waves.

In this study, satellite and radar imagery were used to help determine if convection existed before the gravity wave events or if any developed after the wave was evident. Otherwise, gravity wave maintenance mechanisms were not addressed, primarily because analysis of the mesoscale model fields required for such an investigation is beyond the stated purposes of this study.

3. Methodology

Following a description of the data used in this study, we discuss methods developed for identifying potential wave events in the time series data from the surface mesonetwork, the environment surrounding those events, and gravity wave characteristics.

a. Data

Data sources used in this investigation include surface mesonetwork, special rawinsondes, radar mosaics, and GOES satellite imagery, which were collected as part of the intense data gathering that occurred during STORM-FEST. The majority of the data used in this study was retrieved through the Cooperative Distributed Interactive Atmospheric Catalog System (CODIAC) of the University Corporation of Atmospheric Research Office of Field Project Support.

Surface data included 5-min observations from ASOS, the Illinois Climate Network (ICN), and the National Center for Atmospheric Research (NCAR) Portable Automated Mesonet (PAM), 20-min data from the Aviation Weather Observation System (AWOS), and National Weather Service (NWS) barograph data from the National Climate Data Center (Fig. 3). The barograph data were manually digitized to 15-min intervals, linearly interpolated to 5-min intervals, and combined with the data from the other systems for the purpose of creating a single synthesized dataset. Only 45 of the ASOS stations had been installed by the end of STORM-FEST, primarily in Kansas and Oklahoma. The AWOS data, containing observations from a network of 48 airports in the STORM-FEST domain, were used only in the analysis of the 14 February gravity wave case to fill in gaps resulting from the absence of ASOS data during part of that event. Linear interpolation was used to produce 5-min data from this 20-min-resolution database. The ICN 19-station network, with sites throughout Illinois, was specially equipped for STORM-FEST with high-resolution microbarograph recording devices. The final component of the surface mesonetwork system was a dual network of 45 NCAR PAM II stations, 9 of which were located in a high-density cluster composing the STORM-FEST boundary layer network in northeastern Kansas, while most of the remaining stations were located in Missouri.

The satellite data archived during STORM-FEST were obtained from GOES-7, which produced infrared and visible radiometric images at 30-min intervals. Radar reflectivity data from individual NWS radars were available from CODIAC as radar mosaics (NOWrad) every 15-min during STORM-FEST. Satellite, radar, and surface mesonetwork composites were produced using Zeb software (Corbet et al. 1994).

b. Identifying potential gravity wave occurrences

The initial task was to determine the number of potential gravity wave events. Pressure traces from a majority of the stations in eastern Kansas, Missouri, and Illinois were gathered and perused, since this area was where the station density was greatest, allowing for optimum wave detection. Pressure traces from 50 or more stations for each STORM-FEST day were examined, the time between either peaks or troughs was determined, and the magnitude of the peak-to-peak pressure change (the wave amplitude) was subjectively estimated for disturbances with amplitude ≥0.2 mb and periods >1 h. Shorter period waves could not be followed without ambiguity between stations (the minimum resolvable period is ∼1.3 h assuming an average station spacing of ∼60 km and typical wave phase speed of 25 m s−1). The range of 1–4 h periods encompasses that characterizing mesoscale gravity waves (UK87), and also avoids the effects of the semidiurnal pressure oscillation and synoptic-scale processes. Wave amplitudes were plotted along with the time of wavelike activity on a map. Those days in which the pressure traces displayed wavelike tendencies with acceptable temporal and spatial continuity, periods of 1–4 h, and amplitudes ≥0.2 mb were classified as a pressure pulse event. A simple average of the pressure pulse events from each of the 50 pressure traces from each day involved in the event were used in determining the average wave amplitude.

c. Evaluation of the Uccellini and Koch (1987) gravity wave environment model

Surface and 500-mb wind and height charts at 1200 UTC were studied for each of the 41 days of STORM-FEST in an attempt to determine whether mesoscale gravity waves occurred under the general synoptic pattern described by the UK87 wave environment model. Although they used the 300-mb height fields and winds, the 500-mb charts were readily available to us and were considered to be a viable substitute in this qualitative evaluation of the applicability of the UK87 model. The three criteria examined over the STORM-FEST region were (i) a jet streak near an inflection axis, (ii) a diffluent trough, and (iii) a stationary or warm front to the south or southeast of the “wave” region.

d. Gravity wave analysis methods using surface data

This section describes more exacting methods employed to objectively determine whether the stronger pressure pulse events were truly gravity waves. Because of the great amount of quality control and data analysis required, only the three strongest pressure pulse cases were analyzed using spectral and bandpass filter analysis.

Autospectral analysis was used to determine the significant frequencies present in the pressure traces. Spectral analysis was accomplished on 80 stations from the STORM-FEST region where wave activity was evident. An example of the method is presented in Fig. 4, which shows the histogram of peak frequencies from the autospectral analysis for the 14 February wave event. A single mode is suggested, as is a frequency range of 0.004–0.010 min−1 (corresponding to a period of 100–250 min). We followed the method used in the gravity wave study by Koch and Golus (1988), whereby peak frequencies obtained from the autospectral analysis of the pressure time series were used to determine the appropriate cutoff frequencies for the Lanczos bandpass filter (Duchon 1979). The “narrow” bandpass filter was designed to include only the range of significant spectral peaks. The filter was applied to the pressure time series to obtain the pressure perturbations (p′). The filter response function shows that the most commonly observed frequencies were attenuated by 20%–30% (an unavoidable consequence of the non-Gaussian nature of the histogram and the need to avoid big side lobes in the response function).

The p′ values thus obtained for each station were plotted on a map at 30-min intervals and analyzed both subjectively and objectively, initially without the benefit of time-to-space conversion (TSC). The wave axis tracking method (Koch and Golus 1988) was used to determine the propagation velocity vectors for coherent pressure perturbations from these analyses. These spatially and temporally variable vectors were used as the advection vectors in the TSC analyses.

The filtering process applied to the pressure data was also used on the wind component in the direction of wave propagation. These wave-normal winds (u*) were calculated using wave propagation directions that were allowed to vary both spatially and temporally. Bandpass-filtered u* winds are denoted herein by u*′. The pu*′ correlation values were then calculated over a several hour period and plotted, centered on the time of wave activity for each station. A correlation above 0.32 is statistically significant at the 95% level. Gravity wave theory implies these correlation values will be positive (Fig. 1). Investigators have either implicitly or explicitly used this technique to identify gravity waves, since the pressure–wind correlation is a direct measure of the wave signal-to-noise ratio (Gossard and Sweezy 1974), which is a consequence of the impedance relation for linear gravity waves (Gossard and Hooke 1975).

The TSC Barnes scheme developed by KO97 for gravity wave analysis was adopted here. The traditional TSC principle assumes a simple geophysical phenomenon translating with a constant velocity C, but in the KO97 scheme, the phenomenon (gravity wave) is not required to be in exact steady state. Instead, their technique allows for an inverse exponential decrease in wave conservation with increasing time away from the map time. In the current study, each station generated a series of N = 21 observations (one central on-time and 20 surrounding off-time values, i.e., ±50 min from the map time). These observations were distributed along the wave propagation vector C at distances of CΔt (where Δt = 5 min is the time between successive observations). The Barnes TSC technique assumes an exponentially decreasing degree of steady state of the system over the conversion period τ = (N − 1)Δt, where τ is the period during which the time series data are converted to spatial data. KO97 suggest that for typical mesoscale gravity waves (λ ≅ 200 km) and phase speeds (C ≅ 20 m s−1), a recommended value of τ = λ/4C is ∼45 min. Values used for τ are discussed in the individual case studies and in the sensitivity test section below.

4. Sensitivity tests

This section briefly summarizes the results of various tests performed on the sensitivity of the objective analyses of the bandpass-filtered pressure fields to the choice of filter bandwidth, advection vector(s) in the TSC analysis scheme, TSC conversion interval (τ), and grid mesh. Siedlarz (1996) provides greater details.

a. Choice of filter bandpass interval

Two different filters were applied to the 14 February 1992 dataset to test the sensitivity of the pressure perturbations to the width of the filter: (i) a narrow filter as determined from the power spectral histograms (Fig. 4b), and (ii) a wider filter, identical to that used by KO97 in their analysis of this same gravity wave event (Fig. 4c). The 50% response cutoff frequencies were fN1 = 0.004 min−1 and fN2 = 0.010 min−1 for the narrow filter and fW1 = 0.004 min−1 and fW2 = 0.034 min−1 for the wide filter. These frequencies correspond to wave periods of TN1 = TW1 = 4.2 h, TN2 = 1.7 h, and TW2 = 0.5 h.

Many wave troughs and crests in the original and filtered time series from nearly 50 stations were examined to assess the impact of filter choice on phase errors resulting from the bandpass filter application. It was found that with use of the wide filter, the phase errors produced in the time series were noticeably smaller. This happened because the narrower filter eliminated the higher frequency components (periods shorter than 1.7 h). Wave amplitudes appearing in the TSC Barnes analyses (not shown) also displayed sensitivity to the choice of filter width. These results demonstrate that the effects of bandpass filter characteristics should be considered, even though a spectral histogram may suggest a smaller bandwidth, in order to lessen the broadening effect and hence potential analysis errors. Since the wide filter produced p′ values closer to the actual values deducible from the raw pressure traces, it was used in the analysis of the 14 February event. Slightly different bandpass filters were used in the other two gravity wave events, as determined by the spectral characteristics for each case.

b. Effects of advection vectors in the TSC scheme

We compared results from a customary Barnes objective analysis (Koch et al. 1983) to those from Barnes TSC analyses using both multiple advection vectors and only a single advection vector. Our results confirm the finding of KO97 that the TSC objective analysis method is demonstrably superior to the standard method in the sense that it was much easier to follow gravity waves in the TSC analyses. The conventional Barnes analysis of gravity waves with wavelengths of only 2–4 times the average station spacing produced features of diminished amplitude that disappeared and then reappeared a short time later, and were characterized by circular pressure perturbation patterns instead of wavelike ones. The TSC Barnes analyses suffered from none of these defects.

The single advection vector TSC test using a value of τ = 50 min and a grid length of 25 km produced contour shapes that were sometimes unrealistically planar, but the intensities of the perturbation maxima and wave locations were quite similar to the multiple advection vector results (Fig. 5). Even though the single advection vector displayed difficulty in accurately representing the waves as arcs, it still performed satisfactorily; however, it would probably be wise to use multiple advection vectors when the wave is known to be arc shaped. KO97 have proposed a method to predict the spatial distribution of the phase velocity from wave ducting theory. When clouds are present with the wave features, they can give some additional indication as to the actual shape of the wave.

c. Sensitivity of analysis to TSC interval and to grid length

The TSC results using two different tau values were compared—τ = 15 min (T15) and τ = 50 min (T50). The tau tests displayed the greatest analysis impacts of all the tests conducted with the TSC scheme. As shown later, for the 14 February case, the average wave speed and wavelength were determined to be 21.6 m s−1 and 200 km. The suggested value for this situation is τ = λ/4C = 40 min, meaning that T15 employed a value much below the optimal one. With τ = 15 min, not much TSC was occurring since only three observations on either side of the observation being converted were involved, whereas with τ = 50 min, 10 observations were considered on either side of the central observation, effectively creating off-time data to ±64 km from each station.

The T15 results produced more bull’s-eyes. Most of the time they were less wavelike than the T50 results, and more like the conventional Barnes results (an example appears in Fig. 20 of KO97). It was definitely easier to identify and track the waves in T50 than it was in either T15 or the conventional Barnes analyses. Our results offer additional support for the argument of KO97 that the optimal value for τ based on average mesoscale gravity wave characteristics is τ ∼ 40–50 min.

Two different grid lengths of 25 km and 40 km (the average station spacing for this test was approximately 56 km) were tested for their impact on the resulting objective analyses. Basic patterns and placements were very similar in the two tests (Siedlarz 1996).

The wide filter (details are case-dependent), a TSC interval (τ) of 50 min, and a grid mesh of 25 km were employed in all of the p′ analyses appearing hereafter. The analyses for the 14 and 17 February cases used multiple advection vectors since those waves were highly arc shaped, whereas a single advection vector was used in the more planar wave case on 9 March 1992.

5. Case studies

a. Case selection and generality of the UK87 conceptual model

Table 1 gives a summary of the ranking of the pressure pulse events according to their average amplitudes. Thirteen events were identified during the 41 days of STORM-FEST (32% of the days); alternatively, the number of hours in which a pressure pulse event was identified was 34% of the total hours investigated. Though it is uncertain how many of these events actually involved gravity waves, the detailed analyses of the three strongest events presented below provides strong evidence in support of a gravity wave interpretation for those events. Pressure traces from mesonet stations for the 13 cases in Table 1 displayed various characteristics:all events except 20 February showed some combination of singular waves of depression and elevation, “wavelets” (defined in this study as a single wave cycle occurring within 90 min), and longer-period wave trains.

The entire 41 days (not just the pressure pulse events) were examined to see whether the three synoptic features typically found by Uccellini and Koch (1987) to exist during mesoscale gravity wave events were present. Only three days showed evidence that all three criteria of UK87 existed—but importantly, they occurred on days with strong pressure pulse events (14 and 17 February and 4 March). A surface warm front or stationary front to the southeast of the wave corridor was not apparent in the 9 March case, yet significant gravity wave activity was indicated within the warm sector of the cyclone ahead of the dryline (section 5c). In fact, in each of the three gravity wave cases studied in detail below (14 and 17 February and 9 March), the analyzed waves extended some distance to the south of the warm/stationary fronts. Examination of the soundings from these three gravity wave cases showed temperature inversions present to the north of the frontal boundaries, yet the lower levels were quite stable in the warm sector in each case (no attempt was made to analyze the soundings from all of STORM-FEST). These results considered together suggest that the frontal criterion of UK87 is misleadingly simplistic, and that more careful diagnosis is needed. Of course, if such frontal boundaries are present, the necessary low-level stability required for wave ducting is pretty much assured ahead of the front. On the other hand, the mere existence of static stability is not a sufficient condition to have a mesoscale gravity wave event, since strong stratification is present more often than not in the winter months over the central United States, yet pressure pulse events were found on only 32% of the days.

The results for using just one of the criteria—a 500-mb jet streak near an inflection axis in the 500-mb height field—produced no false alarms. The contingency table for this single criterion (Table 2) suggests that a jet streak approaching the inflection axis is a sufficient—but not necessary—condition for a gravity wave to occur. A two-sample t-test performed on the pressure pulse events of Table 1 showed that at the 95% level of significance, a larger wave amplitude occurred when a jet streak was near the inflection axis (i.e., closer to the inflection axis than either the trough or ridge axes). Logistic regression on the events suggested that for an average pressure amplitude ≥0.35 mb there is a 93% likelihood the jet streak criteria will also be met.

No wave events of large amplitude (average values >2.0 mb) occurred during STORM-FEST, though individual waves did exhibit amplitudes as large as 2–4 mb. Even though many of the perturbation amplitudes were rather weak, this certainly does not mean that the gravity waves were insignificant as far as their impact upon the weather is concerned, as is shown in the following three case events.

b. 14–15 February 1992

1) Synoptic overview

The surface analysis over the STORM-FEST region at 1530 UTC on 14 February (Fig. 6a) shows a warm front extending to the east of a surface cyclone in southwestern Kansas and a dryline over the Texas Panhandle. These features drifted eastward with time, such that by 9 h later (Fig. 6d), the cyclone was positioned over southeastern Kansas. Frontal overrunning led to a large area of light precipitation north of the warm front in Missouri and Illinois during the morning (Figs. 6a,b). Rapid occlusion occurred during this 9-h period, as evidenced in the surface fields and satellite imagery. Initially, the head of the telltale comma cloud was bent back into eastern Colorado, and by 2130 UTC it was positioned along the Kansas–Nebraska border with a pronounced dry wedge over Oklahoma (Fig. 6c). This dry air intrusion developed as air descended the lee slopes of the Rocky Mountains in association with an upper-level jet streak (Jin 1997), which at 1200 UTC was positioned in the Texas Panhandle near the inflection axis between a downstream ridge over Missouri and a diffluent trough axis in eastern New Mexico.

The cloudband associated with gravity wave B first appeared at 1330 UTC in the Oklahoma Panhandle, which puts it in the dry air intrusion, north of the developing warm front, and east of the surface low pressure center. This cloudband subsequently expanded and occupied an increasing fraction of the area cleared by the dry air intrusion. Radar echoes began to appear within this band by ∼1600 UTC in south-central Kansas, though precipitation was not detected at the ground for another 3 h. As the rainband (RB 1) intensified just ahead of the gravity wave trough (Fig. 6b), the trough began to develop a bowed character along the sharp back edge of the cloudband (Fig. 6c). Notice an isolated precipitation feature that appears in the southern part of the bowed rainband at 2130 UTC. This thunderstorm produced very large hail and numerous cloud-to-ground lightning strikes. The bowing effect became more pronounced owing to the continued development of convection in this part of the rainband as it moved into Missouri. A second convective rainband (RB2) appeared by 0030 UTC to the southwest of the original rainband (RB1), and still within the original clear slot (Fig. 6d). This second band traveled along just behind the first one for the next several hours.

2) Gravity wave analysis

Objective analyses of the pressure perturbation fields appear in Fig. 7 at 90-min intervals starting at 1830 UTC. Three gravity waves were detected in these analyses. Table 3 shows average values for the horizontal wavelength, period, and phase velocity (and standard deviations) for each wave as determined from the objective analyses. The three waves displayed quite similar characteristics. Waves A− and B− followed basically the same path extending northeastward from southwestern Kansas (Fig. 8), that is, along and to the north of the warm front.2 Wave C− was a weak wave of depression that was observed behind the dryline, and it followed a more eastward path. Wave A− temporarily weakened as it moved into eastern Kansas (Fig. 7b), but rapidly strengthened upon reaching central Missouri (Fig. 7d). By contrast, wave trough B− attained its strongest amplitude in eastern Kansas, then weakened as it propagated into Missouri. Since A and B were the strongest and most coherent of the three waves, most of the emphasis is placed on them in this study.

Wave B was first detectable in the surface p′ analysis across the Texas and Oklahoma Panhandles at 1530 UTC, which is 2 h after the first indication of the associated cloudband, but 4 h prior to the development of deep convection. This gravity wave is shown during its development period in the 1700 UTC surface mesoanalysis in Fig. 9a. Wave B− is just east of the surface low pressure center, which is passing to the south of Dodge City (DDC). Meteograms from DDC are depicted in Fig. 9b. The sharp pressure dip and recovery from 1550 to 1630 UTC are caused by the mesoscale gravity wave. The southwest wind at Dodge City in Fig. 9a occurs following the passage of a sharp wind shift accompanying this singular wave of depression, which has a timescale of ∼40 min and corresponding spatial scale of ∼50 km. This is clearly of a very different scale than that of the cyclone. The cold front does not pass this station until 1820 UTC, as noted by the sharp decrease in temperature, rise in pressure, and increase in wind speed commencing at that time. It is interesting that the cold air lags the passage of the cyclone center.

Wave A appeared only intermittently in the surface p′ analyses until 1800 UTC, at which time a pressure ridge (A+) began to develop between troughs A− and B− in association with the appearance of precipitation at the ground (Fig. 7a). The moderate amplitude of this ridge was maintained until shortly after 0000 UTC 15 February, when it rapidly intensified following the outbreak of strong thunderstorms in western Missouri. Wave trough B− appears to have been related to rainband RB1 throughout its lifetime, though the gravity wave was first detected in the surface pressure data well before precipitation was reported on the ground. This wave was the dominant one across Kansas prior to the development of convection.

As wave of depression B− strengthened during its traverse across Kansas, a sharp edge developed on the back side of the rainband, in clear association with the sharply falling pressure signature (e.g., at station MHK in Fig. 10). Thus, the wave of depression acted to abruptly terminate the precipitation and clouds in a manner similar to that seen in earlier studies of singular waves of depression (Tepper 1951; Bosart and Seimon 1988). Nevertheless, the most pronounced p′ feature in this event was wave ridge A+. Strong convection led to dramatic changes in the wave structure after 2130 UTC, including making trough B− weaker and less wavelike, dramatically strengthening ridge A+, and creating a new trough B−* (note the progression of the waves from MHK to P18 in Fig. 10).

Generally speaking, wave troughs were located near the back edge as well as the front edge of the rainbands, and pressure ridges were collocated with the rainbands. The strongest wave amplitudes were in northeastern Kansas, northern Missouri, and western Illinois, and were associated with the most widespread convection. In particular, the amplitude of the pressure drop accompanying trough A− increased from 0.5 mb in eastern Kansas to nearly 4.0 mb in northeastern Missouri.

Our results tend to support the observations of Koch et al. (1988) regarding changes in gravity wave characteristics following development of strong convection (which they defined as the unmistakable presence of a gust front). In both studies, wave amplitudes quickly and unmistakably increased as strong convection erupted. The development of a strong mesohigh system, indicative of a rain-cooled hydrostatic response to the convection, was especially pronounced. The isochrones for wave B (Fig. 8b) indicate a deceleration and a rightward deviation at the time that strong convection erupted (as the waves crossed into Missouri). This behavior is also quite similar to the findings of Koch et al. (1988). Likewise, the pu*′ values for wave B (Fig. 11) were largest in those regions where the wave signature remained strongest (Fig. 7). Weaker values occurred along the Kansas–Missouri border, which is where the wave signature was relatively weak—and where severe convection erupted. However, stronger pu*′ values renewed over central Missouri as the wave and associated convection seemed to develop into a mutually supportive system. These results imply that wave-CISK processes were eventually activated here.

Surprisingly, no evidence of a gust front was found in any of the mesonetwork traces. However, since the low-level air was quite moist and stably stratified, which would diminish thunderstorm downdraft effects on the wind field, convection could still explain the temporary reduction in pu*′ correlation values and strong amplification of wave A+. Our results support the model of Johnson and Hamilton (1988) showing a strong covariance between the pressure and wind fields (Fig. 2a) once the convection–gravity wave system had matured into a stable, fully coupled dynamical system. However, during the explosive development stage of the convection, our findings are more consistent with the conceptual model of the solitary thunderstorm (Fig. 2b) by Fujita (1955) and the conceptual model of Koch et al. (1988) regarding convection–gravity wave interactions.

Convection can be ruled out as a potential genesis mechanism for the incipient waves (A, B, and C) since there was no convection in the local vicinity of the gravity wave genesis region.3 However, convection certainly appeared to help sustain the waves and also to have been involved in the development of the B−* feature that appeared behind RB2. Mesoscale model analyses by Jin (1997) indicate a significant role played by geostrophic adjustment processes in generating gravity wave B.

c. 16–18 February 1992

1) Synoptic overview

Similar to the 14 February case, a surface cyclone developed in the lee of the Rocky Mountains over southeastern Colorado on 16–17 February, and drifted eastward throughout the day. By 0500 UTC 17 February, a dryline had formed over the eastern Texas Panhandle while a cold front extended through southern Colorado (Fig. 12a). A strongly diffluent short-wave trough and a pronounced upper-level jet are also features similar to those in the previous case. A distinct difference between the two cases is that on 17 February a weak, north–south-oriented stationary front in Kansas and Oklahoma marked the leading edge of relatively warm, modified Canadian air.

The satellite and radar imagery reveals that arc-shaped cloudbands developed near the surface low as the upper-level cyclonic circulation tightened and became clearly evident. Rainband RB1 formed near the surface stationary front (Fig. 12a) and then crossed into Missouri as it progressively advanced ahead of the front. Rainband RB2 first appeared within the warm sector, then passed over the surface stationary front (Fig. 12b), and thereafter slowly advanced over the cool air ahead of this front. The forcing for these bands, which were the dominant mesoscale precipitation features from 0200 to 1500 UTC, was clearly a mid-to-upper-tropospheric phenomenon, given that the bands propagated nearly orthogonal to the surface winds and did not maintain any association with surface frontal systems. As these bands drifted off to the east, the clearing that developed between the surface low pressure system and the bands suggested a dry air intrusion, similar to the 14 February case. Rainband RB3 formed near the stationary front at ∼1700 UTC in western Missouri. The last cloudband (CB4) was first detected behind the dryline south of the low pressure center (Fig. 12a), then tracked to the northeast through the dry air intrusion in central Kansas to a final position just east of the cyclone (Fig. 12d). Here it produced precipitation in association with a strong veering of the surface winds from the southeast to the southwest.

2) Gravity wave analysis

Isochrones of these four waves identified in the surface p′ analyses are displayed in Fig. 13. A weak wave trough (B−*) was also noted trailing wave B− for a short while (Fig. 13b). Pressure traces from those stations that showed evidence of wave activity portrayed a wave train configuration. The mean horizontal wavelength, period, and phase velocity values (Table 4) are similar for each of the waves, with the exception of wave A. The waves basically followed an east-northeast track through the center of the analysis region.

All the gravity waves were shaped similarly to the cloudbands and tended to travel and even strengthen in tandem with them. A slight arc was evident in the wave fronts. The maximum pressure perturbations were closely associated with the highest cloud tops, while the wave troughs were collocated with the warmer cloud regions. Cloud organization and depth diminished significantly after the waves had weakened, though the cloudbands tended to remain.

Waves A and B first appeared in central Kansas near a region of intensifying convection, whereas waves C and D developed in central Missouri in the absence of such a linear region of preexisting convection. Rainband RB3 suddenly appeared at nearly the same time that wave C− was first detected in the p′ analyses (1700 UTC), yet no cloudband was identifiable prior to this sudden development that could be offered as a convective heat source mechanism for gravity wave generation. Rather, it appears more likely that wave C was first generated aloft by a continuous line or zone of forcing. This hypothesis is lent support in the mesoscale modeling study of this event by Jin (1997), who shows that an unbalanced CFA in the midtroposphere (imbalance being deduced from the large diagnosed Rossby number and violation of the nonlinear balance equation) was the most plausible forcing mechanism for waves B and C.

All four waves displayed areas of significant positive pu*′ correlation with numerous values in excess of 0.7 (Fig. 14). These large correlation values over fairly large areas indicate the presence of strong gravity wave signals. The correlation values for wave A− appeared to change with the strength of the wave (as in the 14 February case), with the strongest values observed in northern Missouri, which is where wave A− was strongest. Likewise, low or even negative correlation values were observed in central Missouri before wave C− became evident, but the values rapidly increased as the wave took form.

In summary, there was significant evidence supporting the suggestion that a gravity wave event occurred during the intensive observing period from 16 to 18 February. This case exhibited surprisingly similar characteristics to the 14 February case, with a cyclone developing on the lee side of the Rocky Mountains, and arc-shaped cloudbands associated with gravity waves forming in the dry air intrusion. We now examine a case that was notably different in several important respects from these two cases.

d. 8–10 March 1992

1) Synoptic overview

The situation on 8 March was similar to the previous two cases in the fact that the Oklahoma and Texas Panhandles were under the exit region of a strong jet that was propagating around an upper-level cutoff low. In addition, a lee cyclone (though stronger in the present case) was over southeastern Colorado, and a dryline extended from the cyclone through the Panhandle (Fig. 15a). Unlike the former cases, however, an arctic front was pushing southward through Nebraska. Blizzard conditions developed as the arctic air moved into northeastern Colorado. In addition, Fig. 15b shows that a CFA was advancing ahead of the dryline in western Oklahoma and southwestern Kansas (Locatelli et al. 1995, hereafter L95). This storm system intensified into a major baroclinic cyclone that eventually had major impacts upon the weather all the way to the Eastern Seaboard.

The first mesoscale rainband (RB1) developed during the afternoon of 8 March. This “pre–dry trough rainband,” (Martin et al. 1995) appearing at 0300 UTC 9 March as a group of disorganized radar echoes stretching from southeastern Nebraska to southwestern Missouri (Fig. 16a), was by this time quickly dissipating. In the meantime, strong convective cells had begun developing along the dryline from southern Kansas through central Texas. According to L95, severe convection (rainband RB2) was triggered as the cold air aloft overran the dryline and the combination of the vertical circulation transverse to the CFA and low-level isallobaric convergence released this instability. Throughout its existence, RB2 tended to move with the CFA, so it will be referred to here, as in L95, as “the CFA rainband.” By 0600 UTC (Fig. 16b), the CFA rainband had clearly pulled away from the dryline. However, a third rainband (RB3) had formed behind the CFA and right along the dryline. RB3 had merged with RB2 by 0900 UTC (Fig. 16c). During this period of time, the satellite and radar imagery revealed that a comma cloud system with an implied center of cyclonic circulation in eastern Colorado had developed. In general, RB1 and RB2 tracked more toward the east while RB3 appeared more directly “connected” to the circulation about the low pressure center and therefore moved in a northeasterly direction. The end result was a spectacular “pinwheel cloudband” pattern.

2) Gravity wave analysis

Five wavelike features were found in the resulting perturbation pressure mesoanalyses, as were a few short-lived features that were of smaller scale than the waves. The isochrones (Fig. 17) suggest that all five waves propagated toward the east or northeast as they passed through the STORM-FEST region. Table 5 shows the characteristics of waves A−, B−, C−, and E− as determined from these analyses.

The objective analyses covered the period from 0200 to 1400 UTC 9 March, though the pressure traces hint at the possible existence of wave A earlier than 0200 UTC in southeastern Kansas. Unfortunately, our analyses began at a time when RB1, RB2, and RB3 (as well as waves A− and B−) were forming or had already developed, so we could not determine whether the waves or the convection had appeared first. Since wave C− developed later (by 0500 UTC) behind RB3 and displayed an orientation similar to that of the rainband (Fig. 16b), it is possible that this wave may have formed in response to the strong convection within RB3. Wave troughs D− and E− became established ahead of convective rainband RB2, which is indicative that convection was involved in their generation as well. The waves and the convective bands in this event tended to travel together and the areas of positive perturbations were shaped similarly to the cloudbands, particularly in the northern and central portions of the region. In general, the wave signatures were neither as strong nor as coherent as in the previous two STORM-FEST cases. Nevertheless, strong pu*′ values were associated with the waves (Fig. 18), and there was some indication once again of a relationship between wave strength and the pu*′ correlation.

Martin et al. (1995) suggest that the pre–dry trough rainband (RB1) developed at the back edge of an area of warm-air advection at middle levels located within a broad region of convective instability. However, the narrowness of this rainband (∼75 km at earlier times) suggests that this mechanism does not fully explain the observations; indeed, a gravity wave (A) is indicated by our surface p′ analyses. Likewise, the complicated nature of RB2 and RB3 suggests that there was more to the severe squall line forcing than just the CFA argued by L95. The surface bandpass analysis suggested the presence of gravity wave signatures closely linked to the cloud/rainbands (and thus, by implication, the CFA features). As Locatelli et al. (1997) has shown, the surface signature of a CFA overhead is enhanced surface convergence associated with rising pressure under the leading nose of cold air aloft, which is virtually indistinguishable from a gravity wave signature. On the other hand, it is most intriguing that their analysis strongly suggests that the CFA represents unbalanced frontogenesis (strong divergence tendency exists so that the nonlinear balance equation is not satisfied). That the CFA cannot be fully understood from quasigeostrophic principles is contrary to earlier claims made by Hobbs et al. (1990). If that is true, then gravity–inertia waves might be expected to be generated near the front aloft, just as in the 17 February case. Mesoscale modeling studies in which the observed precipitation bands are well simulated, and wind profiler and Doppler radar studies of the bands, are necessary to fully unravel these complexities.

6. Conclusions

Coherent pressure pulse events (amplitude ≥ 0.2, period 1–6 h) were identified in surface pressure data over the central United States on 32% of the days during STORM-FEST. This frequency of occurrence is comparable to that found by Grivet-Talocia et al. (1999), in which coherent pressure perturbations were diagnosed 21% of the time during the fall and winter months in central Illinois.

These pulse events were ranked according to their average amplitude value. Three of the four events with the largest average amplitudes occurred in an environment similar to the UK87 (Uccellini and Koch 1987) conceptual model for mesoscale gravity wave events—a jet streak approaching an inflection axis in the upper-level height field, a diffluent trough upstream of the wave corridor, and a surface front to its south or southeast. The fourth case lacked the surface frontal signature, but did show strong static stability below 700 mb (a requirement for linear wave ducting) within the regions that experienced coherent pressure perturbation activity. This suggests that the UK87 criterion for surface fronts should be relaxed to accommodate this feature. The fact that the largest amplitude wave events all occurred only when this configuration was present suggests that the upper-level jet and wave ducting play crucial roles in the generation and maintenance of the most significant mesoscale gravity wave events. On the other hand, the mere existence of static stability does not suffice to have a mesoscale gravity wave event, since strong stratification is present more often than not in the winter months over the central United States, yet we found pressure pulse events only 34% of the time.

The three strongest pressure pulse events (14 and 17 February and 9 March) were analyzed in detail to determine if they were indeed gravity waves. Autospectral analysis, bandpass filtering to obtain pressure perturbations (p′), pressure–wind covariance (pu*′) calculations, and a time-to-space conversion (TSC) adaptation of the Barnes objective map analysis were all performed upon the surface data for these three events, and the results were related to mesoscale features seen in the radar and satellite imagery. All three of the analyzed events displayed evidence of a gravity wave, namely: (i) high pu*′ values appeared in association with the strongest p′ amplitudes, (ii) the perturbations propagated with phase velocities typical of mesoscale gravity waves, and (iii) the waves appeared to exert a controlling influence on the cloud/rainbands in a manner consistent with a ducted gravity wave model. The largest positive pressure–wind field correlation values (pu*′) appeared to be closely related to the strength of the wave signature. A variety of wave types were evident in each of the cases, including short-duration wavelets, longer-duration wave trains, and singular waves of elevation and depression, which in some cases occurred within a wavelet or train. The resultant waves in the 14–15 February event displayed phase velocities of 21.6 ± 5.9 m s−1, wavelengths of 200 ± 41 km, and an average period of 2.3 h. The wave characteristics from the 16–18 February event were 19.9 ± 6.2 m s−1 for phase velocity, 225 ± 31 km for wavelength, and 3.5 h for the mean period. The waves in the 9 March event, which were considerably weaker, displayed phase velocities of 27.9 ± 6.8 m s−1, wavelengths of 260 ± 32 km, and mean period of 2.6 h.

Tests were performed on the data from the 14–15 February case to examine the sensitivity of the TSC objective analysis of the waves to various parameters. The results suggest that caution should be exercised when choosing a bandpass-filter width, since one that rigidly encompasses the dominant wave frequencies resulting from spectral analyses may result in an attenuation of the pressure perturbations and larger phase errors. It was also empirically demonstrated that the TSC needs to be performed for a 45–50-min period centered on the analysis time to allow for the advantages of the TSC technique to be realized, just as suggested by KO97. It was much easier to follow the waves using this analysis scheme than in either subjective or customary Barnes objective analyses. The comparison between the TSC results involving a single TSC advection vector and multiple advection vectors suggested that a single vector would be adequate for analysis of a planar wave, while multiple advection vectors would be recommended for waves that are highly arc shaped.

The strong wave of depression in the 14–15 February case was evident in the surface time series data 2 h before the commencement of surface precipitation and 4 h before the development of deep convection. This relation clearly shows that convection did not force this wave (which is rare in gravity wave case studies). The fact that a cloudband was associated with the wave throughout its lifetime and was evident even before the detection of the wave in microbarograms implies that it was generated at sufficiently high levels in the troposphere, such that ground-based pressure sensors were initially unable to detect its presence. This lack of a perceptible surface signal of the gravity wave in the first couple of hours after the cloudband first became perceptible can be understood from the concept of the natural filtering properties of the intervening atmosphere. Given that within the wave generation region, no wave duct existed, the surface pressure would be diminished by the factor of sinh(nlH), since the pressure in the lower layer Pl compared to that at the source height PH can be expressed
PlPHnlznlz
where nl is the vertical wavenumber in the lower layer (Gossard and Hooke 1975, p. 150). Note that since cosh(nlz) tends to unity as z → 0, the pressure perturbation does not diminish to zero at the ground, yet the sinh(nlH) effect has a strong suppressive effect, as follows. The gravity wave was generated near the 6-km level according to investigations of this gravity wave with wind profiler data (Trexler et al. 1998) and a mesoscale numerical model (Jin 1997). In the wave generation region of the Panhandle, in which strong low-level static stability was lacking, a gravity wave forced at an altitude of H = 6 km would have produced a surface pressure signal only 10% of one that would have been detected by pressure sensors from central Kansas eastward, where a strong wave duct was present in the lowest 2 km (KO97). These specific results have major general implications for our ability to discern the presence of mesoscale gravity waves in the atmosphere, which until now have been primarily studied with surface data.

Some of the gravity waves studied in the three cases formed in western Kansas and the Panhandle region in the absence of convection. Those waves that seemed to form in response to convection generally appeared farther east. In particular, most of the waves identified in the 17 February and 9 March cases were associated with convection right from their beginnings. Nevertheless, convection cannot be considered a fully satisfactory explanation for the sudden appearance of a cloudband and associated gravity wave signature in surface pressure sensors. Linearly shaped mesoscale forcing mechanisms must be invoked to adequately explain such features.

A strong relationship between the rainbands and the waves was observed in all the three cases as the cloudbands and the gravity waves tended to travel together. The positive pressure perturbation areas (wave crests) appeared to be very closely related to the location of the rainbands, which is consistent with a ducted gravity wave model. Furthermore, the highest cloud tops generally appeared at the same time as the largest perturbation pressure values. Our results suggest that a strong covariance appears between the pressure and wind fields once the convection–gravity wave system matures into a stable, fully coupled dynamical system, consistent with the Johnson and Hamilton (1988) model. On the other hand, our results are more consistent with the conceptual model of the solitary thunderstorm (Fujita 1955; Koch et al. 1988) during the explosive development stage of the convection.

Indirect evidence suggests that topography may have played an important role in the excitation of the gravity wave events during STORM-FEST. Waves in each of the three strongest events appeared to be associated with the transformation of a lee cyclone into a baroclinic system and to occur downstream from where a jet streak was approaching an inflection axis in the upper-level height field. The majority of the cloudbands associated with the gravity waves systematically formed in the dry air intrusion, and the pinwheel-like appearance of the cloudbands that developed here strongly suggested that they were connected to the cyclonic circulation. This suggestion needs to be explored further with numerical models to understand if and how propagating gravity waves may be generated by the flow of a jet streak over complex terrain.

Additional research should focus on the interrelationships between gravity waves and complex frontal features, including cold fronts aloft. Conclusive determination of the responsible wave genesis and maintenance mechanism(s) requires detailed analysis of the special upper-level data and/or mesoscale model simulations. Finally, since some of these events did involve severe weather, it could prove beneficial to investigate the generality of the conclusions reached in this study.

Acknowledgments

We are grateful to Chris O’Handley and Devin Kramer, who provided guidance on the use of the programs used in the bandpass filter analysis and the TSC program. The Zeb data-display software developed by NCAR was used for the satellite and radar imagery. Lisa Gray did many of the complex overlay graphics in this paper. Funding for this research was obtained from the National Science Foundation under ATM-9319345 and ATM-9700626, COMET Outreach Project Grant #S96-75675, and the Air Force Institute of Technology.

REFERENCES

  • Adams, M., 1996: Terrain induced mid-tropospheric frontogenesis and jet streak development during STORM-FEST IOP 17, 8 and 9 March 1992. Ph.D. dissertation, North Carolina State University, 214 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Barnes, S. L., F. Caracena, and A. Marroquin, 1996: Extracting synoptic-scale diagnostic information from mesoscale models: The Eta Model, gravity waves, and quasigeostrophic diagnostics. Bull. Amer. Meteor. Soc.,77, 519–528.

  • Bosart, L. F., and J. P. Cussen Jr., 1973: Gravity wave phenomena accompanying East Coast cyclogenesis. Mon. Wea. Rev.,101, 446–454.

  • ——, and F. Sanders, 1986: Mesoscale structure in the megalopolitan snowstorm of 11–12 February 1983. Part III: A large-amplitude gravity wave. J. Atmos. Sci.,43, 924–939.

  • ——, and A. Seimon, 1988: A case study of an unusually intense atmospheric gravity wave. Mon. Wea. Rev.,116, 1857–1886.

  • Corbet, J., C. Mueller, C. Burghart, K. Gould, and G. Granger, 1994:Zeb: Software for integration, display, and management of diverse environmental datasets. Bull. Amer. Meteor. Soc.,75, 783–792.

  • Duchon, C. E., 1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor.,18, 1016–1022.

  • Einaudi, F., A. J. Bedard, and J. J. Finnigan, 1989: A climatology of gravity waves and other coherent disturbances at the Boulder Atmospheric Observatory during March–April 1984. J. Atmos. Sci.,46, 303–329.

  • Eom, J. K., 1975: Analysis of the internal gravity wave occurrence of 19 April 1970 in the Midwest. Mon. Wea. Rev.,103, 217–226.

  • Friday, E. W., Jr., 1993: Associated restructuring of the National Weather Service: An overview. Bull. Amer. Meteor. Soc.,75, 43–48.

  • Fujita, T. T., 1955: Results of detailed synoptic studies of squall lines. Tellus,7, 405–436.

  • Gossard, E. E. and W. B. Sweezy, 1974: Dispersion and spectra of gravity waves in the atmosphere. J. Atmos. Sci.,31, 1540–1548.

  • ——, and W. H. Hooke, 1975: Waves in the Atmosphere. Elsevier, New York, 456 pp.

  • Grivet-Talocia, S., F. Einaudi, W. L. Clark, R. D. Dennett, G. D. Nastrom, and T. E. VanZandt, 1999: A 4-yr climatology of pressure disturbances using a barometer network in central Illinois. Mon. Wea. Rev.,127, 1613–1629.

  • Hauf, T., U. Finke, J. Neisser, G. Bull, and J.-G. Stagenberg, 1996: A ground-based network for atmospheric pressure fluctuations. J. Atmos. Oceanic Technol.,13, 1001–1023.

  • Hobbs, P. V., J. D. Locatelli, and J. E. Martin, 1990: Cold fronts aloft and the forecasting of precipitation and severe weather east of the Rocky Mountains. Wea. Forecasting,5, 613–626.

  • ——, ——, and ——, 1996: A new conceptual model for cyclones generated in the lee of the Rocky Mountains. Bull. Amer. Meteor. Soc.,77, 1169–1178.

  • Jin, Y., 1997: A numerical model study of the role of mesoscale gravity waves in rainband dynamics in the central United States during STORM-FEST. Ph. D. dissertation, North Carolina State University, 318 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Johnson, R. H., and P. J. Hamilton, 1988: The relationship of surface pressure features to the precipitation and airflow structure of an intense midlatitude squall line. Mon. Wea. Rev.,116, 1444–1472.

  • Kaplan, M. L., S. E. Koch, Y.-L. Lin, and R. Weglarz, 1997: Numerical simulations of a gravity wave event over CCOPE. Part I: The role of geostrophic adjustment in mesoscale jetlet formation. Mon. Wea. Rev.,125, 1185–1211.

  • Keyser, D., B. D. Schmidt, and D. G. Duffy, 1989: A technique for representing three-dimensional vertical circulations in baroclinic disturbances. Mon. Wea. Rev.,117, 2463–2494.

  • Koch, S. E., and P. B. Dorian, 1988: A mesoscale gravity wave event observed during CCOPE. Part III: Wave environment and probable source mechanisms. Mon. Wea. Rev.,116, 2570–2592.

  • ——, and R. E. Golus, 1988: A mesoscale gravity wave event observed during CCOPE. Part I: Multi-scale statistical analysis of wave characteristics. Mon. Wea. Rev.,116, 2527–2544.

  • ——, and C. O’Handley, 1997: Operational forecasting and detection of mesoscale gravity waves. Wea. Forecasting,12, 253–281.

  • ——, M. desJardins, and P. Kocin, 1983: An interactive Barnes objective map analysis scheme for use with satellite and conventional data. J. Climate Appl. Meteor.,22, 1487–1503.

  • ——, R. E. Golus, and P. B. Dorian, 1988: A mesoscale gravity wave event observed during CCOPE. Part II: Interactions between mesoscale convective systems and the antecedent waves. Mon. Wea. Rev.,116, 2545–2569.

  • ——, D. Hamilton, D. Kramer, and A. Langmaid, 1998: Mesoscale dynamics in the Palm Sunday tornado outbreak. Mon. Wea. Rev.,126, 2031–2060.

  • Lin, Y.-L., 1994: Airflow over mesoscale heat sources. Part II: Responses in a shear flow. Proc. Natl. Sci. Council ROC (A),18, 119–150.

  • ——, and R. C. Goff, 1988: A study of a mesoscale solitary wave in the atmosphere originating near a region of deep convection. J. Atmos. Sci.,45, 194–205.

  • Lindzen, R. S., and K. K. Tung, 1976: Banded convective activity and ducted gravity waves. Mon. Wea. Rev.,104, 1602–1617.

  • 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.

  • ——, M. T. Stoelinga, R. D. Schwartz, and P. V. Hobbs, 1997: Surface convergence induced by cold fronts aloft and prefrontal surges. Mon. Wea. Rev.,125, 2808–2820.

  • Loughe, A. F., C.-C. Lai, and D. Keyser, 1995: A technique for diagnosing three-dimensional ageostrophic circulations in baroclinic disturbances on limited-area domains. Mon. Wea. Rev.,123, 1276–1303.

  • Martin, J. E., J. D. Locatelli, P. V. Hobbs, P.-Y. Wang, and J. A. Castle, 1995: Structure and evolution of winter cyclones in the central United States and their effects on the distribution of precipitation. Part I: A synoptic-scale rainband associated with a dryline and lee trough. Mon. Wea. Rev.,123, 241–264.

  • Pecnick, M. J., and, J. A. Young, 1984: Mechanics of a strong subsynoptic gravity wave deduced from satellite and surface observations. J. Atmos. Sci.,41, 1850–1862.

  • Pokrandt, P. J., G. J. Tripoli, and D. D. Houghton, 1997: Processes leading to the formation of mesoscale waves in the midwest cyclone of 15 December 1987. Mon. Wea. Rev.,124, 2726–2752.

  • Powers, J. G., 1997: Numerical model simulation of a mesoscale gravity-wave event: Sensitivity tests and spectral analyses. Mon. Wea. Rev.,125, 1838–1869.

  • ——, and R. J. Reed, 1993: Numerical simulation of the large-amplitude mesoscale gravity-wave event of 15 December 1987 in the central United States. Mon. Wea. Rev.,121, 2285–2306.

  • Ralph, F. M., M. Crochet, and S. V. Venkateswaran, 1993: Observations of a mesoscale ducted gravity wave. J. Atmos. Sci.,50, 3277–3291.

  • Ramamurthy, M. K., R. M. Rauber, B. P. Collins, and N. K. Malhotra, 1993: A comparative study of large-amplitude gravity wave events. Mon. Wea. Rev.,121, 2951–2974.

  • Raymond, D. J., 1983: Wave-CISK in mass form. J. Atmos. Sci.,40, 2561–2571.

  • Rottman, J. W., and F. Einaudi, 1993: Solitary waves in the atmosphere. J. Atmos. Sci.,50, 2116–2136.

  • Schneider, R. S., 1990: Large-amplitude mesoscale wave disturbances within the intense Midwest extratropical cyclone of 15 December 1987. Wea. Forecasting,5, 533–557.

  • Siedlarz, L. M., 1996: A climatology of mesoscale wave disturbances seen in mesonet data during STORM-FEST. M.S. thesis, North Carolina State University, 200 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Stobie, J. G., F. Einaudi, and L. W. Uccellini, 1983: A case study of gravity waves–convective interactions: 9 May 1979. J. Atmos. Sci.,40, 2804–2830.

  • Szoke, E., J., J. M. Brown, J. A. McGinley, and D. Rodgers, 1994: Forecasting for a large field program: STORM-FEST. Wea. Forecasting,9, 593–605.

  • Tepper, M., 1951: On the desiccation of a cloud bank by a propagating pressure wave. Mon. Wea. Rev.,79, 61–70.

  • Trexler, C. M., Y. Jin, and S. E. Koch, 1998: Vertical structure of a mesoscale gravity wave event detected during STORM-FEST. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, AZ, Amer. Meteor. Soc., 463–465.

  • Uccellini, L. W., 1975: A case study of apparent gravity wave initiation of severe convective storms. Mon. Wea. Rev.,103, 497–513.

  • ——, and S. E. Koch, 1987: The synoptic setting and possible energy sources for mesoscale wave disturbances. Mon. Wea. Rev.,115, 721–729.

Fig. 1.
Fig. 1.

Height variation in the phase relations between the horizontal wind component in the direction of wave propagation (u*) and the wave-induced vertical motions (w′) for: (a) a gravity wave displaying no vertical tilt and (b) an upstream-tilted wave, both of which are propagating toward the right. Wave phase velocity (C) indicates that the upstream-tilted wave is vertically propagating, in contrast to the nontilted wave. Also shown is location of maximum cloudiness relative to gravity wave, assumed in both cases to be π/8 behind the maximum vertically integrated updraft. Here H and L refer to high and low pressure surface perturbations (after Koch and Golus 1988).

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 2.
Fig. 2.

Surface pressure and wind fields associated with deep convection: (a) the mature stage of a squall line with a trailing stratiform precipitation region according to the conceptual model of Johnson and Hamilton (1988), and (b) the thunderstorm conceptual model of Fujita (1955), in which small arrows indicate surface wind and large arrows the relative flow into the wake depression. Contours are isobars.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 3.
Fig. 3.

Locations of the STORM-FEST surface mesonet stations used in this study: ASOS, ICN, and PAM II stations (open triangles), and barograph sites (filled triangles). Note the PAM II boundary layer array in northeastern Kansas. Stations whose meteograms are displayed in Fig. 10 are highlighted. An additional 48 AWOS sites (not shown) were also used in the analysis of the 14 Feb case.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 4.
Fig. 4.

(a) Bar graph histogram of significant peaks identified from the autospectral analysis for the 14 Feb 1992 gravity wave event, and the wide-filter (f1 = 0.004 min−1, f2 = 0.034 min−1) and narrow-filter (f1 = 0.004 min−1, f2 = 0.010 min−1) response curves; (b) and (c) the narrow-filter and wide-filter reponses plotted over one-half of the Nyquist frequency interval (0.10 min−1).

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 5.
Fig. 5.

TSC objective analysis plots of pressure perturbation fields from the 14 Feb gravity wave event showing the effects of advection vectors in the TSC process: (a) multiple advection vectors and (b) single advection vector (220°, 15.7 m s−1). Contour interval is 0.2 mb, with darker shading every 0.4 mb for negative values. Thick lines depict locations of wave troughs A−, B−, and C−.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 6.
Fig. 6.

Surface mesonet temperatures (C) and wind vectors, and radar reflectivity mosaic overlay (only below the east–west line in northern Kansas), on 14 Feb 1992 at (a) 1530 UTC, (b) 1810 UTC, and (c) 2130 UTC, and on 15 Feb at (d) 0030 UTC, and meteorological features, including the surface low pressure center, cold and warm fronts, and dryline. Enhanced infrared satellite imagery is overlaid on panel (c) radar image. Arc lines indicate gravity wave troughs in surface analyses. RB1 and RB2 are rainbands discussed in text.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 7.
Fig. 7.

Objectively analyzed pressure perturbation fields on 14 Feb 1992 at (a) 1830 UTC, (b) 2000 UTC, (c) 2130 UTC, and (d) 2300 UTC, and on 15 Feb 1992 at (e) 0030 UTC and (f) 0200 UTC following the same format as in Fig. 5.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 8.
Fig. 8.

Isochrones for gravity wave troughs (a) A−, (b) B−, and (c) C− obtained from the objectively analyzed p′ fields for the 14 Feb gravity wave event. Solid lines represent the primary wave isochrones. Dotted lines in (b) represent coherent wave signatures (B−*) that developed behind the primary wave B−. Other dotted lines signify incoherent early waves.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 9.
Fig. 9.

(a) Surface mesonet station plots, enhanced infrared satellite imagery, and surface analysis at 1700 UTC 14 Feb 1992. The station plots consist of temperatures and dewpoints (F) and wind barbs (long barb = 5 m s−1, short barb = 2.5 m s−1). Analysis shows the surface low pressure center, cold and warm fronts, and dryline, and isobars (2-hPa intervals). Gravity wave trough B− is the small closed isobar due east of the low pressure center. Dodge City is located directly north of the low pressure center. (b) Time series of 5-min surface measurements from Dodge City for the interval 1000–2000 UTC 14 Feb 1992, consisting of surface pressure (thick solid line), temperature (dotted line), dewpoint (gray line), wind speed (dashed line), and wind direction (thin solid line).

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 10.
Fig. 10.

Selected 5-min resolution pressure traces (mb) from the 14 Feb 1992 gravity wave event (1530 UTC 14 Feb–0330 UTC 15 Feb) for stations DDC, MHK, P16, P17, and P18. The pressure trace for DDC shows a singular wave of depression. A wavelet is indicated at MHK (A− to A+ to B−) and a wave train at P18.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 11.
Fig. 11.

Average pu*′ correlation values (*100) determined over the 6-h period of wave activity at each station for the primary gravity wave (B) on 14 Feb 1992.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 12.
Fig. 12.

Surface mesonet temperatures (°C) and wind vectors, and radar reflectivity mosaic overlays (only below the east–west line in central Nebraska) for the 17 Feb 1992 case at (a) 0502 UTC, (b) 1002 UTC, (c) 1502 UTC, and (d) 2002 UTC, and meteorological features, including the surface low pressure center, fronts, and dryline. Radar reflectivity overlaid on enhanced infrared satellite imagery in panel (a) and (d) is shown with contours instead of shading. Arc lines indicate gravity wave troughs in surface analyses. RB1 and RB2 are rainbands discussed in text. Cloudband CB4 discussed in text is labeled on the infrared satellite imagery in panel (a), but does not produce a rainband (near the cyclone center) until just before 1502 UTC.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 13.
Fig. 13.

As in Fig. 8 except for the 17 Feb 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, and (d) D−. Short-lived wave trough B−* is indicated by the dotted lines in (b).

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 14.
Fig. 14.

Average pu*′ correlation values (*100) determined over a 3-h period of wave activity at each station for the 17 Feb 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, and (d) D−.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 15.
Fig. 15.

Subjective analyses at 0000 UTC 9 Mar 1992 derived from conventional surface and STORM-FEST special sounding observations:(a) surface isobars (solid, 4-mb intervals), isentropes (dashed, 4 K intervals), radar echoes (shaded, 10-dBZ intervals), and station model plots showing potential temperature, equivalent potential temperature, mean sea level pressure, observed weather, and wind barbs (full staff—5 m s−1, half staff—2.5 m s−1); (b) 500-mb geopotential height contours (solid, 6-dcm intervals), isentropes (dashed, 2 K intervals), isotachs (dotted, 5 m s−1 intervals, shaded above 30 m s−1), and station model plots showing potential temperature, geopotential height, and wind barbs. Frontolysis is depicted by broken lines separated by crosses, CFA by line of open triangles, and dryline by scalloped line (after Adams 1996).

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 16.
Fig. 16.

As in Fig. 6 except for 9 Mar 1992 case at (a) 0300 UTC, (b) 0600 UTC, and (c) 0900 UTC. Enhanced infrared satellite imagery is overlaid on panel (b) radar image.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 17.
Fig. 17.

As in Fig. 8 except for the 9 Mar 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, (d) D−, and (e) E−.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Fig. 18.
Fig. 18.

Average pu*′ correlation values (*100) determined over a 3-h period of wave activity at each station for the 9 Mar 1992 event for gravity waves (a) A−, (b) B−, (c) C−, and (d) E−.

Citation: Monthly Weather Review 127, 12; 10.1175/1520-0493(1999)127<2854:MGWATE>2.0.CO;2

Table 1.

Amplitude ranking of the 13 identified pressure pulse events in STORM-FEST, and their occurrence with respect to the Uccellini and Koch (1987) conceptual model of the wave environment. The first hour the event was observed through the last hour of activity is included in the duration (day/UTC). The results of the UK87 verification are included for all three criteria—a jet streak near the inflection axis, a diffluent trough and a front to the south or southeast of the wave region, as well as for when a jet streak near the inflection axis was considered alone.

Table 1.
Table 2.

Contingency table for prediction of coherent mesoscale pressure disturbance event for cases in which a 500-mb jet streak occurred near a height field inflection axis. “Predicted gravity waves” exhibited this feature, whereas “no predictions” did not. The total number of days in the sample was 41.

Table 2.
Table 3.

Phase speeds and directions, horizontal wavelengths, and wave periods for the wave troughs identified in the 14–15 Feb gravity wave event. The mean periods were calculated from the phase speeds and wavelengths.

Table 3.
Table 4.

Phase speeds and directions, horizontal wavelengths, and average wave periods for the wave troughs identified in the 16–18 Feb gravity wave event. The mean periods were calculated from the phase speeds and wavelengths.

Table 4.
Table 5.

Phase speeds and directions, horizontal wavelengths, and wave periods for the wave troughs identified in the 9 Mar gravity wave event. The mean periods were calculated from the phase speeds and wavelengths.

Table 5.

1

This statement is precisely true only for straight jets. Although curvature effects have been shown by Keyser et al. (1989) and Loughe et al. (1995) to produce nondivergent cross-stream flow, these studies do not reveal “unbalanced” cross-stream flow (i.e., flow directed toward the cyclonic side of the upper-level jet in the exit region of the geostrophic wind maximum).

2

The shorthand notation used herein for wave troughs and ridges is as follows, for example, for wave A: A− refers to the trough and A+ refers to the wave crest.

3

Precipitation produced by wave B did not reach the ground until 1900 UTC. It is highly unlikely that deep convection could have provided a significant source of latent heat energy for the waves until at least that time. Weak wave coherence in the TSC analyses prior to 1600 UTC was primarily the result of the paucity of observations in the wave source region (Fig. 3), rather than the lack of a gravity wave itself, since the wave was quite apparent in the microbarogram data (Fig. 9).

Save
  • Adams, M., 1996: Terrain induced mid-tropospheric frontogenesis and jet streak development during STORM-FEST IOP 17, 8 and 9 March 1992. Ph.D. dissertation, North Carolina State University, 214 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Barnes, S. L., F. Caracena, and A. Marroquin, 1996: Extracting synoptic-scale diagnostic information from mesoscale models: The Eta Model, gravity waves, and quasigeostrophic diagnostics. Bull. Amer. Meteor. Soc.,77, 519–528.

  • Bosart, L. F., and J. P. Cussen Jr., 1973: Gravity wave phenomena accompanying East Coast cyclogenesis. Mon. Wea. Rev.,101, 446–454.

  • ——, and F. Sanders, 1986: Mesoscale structure in the megalopolitan snowstorm of 11–12 February 1983. Part III: A large-amplitude gravity wave. J. Atmos. Sci.,43, 924–939.

  • ——, and A. Seimon, 1988: A case study of an unusually intense atmospheric gravity wave. Mon. Wea. Rev.,116, 1857–1886.

  • Corbet, J., C. Mueller, C. Burghart, K. Gould, and G. Granger, 1994:Zeb: Software for integration, display, and management of diverse environmental datasets. Bull. Amer. Meteor. Soc.,75, 783–792.

  • Duchon, C. E., 1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor.,18, 1016–1022.

  • Einaudi, F., A. J. Bedard, and J. J. Finnigan, 1989: A climatology of gravity waves and other coherent disturbances at the Boulder Atmospheric Observatory during March–April 1984. J. Atmos. Sci.,46, 303–329.

  • Eom, J. K., 1975: Analysis of the internal gravity wave occurrence of 19 April 1970 in the Midwest. Mon. Wea. Rev.,103, 217–226.

  • Friday, E. W., Jr., 1993: Associated restructuring of the National Weather Service: An overview. Bull. Amer. Meteor. Soc.,75, 43–48.

  • Fujita, T. T., 1955: Results of detailed synoptic studies of squall lines. Tellus,7, 405–436.

  • Gossard, E. E. and W. B. Sweezy, 1974: Dispersion and spectra of gravity waves in the atmosphere. J. Atmos. Sci.,31, 1540–1548.

  • ——, and W. H. Hooke, 1975: Waves in the Atmosphere. Elsevier, New York, 456 pp.

  • Grivet-Talocia, S., F. Einaudi, W. L. Clark, R. D. Dennett, G. D. Nastrom, and T. E. VanZandt, 1999: A 4-yr climatology of pressure disturbances using a barometer network in central Illinois. Mon. Wea. Rev.,127, 1613–1629.

  • Hauf, T., U. Finke, J. Neisser, G. Bull, and J.-G. Stagenberg, 1996: A ground-based network for atmospheric pressure fluctuations. J. Atmos. Oceanic Technol.,13, 1001–1023.

  • Hobbs, P. V., J. D. Locatelli, and J. E. Martin, 1990: Cold fronts aloft and the forecasting of precipitation and severe weather east of the Rocky Mountains. Wea. Forecasting,5, 613–626.

  • ——, ——, and ——, 1996: A new conceptual model for cyclones generated in the lee of the Rocky Mountains. Bull. Amer. Meteor. Soc.,77, 1169–1178.

  • Jin, Y., 1997: A numerical model study of the role of mesoscale gravity waves in rainband dynamics in the central United States during STORM-FEST. Ph. D. dissertation, North Carolina State University, 318 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Johnson, R. H., and P. J. Hamilton, 1988: The relationship of surface pressure features to the precipitation and airflow structure of an intense midlatitude squall line. Mon. Wea. Rev.,116, 1444–1472.

  • Kaplan, M. L., S. E. Koch, Y.-L. Lin, and R. Weglarz, 1997: Numerical simulations of a gravity wave event over CCOPE. Part I: The role of geostrophic adjustment in mesoscale jetlet formation. Mon. Wea. Rev.,125, 1185–1211.

  • Keyser, D., B. D. Schmidt, and D. G. Duffy, 1989: A technique for representing three-dimensional vertical circulations in baroclinic disturbances. Mon. Wea. Rev.,117, 2463–2494.

  • Koch, S. E., and P. B. Dorian, 1988: A mesoscale gravity wave event observed during CCOPE. Part III: Wave environment and probable source mechanisms. Mon. Wea. Rev.,116, 2570–2592.

  • ——, and R. E. Golus, 1988: A mesoscale gravity wave event observed during CCOPE. Part I: Multi-scale statistical analysis of wave characteristics. Mon. Wea. Rev.,116, 2527–2544.

  • ——, and C. O’Handley, 1997: Operational forecasting and detection of mesoscale gravity waves. Wea. Forecasting,12, 253–281.

  • ——, M. desJardins, and P. Kocin, 1983: An interactive Barnes objective map analysis scheme for use with satellite and conventional data. J. Climate Appl. Meteor.,22, 1487–1503.

  • ——, R. E. Golus, and P. B. Dorian, 1988: A mesoscale gravity wave event observed during CCOPE. Part II: Interactions between mesoscale convective systems and the antecedent waves. Mon. Wea. Rev.,116, 2545–2569.

  • ——, D. Hamilton, D. Kramer, and A. Langmaid, 1998: Mesoscale dynamics in the Palm Sunday tornado outbreak. Mon. Wea. Rev.,126, 2031–2060.

  • Lin, Y.-L., 1994: Airflow over mesoscale heat sources. Part II: Responses in a shear flow. Proc. Natl. Sci. Council ROC (A),18, 119–150.

  • ——, and R. C. Goff, 1988: A study of a mesoscale solitary wave in the atmosphere originating near a region of deep convection. J. Atmos. Sci.,45, 194–205.

  • Lindzen, R. S., and K. K. Tung, 1976: Banded convective activity and ducted gravity waves. Mon. Wea. Rev.,104, 1602–1617.

  • 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.

  • ——, M. T. Stoelinga, R. D. Schwartz, and P. V. Hobbs, 1997: Surface convergence induced by cold fronts aloft and prefrontal surges. Mon. Wea. Rev.,125, 2808–2820.

  • Loughe, A. F., C.-C. Lai, and D. Keyser, 1995: A technique for diagnosing three-dimensional ageostrophic circulations in baroclinic disturbances on limited-area domains. Mon. Wea. Rev.,123, 1276–1303.

  • Martin, J. E., J. D. Locatelli, P. V. Hobbs, P.-Y. Wang, and J. A. Castle, 1995: Structure and evolution of winter cyclones in the central United States and their effects on the distribution of precipitation. Part I: A synoptic-scale rainband associated with a dryline and lee trough. Mon. Wea. Rev.,123, 241–264.

  • Pecnick, M. J., and, J. A. Young, 1984: Mechanics of a strong subsynoptic gravity wave deduced from satellite and surface observations. J. Atmos. Sci.,41, 1850–1862.

  • Pokrandt, P. J., G. J. Tripoli, and D. D. Houghton, 1997: Processes leading to the formation of mesoscale waves in the midwest cyclone of 15 December 1987. Mon. Wea. Rev.,124, 2726–2752.

  • Powers, J. G., 1997: Numerical model simulation of a mesoscale gravity-wave event: Sensitivity tests and spectral analyses. Mon. Wea. Rev.,125, 1838–1869.

  • ——, and R. J. Reed, 1993: Numerical simulation of the large-amplitude mesoscale gravity-wave event of 15 December 1987 in the central United States. Mon. Wea. Rev.,121, 2285–2306.

  • Ralph, F. M., M. Crochet, and S. V. Venkateswaran, 1993: Observations of a mesoscale ducted gravity wave. J. Atmos. Sci.,50, 3277–3291.

  • Ramamurthy, M. K., R. M. Rauber, B. P. Collins, and N. K. Malhotra, 1993: A comparative study of large-amplitude gravity wave events. Mon. Wea. Rev.,121, 2951–2974.

  • Raymond, D. J., 1983: Wave-CISK in mass form. J. Atmos. Sci.,40, 2561–2571.

  • Rottman, J. W., and F. Einaudi, 1993: Solitary waves in the atmosphere. J. Atmos. Sci.,50, 2116–2136.

  • Schneider, R. S., 1990: Large-amplitude mesoscale wave disturbances within the intense Midwest extratropical cyclone of 15 December 1987. Wea. Forecasting,5, 533–557.

  • Siedlarz, L. M., 1996: A climatology of mesoscale wave disturbances seen in mesonet data during STORM-FEST. M.S. thesis, North Carolina State University, 200 pp. [Available from North Carolina State University, Dept. of Marine, Earth, and Atmospheric Sciences, Campus Box 8208, Raleigh, NC 27695-8208.].

  • Stobie, J. G., F. Einaudi, and L. W. Uccellini, 1983: A case study of gravity waves–convective interactions: 9 May 1979. J. Atmos. Sci.,40, 2804–2830.

  • Szoke, E., J., J. M. Brown, J. A. McGinley, and D. Rodgers, 1994: Forecasting for a large field program: STORM-FEST. Wea. Forecasting,9, 593–605.

  • Tepper, M., 1951: On the desiccation of a cloud bank by a propagating pressure wave. Mon. Wea. Rev.,79, 61–70.

  • Trexler, C. M., Y. Jin, and S. E. Koch, 1998: Vertical structure of a mesoscale gravity wave event detected during STORM-FEST. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, AZ, Amer. Meteor. Soc., 463–465.

  • Uccellini, L. W., 1975: A case study of apparent gravity wave initiation of severe convective storms. Mon. Wea. Rev.,103, 497–513.

  • ——, and S. E. Koch, 1987: The synoptic setting and possible energy sources for mesoscale wave disturbances. Mon. Wea. Rev.,115, 721–729.

  • Fig. 1.

    Height variation in the phase relations between the horizontal wind component in the direction of wave propagation (u*) and the wave-induced vertical motions (w′) for: (a) a gravity wave displaying no vertical tilt and (b) an upstream-tilted wave, both of which are propagating toward the right. Wave phase velocity (C) indicates that the upstream-tilted wave is vertically propagating, in contrast to the nontilted wave. Also shown is location of maximum cloudiness relative to gravity wave, assumed in both cases to be π/8 behind the maximum vertically integrated updraft. Here H and L refer to high and low pressure surface perturbations (after Koch and Golus 1988).

  • Fig. 2.

    Surface pressure and wind fields associated with deep convection: (a) the mature stage of a squall line with a trailing stratiform precipitation region according to the conceptual model of Johnson and Hamilton (1988), and (b) the thunderstorm conceptual model of Fujita (1955), in which small arrows indicate surface wind and large arrows the relative flow into the wake depression. Contours are isobars.

  • Fig. 3.

    Locations of the STORM-FEST surface mesonet stations used in this study: ASOS, ICN, and PAM II stations (open triangles), and barograph sites (filled triangles). Note the PAM II boundary layer array in northeastern Kansas. Stations whose meteograms are displayed in Fig. 10 are highlighted. An additional 48 AWOS sites (not shown) were also used in the analysis of the 14 Feb case.

  • Fig. 4.

    (a) Bar graph histogram of significant peaks identified from the autospectral analysis for the 14 Feb 1992 gravity wave event, and the wide-filter (f1 = 0.004 min−1, f2 = 0.034 min−1) and narrow-filter (f1 = 0.004 min−1, f2 = 0.010 min−1) response curves; (b) and (c) the narrow-filter and wide-filter reponses plotted over one-half of the Nyquist frequency interval (0.10 min−1).

  • Fig. 5.

    TSC objective analysis plots of pressure perturbation fields from the 14 Feb gravity wave event showing the effects of advection vectors in the TSC process: (a) multiple advection vectors and (b) single advection vector (220°, 15.7 m s−1). Contour interval is 0.2 mb, with darker shading every 0.4 mb for negative values. Thick lines depict locations of wave troughs A−, B−, and C−.

  • Fig. 6.

    Surface mesonet temperatures (C) and wind vectors, and radar reflectivity mosaic overlay (only below the east–west line in northern Kansas), on 14 Feb 1992 at (a) 1530 UTC, (b) 1810 UTC, and (c) 2130 UTC, and on 15 Feb at (d) 0030 UTC, and meteorological features, including the surface low pressure center, cold and warm fronts, and dryline. Enhanced infrared satellite imagery is overlaid on panel (c) radar image. Arc lines indicate gravity wave troughs in surface analyses. RB1 and RB2 are rainbands discussed in text.

  • Fig. 7.

    Objectively analyzed pressure perturbation fields on 14 Feb 1992 at (a) 1830 UTC, (b) 2000 UTC, (c) 2130 UTC, and (d) 2300 UTC, and on 15 Feb 1992 at (e) 0030 UTC and (f) 0200 UTC following the same format as in Fig. 5.

  • Fig. 8.

    Isochrones for gravity wave troughs (a) A−, (b) B−, and (c) C− obtained from the objectively analyzed p′ fields for the 14 Feb gravity wave event. Solid lines represent the primary wave isochrones. Dotted lines in (b) represent coherent wave signatures (B−*) that developed behind the primary wave B−. Other dotted lines signify incoherent early waves.

  • Fig. 9.

    (a) Surface mesonet station plots, enhanced infrared satellite imagery, and surface analysis at 1700 UTC 14 Feb 1992. The station plots consist of temperatures and dewpoints (F) and wind barbs (long barb = 5 m s−1, short barb = 2.5 m s−1). Analysis shows the surface low pressure center, cold and warm fronts, and dryline, and isobars (2-hPa intervals). Gravity wave trough B− is the small closed isobar due east of the low pressure center. Dodge City is located directly north of the low pressure center. (b) Time series of 5-min surface measurements from Dodge City for the interval 1000–2000 UTC 14 Feb 1992, consisting of surface pressure (thick solid line), temperature (dotted line), dewpoint (gray line), wind speed (dashed line), and wind direction (thin solid line).

  • Fig. 10.

    Selected 5-min resolution pressure traces (mb) from the 14 Feb 1992 gravity wave event (1530 UTC 14 Feb–0330 UTC 15 Feb) for stations DDC, MHK, P16, P17, and P18. The pressure trace for DDC shows a singular wave of depression. A wavelet is indicated at MHK (A− to A+ to B−) and a wave train at P18.

  • Fig. 11.

    Average pu*′ correlation values (*100) determined over the 6-h period of wave activity at each station for the primary gravity wave (B) on 14 Feb 1992.

  • Fig. 12.

    Surface mesonet temperatures (°C) and wind vectors, and radar reflectivity mosaic overlays (only below the east–west line in central Nebraska) for the 17 Feb 1992 case at (a) 0502 UTC, (b) 1002 UTC, (c) 1502 UTC, and (d) 2002 UTC, and meteorological features, including the surface low pressure center, fronts, and dryline. Radar reflectivity overlaid on enhanced infrared satellite imagery in panel (a) and (d) is shown with contours instead of shading. Arc lines indicate gravity wave troughs in surface analyses. RB1 and RB2 are rainbands discussed in text. Cloudband CB4 discussed in text is labeled on the infrared satellite imagery in panel (a), but does not produce a rainband (near the cyclone center) until just before 1502 UTC.

  • Fig. 13.

    As in Fig. 8 except for the 17 Feb 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, and (d) D−. Short-lived wave trough B−* is indicated by the dotted lines in (b).

  • Fig. 14.

    Average pu*′ correlation values (*100) determined over a 3-h period of wave activity at each station for the 17 Feb 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, and (d) D−.

  • Fig. 15.

    Subjective analyses at 0000 UTC 9 Mar 1992 derived from conventional surface and STORM-FEST special sounding observations:(a) surface isobars (solid, 4-mb intervals), isentropes (dashed, 4 K intervals), radar echoes (shaded, 10-dBZ intervals), and station model plots showing potential temperature, equivalent potential temperature, mean sea level pressure, observed weather, and wind barbs (full staff—5 m s−1, half staff—2.5 m s−1); (b) 500-mb geopotential height contours (solid, 6-dcm intervals), isentropes (dashed, 2 K intervals), isotachs (dotted, 5 m s−1 intervals, shaded above 30 m s−1), and station model plots showing potential temperature, geopotential height, and wind barbs. Frontolysis is depicted by broken lines separated by crosses, CFA by line of open triangles, and dryline by scalloped line (after Adams 1996).

  • Fig. 16.

    As in Fig. 6 except for 9 Mar 1992 case at (a) 0300 UTC, (b) 0600 UTC, and (c) 0900 UTC. Enhanced infrared satellite imagery is overlaid on panel (b) radar image.

  • Fig. 17.

    As in Fig. 8 except for the 9 Mar 1992 event for gravity wave troughs (a) A−, (b) B−, (c) C−, (d) D−, and (e) E−.

  • Fig. 18.

    Average pu*′ correlation values (*100) determined over a 3-h period of wave activity at each station for the 9 Mar 1992 event for gravity waves (a) A−, (b) B−, (c) C−, and (d) E−.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 501 80 9
PDF Downloads 253 54 6