Cynthia K. Mueller, James W. Wilson, and N. Andrew Crook
National Center for Atmospheric Research, Boulder Colorado
Weather and Forecasting
Vol. 8,No. 1, March 1993, 132-146
ABSTRACT
Previous studies have shown that thunderstorms often form along boundary-layer convergence lines (boundaries) detected by sensitive Doppler radar similar to the WSR-88D. In this paper, high-resolution mesonet observations (10-15-km spacing and 1-min averages) and sounding data (eight stations within 25 000 km2 and 1-6-h frequency) collected in northeast Colorado are examined to determine their utility for forecasting precisely when and where storms initiate along boundaries. Stability indices derived from mesonet and sounding data were useful in identifying stable regions where storm initiation was unlikely. However, in regions where indices indicated a degree of latent instability, storms often did not form and if they did their intensities were not correlated to the magnitude of the instability.
Two-dimensional numerical model studies show that in a near-neutral environment (as typical during a Denver, Colorado summer afternoon), surface temperature and/or dewpoint fluctuations of 2-4 deg C can be significant for storm initiation. Small scale fluctuations of this magnitude are common. In addition, observations and numerical model results suggest that in the High Plains, profiles of boundary-layer moisture are necessary to identify the precise locations of storm initiation along convergence lines. Mesonets cannot provide this information and it is impractical to obtain it solely from soundings. Cumulus clouds identify regions where moisture is mixed to the cloud condensation level. Therefore, monitoring cloud location and development with visual observations, very sensitive radar, and satellite imagery is a useful indirect means for identifying regions of deep moisture.
It is argued that the rule for forecasting short-term, time-specific locations of thunderstorm initiation, presented in a previous paper by the authors are not significantly changed by the addition of high-resolution mesonet and sounding data. Observations and numerical model results reinforce the importance of using the locations of clouds, stationary boundaries, and horizontal rolls as potential indicators of deep moisture and potential locations of thunderstorm initiation. Mesonet and sounding data are primarily useful for identifying the potential within a mesoscale air mass for thunderstorm initiation. Therefore, mesonet spacing of 25 to 50 km and access to a morning sounding are felt to be adequate.
This study was initiated to determine if high-resolution spatial and temporal thermodynamic data from mesonets and soundings would improve the ability to forecast specific locations/times of thunderstorm initiation. To the authors' surprise, data at the resolution available from CINDE were of marginal utility for this purpose.
Soundings collected in proximity to radar-detected boundaries were marginally useful in determining the likelihood of storm initiation. MLI indices, negative area below the level of free convection, and vertical shear of the horizontal wind were examined. The proximity soundings were useful in distinguishing cases that would not produce storms (MLI's > -1). However, in most cases the morning sounding and a nearby mesonet station would give the same information. A combination of negative MLI and negative area below the LFC showed a weak correlation with thunderstorm initiation. Wind profiles relative to boundaries were examined; however, there was insufficient evidence to indicate the role of shear in thunderstorm initiation. In addition as proximity soundings were logistically difficult, an accurate 2-h forecast of boundary location/evolution was required to place soundings. Small-scale fluctuations made sampling of the updraft region difficult.
Spatial differences in the MLI of soundings take within 30 min of each other have a standard deviation of 1.4 deg C. The standard deviation of MLI values in time are similar: 1.9 deg C for the 0500 versus 1400 and 1.6 deg C for 1100 versus 1400 soundings. Some of the causes for the variation are meteorological; that is differing mesoscale air masses, stationary boundaries, and horizontal rolls. Another cause may be instrument error; intercomparison showed a standard deviation of 0.5 deg C in MLI.
Numerical model results show that the High Plains atmosphere is very sensitive to small fluctuations in the temperature and moisture. Sensitivity studies were done using observed variations. Simulations with a sounding whose MLI value was -3.2 deg C produced a 40 dBZ cell. When the MLI was increased to -1.6 deg C by reducing the boundary-layer moisture by 1 g/kg, no precipitable water formed. Also, no convection formed when the depth of the moist layer was decreased to ~ 1 km AGL.
Although sounding and mesonet data were of marginal utility for specific forecasts of thunderstorm initiation, they were useful in defining thunderstorm potential. A dry stable sounding and corresponding afternoon mesonet data were good indicators of no thunderstorm development. The data were valuable for identifying regions where storms were not likely to form. Usually these areas were rather large in size. The authors fell that thunderstorm potential would have been sampled as well with mesonet station spacing of 25-50 km and a late morning sounding.
These results are valid for the High Plains region where the summertime static stability is roughly neutral. They are not valid in regions such as the Midwest, where potential energy is capped by inversions that cannot be eliminated by solar heating alone. In these cases small changes to the boundary-layer moisture and temperature profiles are not major factors to the initiation of convection. Similarly in moist regions, such as Florida, the depth of the low-level moisture is probably not crucial because there is generally plenty of moisture throughout the boundary layer. In these regions negative area and vertical shear of the horizontal wind may play a more important role.
The authors believe that time-/space-specific thunderstorm initiation forecasts require detailed knowledge of the boundary-layer moisture profile. Wilson and Mueller (1993) note that areas of preferred storm initiation were a) intersection points between horizontal rolls and stationary boundaries; b) collisions between a moving and stationary boundary; and c) regions where a moving boundary passes under a field of existing cumulus clouds. These locations are believed to be localized regions of deep moisture due to vertical transport by sustained updraft. Numerical model results show that varying the depth of moisture has a profound influence on whether clear skies, cumulus clouds, or thunderstorms form. Mesonet stations cannot measure moisture depth and it is difficult/impractical to use soundings. Unfortunately, there is no practical way to routinely monitor moisture depth. The presence of cumulus clouds indicates regions where the moisture is mixed to the condensation level and at least for the High Plains provides important information for thunderstorm nowcasts. Until means are available for mapping boundary-layer moisture in detail, specific time/space thunderstorm initiation forecasting is best accomplished by observing boundary-layer features and echoes aloft using sensitive Doppler radar and mapping cloud locations. Possible methods for monitoring cloud activity may be high-resolution satellite data, visual observations, or sensitive 10-cm radars.