Cross-Sectional Analysis of GFS Grid-scale Convection

Vertical cross-sections (1000-100 hPa) can be seen here along 42°N between 105° and 85°W preceeding and during the IA grid-scale precipitation bomb event. The top panel shows the horizontal winds, vertical motion and absolute vorticity forecasts from 3 hours to 99 hours into the 12 UTC 20 August 2002 GFS forecast, the middle panel shows the vertical temperature and moisture structure, and the bottom panel shows the convective (red) and total (green) precipitation, with the grid-scale precipitation represented by the difference.

These cross-sections clearly show the areas of upward motion are coincident with moist areas that are conditionally unstable (grey shading, where the environmental lapse rate is greater than moist-adiabatic). Before the grid-scale convection begins (between 09 UTC and 15 UTC 22 August), we can see

• In the moisture field
  • A layer of moist (RH > 90%) air between 500 and 400-hPa
  • Somewhat drier air between 500 and 800-hPa
• Upward vertical motion exists between 102° and 100°W, but is not collocated with the moistest air or a conditionally unstable layer.
• In the lower troposphere
  • Upslope flow from the east and southeast downwind from the vertical motion maximum
  • Moist air in the planetary boundary layer (below 850-hPa).
  • The upslope is slowly deepening the moist layer, and daytime heating and surface evaporation is destabilizing the air as well.

The effect of convective parameterization on the environment

To see how the convective parameterization in the GFS affects the grid-scale environment,

Development of a grid-scale thunderstorm

At 15Z 22 August, however, the upward motion maximum and conditionally unstable, high RH air do become collocated between 500-600 hPa, at which time the grid-scale convection begins. At this time, only a small amount of the precipitation is generated by the grid-scale scheme, and about 80-90% is convective. However, the convective scheme seems unable to adequately dry and cool the lower troposphere, perhaps because of solar heating, the upslope flow, and/or evaporation of falling grid-scale precipitation into the lower troposphere. By 18 UTC,

• The majority of the precipitation is grid-scale,
• We have a deep, conditionally unstable moist layer from 850-400 hPa in which
  • Grid-scale condensating is taking place and from which
  • Grid-scale precipitation is falling out
• A strengthening grid-scale updraft which has reached 2 Pa/sec.

Vertical motion increases to greater than 4 Pa/sec at 21 UTC 22 August, and the 0° and -15°C isotherms bow upward at the windward side of the moist grid-scale updraft, indicating diabatic heating from grid-scale condensation is occurring.

Dynamical effects in the vertical:

• The life cycle of this particular grid-scale convective system is in excess of 24 hours!
• Early on, the updraft maximum is at between 300 and 600-hPa, somewhat above the level of maximum grid-scale heating (not shown).
  • This is at a somewhat lower pressure (greater height) than one might expect from grid-scale diabatic heating alone;
  • It represents effects of both grid-scale and convective parameterization condensational heating.
  • The updraft is initially tilted upshear with height, becomes vertical as it matures, and finally tilts downshear with height as it weakens.
  • The updraft maximum gradually lowers with time, before weakening begins.
• Divergence is increasing above the maximum vertical motion, and convergence is increasing below the maximum vertical motion
• The updraft is generating positive vorticity beneath and slightly upwind from it, consistent with the dynamical processes presented at http://meted.ucar.edu/nwp/pcu3/cases/etapbomb/page2.htm (courtesy of Dr. Stephen Jascourt, COMET Project Scientist).
  • The major contribution to vorticity generation is vertical stretching in the grid-scale updraft.
  • The level of maximum vorticity lowers as the vertical velocity maximum lowers.

Physical effects:

• Precipitation rates (bottom panel of animation) are near or in excess of 50-mm (greater than 2") per three hours between 00 and 12 UTC 23 August, with two-thirds to three-quarters of it being generated by the grid-scale scheme.
• The high RH takes on a "thunderstorm look" with high RH outflow at anvil level (150-200 hPa) where the updraft detrains, with high RH through the depth of the troposphere below (in the core of the grid-scale updraft).
• The lower-tropospheric convergence into the grid-scale updraft maintains a feed of moist, conditionally unstable air into the grid-scale convective system.

What "kills" a grid-scale thunderstorm before it has a chance to grow?

A couple of interesting features appear in the animation, separate from the main grid-scale precipitation bomb. Starting at both 21 UTC 20 August and 03Z 23 August, moist conditionally unstable air again becomes collocated with upward vertical motion between 300-500 hPa at the western edge of the cross-section, and the updrafts intensify. Both these situations are initially associated mostly with light-to-moderate grid-scale precipitation, with additional convective precipitation at about a 2-3 to one ratio. The updraft intensification in these cases, however, only lasts for 3-6 hours and then the updrafts collapse.

What is different in these cases is the existence of enough dry air beneath the grid-scale updraft and precipitation to create a compensating grid-scale downdraft, probably as a result of evaporational cooling within the drier layer. The downdraft is quite prominent, for example, beneath the updraft at 09Z 23 August in the latter case. Divergence can be seen in the cross-section of the wind field in the lower troposphere at this time. Additionally, the 0° and -15°C isotherms can be seen to bow downward in the grid-scale downdraft, indicative of diabatic cooling.