Slide number 
Title 
Slide Text (bold) and comments 
1 
Front page 
The first session in this series focused on integrated sensor applications to analyzing lowlevel boundaries. This session now extends beyond to how you can use integrated sensor data to analyzing lowlevel boundary characteristics considered to be relevant to initiating thunderstorms. 
2 
Objectives 
1. To diagnose boundary characteristics considered relevant to initiating deep convection.
For the first goal, there are many boundary characteristics you should be able to determine starting with the available CAPE and CIN of the environment, strength of the lowlevel convergence, depth of the boundary, environmental shear and boundaryrelative flow, and alongboundary variations like intersections.
2. Improve short term forecasts of convective initiation.
At the end of this session, you should be aware of the most recent theoretical work on boundary characteristics that modulate the initiation of convection. You should also be aware of several techniques that you can perform using AWIPS to apply these theories to short term forecasting of thunderstorms.

3 
Scope of this session 
1. Convective initiation refers to deep, moist convection.
The definition here of deep moist convection is that which produces precipitation. Any cumulonimbus cloud applies in this case.
2. Limited to initiation directly resulting from surface boundaries.
We exclude mentioning initiation resulting from elevated convection caused by warm air ascending well behind a boundary such as a warm front. Surfacebased convection forms from direct parcel ascent from near the surface (within 100’s of meters AGL). However, we consider only the surfacebased convection that forms within 50 km or 40 minutes of an identifiable boundary rooted in the Planetary Boundary Layer (PBL).
3. Initiation mechanisms are considerations, not hard fact.
Convective initiation is one of the least observed and least understood phenomena in meteorology. What is presented to you in this session are the currently understood theories and considerations on what boundary characteristics are important in convective initiation. In the next couple years, the theories presented here may change significantly based on new modeling and observational studies. There is a planned field project called TIMEX (Thunderstorm Initiation Mobile Experiment) in 2002 that will sample boundaries with unprecedented detail. We should learn more from this one experiment than we know about convective initiation to date.
The TIMEX web page is at: www.nssl.noaa.gov/timex

4 
Topics covered 
Objectives for each topic are listed in this font. This page will show up as we begin each new topic.
1. Assessing CAPE/CIN
A. Be aware of how mesoscale ascent can radically change Mixed Layer CAPE.
B. Understand the concept of the vertical lifting distance to the LFC and its vertical profile. Understand what is a favorable vertical profile for convective initiation.
2. Assessing lowlevel convergence across a boundary.
3. Measuring depth of ascent zone in a boundary.
A. Depth relative to the LCL and LFC.
4. Estimating potential depth of ascent zone.
A. Boundaryrelative steeringlayer flow
B. Boundaryorthogonal shear across boundary.
5. Identifying alongboundary variations and interactions.
A. Boundary mergers, collisions.
B. Interactions with free convective cells, rolls.
Mention this is a table of contents on what we’ll be covering.

5 
CAPE/CIN 
Positive CAPE is a necessary but not sufficient condition for convective initiation. Environmental CIN is also not a sufficient condition for initiation. Therein lies the problem.
Take a look at sounding or model based CIN fields in the warm season and you will see vast areas of zero CIN and positive CAPE with no thunderstorms present. Clearly removing CIN is not a sufficient condition to initiate convection.
Considerations:
• Strong mesoscale updrafts can force air through large CIN. However, initiation not likely in zero CIN unless there is some forced updraft.
What is meant by forced updraft is a mesoscale region of ascending air that helps provide a support base for subsequent parcels to reach the LFC. This mesoscale updraft can arise from free atmospheric convective cells such as rolls, rayleigh convection. Mostly, boundaries, topography create the mesoscale ascent.
• Initial cumulus and TCU rarely follow the theoretical surfacebased moist adiabat.
This will be shown in frame 7 where we examine where entrainment forces the parcel thermodynamics to deviate from the theoretical curve.
• Virtual temperature corrections have an uncertain impact on CAPE/CIN. The next slide shows an example of this.

6 
CAPE/CIN: Schematic of a boundary 
• Updraft from density gradient and/or lowlevel convergence, CL can double MLCAPE (Mixed Layer CAPE) and locally reduce CIN to zero.
This is not surprising since Mixed Layer CAPE uses a layer and moisture needs to be well mixed . What maybe surprising and a bit sobering is that MLCAPE can double or more depending on whether the sounding was launched inside out outside of a boundary. Even Horizontal Convective Rolls (HCRs) can lead to the same differences in CAPE. Evening soundings half a roughly 50/50 chance of being launched inside or outside an HCR updraft zone.

7 
LFC Considerations 
Assume a 1.5 km deep outflow boundary pushing air upward
Surface air will be displaced upward at least 1.5 km
Air at higher initial levels also is forced upward but with less total displacement.
Convection is likely if an airparcel at some level is lifted to the LFC corresponding to that parcel.
Imagine air flowing over a solid obstacle such as a mountain range or, in this case, a sharp outflow boundary. Air trajectories would likely follow the paths outlined in this figure. The lowest level air would be forced upward the most. However, as the initial parcel level increases, the forced vertical displacement diminishes. This kind of lifting profile is a rough estimation of reality derived from laboratory observations by Simpson (1969) (QJR. Meteorl Soc., 104, pp 413)and radar observations by Wakimoto (1982)(MWR, 110, pp 1060)
The flow patterns may not match what is shown here but the vertical displacements relative to the LFC can be illustrated here. The LFC changes according to the sounding and the level of the parcel origin. In this diagram, the parcel lifted along the yellow trajectory successfully reaches its LFC to initiate deep convection.
Foote (1984) used the theoretical lifting profile of air forced upward by a gustfront to the vertical profile of lifting distance needed to air parcels at various levels to reach their respective LFCs. The application, which will be shown in the next few frames, shows which vertical moisture profiles are favorable for boundaries to initiate convection.

8 
LFC: GOES vis loop centered at KEDW 
To serve as an example, a convectively active day in Southern California (Sept 02, 1998) was chosen to examine the propensity for the atmosphere to produce convection. We will 17 UTC sounding from Edwards AFB (highlighted by the red arrow)which was launched prior to first convection forming on the topography surrounding the LA basin. During this loop, KEDW remained on the western edge of the most active convection except for the one small thunderstorm that passed overhead.
As an aside….The first frame of the loop shows the ceiling heights at about 6500 ft AGL which, if is the base of cumulus clouds and the boundary layer was well mixed, would yield a dewpoint depression of 30° F. The actual dewpoint depression was 37° F.

9 
Consider this sounding, KEDW 09021998, 18 UTC 
• Would convection form with a 1.5 km deep outflow boundary? Given an outflow boundary plowing into the airmass represented by this sounding, could deep convection form?
• Consider the amount of lifting required to bring an air parcel from its initial height to its LFC called H_{lfcz}. For example, a surface air parcel would need to be lifted 2.1 km. Therefore, H_{lfcz} = 2.1.
• Then consider a vertical profile H_{lfcz} and compare it to a theoretical vertical profile of vertical air displacements by Foote, 1984.
We will lift air parcels up to their LFCs starting at the surface, then at 925mb, 900mb,…, 725mb. Then we can compute a vertical profile of H_{lfcz} . In two frames there will be a graph showing the vertical profile of H_{lfcz} and a theoretical vertical lifting distance profile from the outflow boundary. But first…
The next slide will be a modification of the KEDW sounding.

10 
Modified KEDW Sounding 
• This modification could have come from moving the sounding launch site underneath a group of cumulus.
• What would happen now if a gust front lifted this air?
• We’ll compare both H_{lfcz} profiles to a theoretical lifting profile by Foote, 1984.

11 
Foote 1984 diagram 
• Look for where the H_{lfcz} profile merges or crosses over the theoretical lifting profile (green dashed lines)A 1.5 km deep gust front could potentially lift air residing at 750mb to the LFC in the modified KEDW sounding represented by the dashed orange profile. Or in other words, the air at 750 mb would need the same amount of lifting as provided by the forced lifting along the gustfront.
• The unmodified sounding would remain dry represented by the solid blue profile.
The ordinate is the parcel starting pressure level. The abscissa is the vertical lifting distance H_{lfcz}.
The area between the green dashed lines represent the potential vertical lifting profile from a barrierlike outflow boundary. For example, the air initially residing at 850 mb would be lifted from 1 to 1.3 km.

12 
LFC Forecast considerations 
• Application of Foote’s 1984 technique possibly most useful for advancing density current boundaries?
• Favorable convective soundings occur where:
• LFC heights are low
• Boundary depth is deep relative to the Hlfcz profile.
A decreasing lifting distance with height Hlfcz indicates the sounding is getting closer to saturation and that air from multiple levels can initiate deep convection. Low LFC heights makes it more possible for weaker boundaries to initiate deep convection.

13 
LFC Caveats 
• For different frontal heights, scale the forced displacement curve accordingly.
For example, if you know that an outflow boundary is 2km high, then multiply every point on the green curve by 2/1.5.
• The theoretical lifting profile:
• Over a gust front may not be valid.
• Does not apply to nondensity current boundaries.
This is a rather old theoretical treatment of lifting over a density current boundary. Air parcel trajectories may differ greatly and in fact we’ll see that shear and mean flow profiles relative to boundary motion can impact the depth of the ascending zone. Non density current boundaries will likely have a completely different set of air trajectories within the ascending zone.
• LFC distance profiles (Hlfcz) may differ considerably in the space of a few km.
Consider the large moisture profile differences from within a convective roll updraft to that just outside.

14 
Topics: Estimating Boundary Depth 

15 
Measuring the depth of an ascent zone: And the depth relative to the LCL, LFC. 
As the depth of ascending air column relative to the LFC increases, so do the probabilities of convective initiation.
There is no guarantee of convection if the ascending air zone does reach the LFC. There is still entrainment to contend with above the LFC. But the odds are improved.
An AWIPS user has a few observational tools to make broad estimations on the depth of a boundary and its ascent zone.

16 
Depth of a Boundary Ascent Zone: radar estimations. 
• The beam is entirely occupied by echoes closer than point A.
Reflectivity should then not be as much of a function of distance for points closer than A. The only thing that can affect reflectivity is the signal to noise ratio for weakly returning fineline echoes.
• The fineline partially fills the beam between points A and B.
• Beyond point B, there is too little echo for detectability. However, there could still be echo in the beam volume. The echoes might fall below the signal to noise ratio.

17 
Depth of a Boundary Ascent Zone: radar estimations. 
• One does not know where the boundary height is relative to the beam center height at point B.
• Once the beam top exits the top of the boundary, reflectivity decreases at point A.
The beam top starts exiting the top of the boundary fineline and the returned reflectivity decreases more rapidly as the percentage of target coverage drops. Knowing the height of the beam top is more crucial than knowing the beam center height. One does not know where in that beam volume the fineline top is located.
• Therefore, use the beam top elevation to estimate the depth of a boundary.
Therefore the best method for detecting the top of a linear feature like a fineline is knowing how high the beam top is when the reflectivity starts to fall. For that reason, we have developed an applet for the web that is located on the front web page for this session.

18 
Depth of a Boundary Ascent Zone: 88D Base Velocity Considerations 
• Typically, air should be ascending above the maximum height convergence is found.
Unlike reflectivity where you seeing direct evidence of ascending air manifesting as finelines. Here, you are only observing convergence/divergence. Calculating the integrated values of convergence below and up to the level of the radar beam is the only real way of knowing the sign of the vertical velocity. If you see convergence in the velocity display, most likely there is ascent occurring at that level. In a classic frontallike circulation, the level of nondivergence is where the maximum ascent occurs. Therefore, ascending air should exceed the height of any convergence you see.
• However, detecting convergence is difficult due to the large default velocity increments of ± 10 kts.
Most weakly forced boundaries exhibit less than 20kts of crossboundary wind differential. The current velocity increments would never adequately capture most boundaries, especially when the wind directions are mostly tangential.
• Solution, when possible reduce the velocity increments in the low range at the UCP.
Look in the Selection of product parameters menu in the UCP then pull of the velocity submenu. You can reduce the increment levels in there when you’re experiencing difficulty in discerning boundaries. Don’t forget to change it back when severe storms start forming.

19 
Depth of a Boundary Ascent Zone: 88D Base Velocity Convergence Depth (contd) 
• Velocity gradients may also be entirely tangential leading to poor detection as seen in this image at pt. C.
A boundary passing over a radar maybe completely invisible in the base velocity if the winds on both sides of the boundary are tangential. Try to draw various windfields that might not show in a clear air base velocity display.
• However, velocity echoes display at lower threshold reflectivity.
The precipitation mode reflectivity displays do not show the lowest reflectivity that the radar observes. The base velocity shows velocity for all observable echoes. Therefore, you can observe boundaries on occasion out to greater ranges in the base velocity when the radar is in precipitation mode.

20 
Depth of a Boundary Ascent Zone: Satellite Applications 
• An organized cumulus line implies that mesoscale ascent is reaching the LCL.
This is a simple but very useful benchmark for knowing when a boundary ascent zone reaches the LCL. For all the capabilities of radar, it simply cannot observe cumulus clouds. And it’s always a good idea to have cumulus to develop before a storm develops. The image on the right is a highly magnified view of a boundary in GOES8 visible imagery. There are two eastwest cumulus lines that indicate boundary lifting through the LCL.
• Organized TCU indicates deeper boundary lifting and more likely initiation.
This is not so easy to discern. There is no benchmark for observing when cumulus break through the LFC. Towering cumulus are also difficult to define from satellite data alone. One good candidate for a benchmark is the 3.9 micron reflectivity or albedo data during the day and the fog product or nocturnal 3.9 micron albedo product at night. These products (especially the daylight ones) are very sensitive to the onset of glaciation. In rapid scan imagery, the timing of glaciation in cumulus can be an important benchmark to anticipating storm development in a short term forecast.
• Moisture/temp gradients on lowlevel water vapor imagery indicates deep boundary.
This idea is uncertain. However, the deeper the air on the cool or dry side of a boundary, the more likely, the lowlevel water vapor sounder imagery will observe a temperature/moisture gradient.

21 
GOES8 vis loop centered @ LBB 0531952230 UTC 
There are two curious manifestations of outflow boundaries in this loop:
Boundary # 1 is just emerging from a group of thunderstorms as an arc of stratiform clouds. This is not surprising judging from the lack of cumulus in the air preceding this boundary. Although there isn’t enough information to make any claims about the stability of the boundary yet. In any case, the forced lifting is reaching the LCL of the environment, whatever depth that might be.
Boundary #2 is not easy to define. Upon looping the visible imagery, notice the cumulus clouds moving east and dissipating along a north to south line just south of the label. There is a boundary along that line of dissipation. Go on to the next two slides for an closer look.

22 
The KMAF sounding 19950601 00 UTC 
This sounding was taken a bit south but probably in the same environment as the area west of Boundary #2. The LCL is around 640 mb and the mean subcloud flow is westerly. 
23 
Radar reflectivity, velocity and GOES8 VIS comparison @ LBB 0531952252 UTC. 
The Base reflectivity image shows a strong fineline associated with the boundary #2. Recalling from Session 1, finelines with reflectivities > 15dBZ likely arise from particulate scattering from insects or plant debris lofted vertically in an updraft. Therefore, ascending air is most likely up to a certain level along the boundary. Base Velocity supports the idea of ascent because of the strong convergence visible along the boundary. Ascending air reaches above the level of convergence. Ascending air is not likely, however, at the LCL of the western airmass since there is no enhancement of the cumulus in the GOES8 VIS image.
The thin green line crossing NW to SE across the radar site representes a velocity cross section in the next page.

24 
Crosssection following the green line 
This crosssection cleanly shows the depth of the convergence zone along the leading edge of the boundary reaching a height around 3000 ft ARL. Note a horizontal belt of echoes from 9000 to 11000 ft ARL left of the red pointer. These are probably cumulus with bases around 9000 ft ARL. The crosssection shows well the large space between the top of the boundary and the LCL.
Also note that the shape of the boundary head consists of an initial rise in depth followed by a depression. This matches the shape of outflow boundaries produced by numerical and theoretical results.

25 
The Depth of a Boundary Ascent Zone: Wind Profile considerations 
In a timeheight profile, the boundary motion vector needs to be known so that a time to space conversion can be done to convert the time from the boundary passage to distance from the boundary passage. For the VWP on the right, the surface boundary passed through around 1808 UTC. Given a boundary motion (estimated elsewhere) of 15kt from the north, each volumescan away from its passage represents about 1.5 nm further distant from the boundary. You can see the windshift is not immediate but rather more gradual with such a fine resolution wind profile as the VWP.
• The depth of the windshift might be related to the depth of the convergence.
This is a similar problem to that of the 88D base velocity. Convergence can be estimated by the shift of the winds as the boundary passes. But to really calculate the vertical velocity profile, the convergence between two wind profiles on either side of the boundary needs to be calculated and then integrated. For visual inspection, the deeper the initial windshift, the more likely the convergence is deeper.
In the example on the right, the wind shifts immediately for the first two elevation gates (1 and 2 kft) above the radar in one volume scan. The 3kft gate wind remains unchanged for another 20 minutes (or 5nm behind the boundary). Finally, the 4kft wind (4th level) remains unchanged for 28 minutes before going to ‘No Data’. Most likely, the boundary interface has reached close to the 4kft range in 28 minutes.
It is difficult to say which wind profiles should be compared to get a picture of what the convergence is between the two sides of the boundary. In this case, say the width of the fineline is about 3nm. The wind profile at 1808 UTC would be inside the fineline. The 1803 profile sampled the air previous to the boundary. The 1813 profile could be close to the back edge of the fineline while the 1818 UTC should be clear of the back edge of the fineline. Assuming the winds directly in front of and behind the fineline are characteristic of the pre and postboundary winds, and the convergence depth is equal to the depth of the windshift, then the convergence depth is between 2 and 3kft above radar level.
• The level of maximum ascent will likely be at the top of the convergence zone.
Cautions:
• Large VAD range will smooth out true wind gradients.
If the VAD range is very high, both sides of a boundary will be sam pled and a wind direction estimate will be attempted using data from disparate windfields. The default VAD range is 33km. A boundary moving at 20km/hr means there will be disparate wind fields for over an hour. If possible, bring in the VAD range as close as possible to the cone of silence at times of boundary passages.
The VAD range is the range out from the radar at which horizontal winds are calculated.
• Poor vertical resolution will allow for only .
• Horizontal variations in boundary depth not accounted.
Just like in the surface convergence estimations, wind profiles can only be used to estimate broad estimations of boundary depth convergence when you can safely ignore alongboundary variations in wind.

26 
VWP comparisons at KLBB 19950531 2252 UTC 
The three frames in this page loop through a VWP taken at KLBB also at 2252 UTC. Of the three VAD ranges chosen, only the 10km range resolves the southeasterly lowlevel winds on the cool side of the boundary. Compare this page with the base velocity image on page 23.
This illustrates the importance of being cognizant of how the VWP realistically depicts the environmental winds in regions of strong windshear. Decreasing the VAD range not only helps to resolve shallow windshear layers but it can also more accurately measure the lowlevel velocity gradients of boundary passages.

27 
Depth of a Boundary Ascent Zone:
Interim Summary 
• From radar reflectivity:
• The depth of the radar fineline marks the depth of ascending air.
• As the top of the beam exits the top of a fineline, reflectivity will decrease sharply,
Estimating the depth of the radar fineline may underestimate the depth of the ascending air column especially for higher slices. It is not always the case that echo causing particulates reach as high as the ascending air column.
• From wind profiles and velocity:
• Depth of convergence can be estimated
• Ascending air should be higher than the convergence depth.
It is not clear how much higher the ascent reaches above the convergence layer.
• From satellite:
• Presence of cumulus indicates ascending air reaching LCL.
The uses of satellite will probably be greater than the other sensor sources when you actually forecast convection. It is the only tool that explicitly observes growing cumulus convection.

28 
Topics
TwoDimensional Boundary Considerations – Assessing Lowlevel Convergence. 

29 
Lowlevel Convergence 
Boundaries with the greatest persistent convergence would have the highest potential for initiating convection given all other parameters being equal. More details two sections later. The persistence helps account for how the boundary lies in relation to the surface observations. Often what may seem to be convergence could be a bad wind ob, or minor fluctuations in the reported winds. Persistence adds credibility that something is real there. Cautions:
• Most boundaries converge air on scales too small for even mesonetworks to resolve (Crook 1996).
The next several frames point this caution out in even more detail.
• Strength of convergence doesn’t always mean deep convergence. A boundary may have strong convergence down low and be capped well below the LFC.
This is a problem especially when the LFC is high and/or if there’s a significant capping layer below it. Drylines are notorious for exhibiting strong sfc convergence and yet shallow circulations.
One thing to remember is that when looking at the vertical continuity equation, the vertical velocity at some level depends on the integrated values of convergence/divergence below. Vertical velocity is highest at the level of nondivergence where convergence resides below and divergence above. In most cases, the deeper the convergence, the higher the level of maximum vertical velocity will become.

30 
reconciling station spacing 
• Consider air converging uniformly up and down the boundary (green vectors).
One could easily imagine a boundary as a two dimensional interface between two regions of differing winds. In this example, there are no irregularities in the wind field on either side of the boundary as one travels up or down the boundary.
• Then the only component of air doing the converging is that oriented perpendicular to the boundary (red vectors).
• In this case, convergence can be calculated by calculating differential velocity parallel to the baseline (yellow line). Which baseline distance?
Convergence doesn’t need to be calculated along X and Y cartesian coordinates. There is radial convergence and in this case, convergence calculated along a rotated coordinate system. The only axis the red vector winds are converging upon is the axis of the boundary. Therefore, the baseline that we calculate convergence is the only baseline down which the winds change which are the ones perpendicular to the axis of the boundary. You can almost think of convergence calculated in natural coordinates where n is the distance along the yellow baselines. Then dv/dn is the way to calculate convergence.
The problem is, over what length (n) do you calculate convergence? The best length is that where the winds change. The next frames show a typical boundary and typical convergences calculated using typical station spacings and a realistic convergence over the right baseline length.

31 
Low level Convergence: What is the right distance? 
First consider a baseline typical of poor METAR coverage (140km). Dv/dn for the longer line left is 2.8 X 10^{6 }s^{1}.
The dv/dn = convergence where n= 140km and is along the long yellow baseline. Since we only know the winds from the two METAR stations at the end of the baseline, convergence can only be calculated over this huge distance.
Now consider dense station coverage of 46 km. Dv/dn along the shorter line left increases to 7.1 X 10^{5} s^{1}.
Now adding the mesonetwork of stations, we have a much shorter station spacing. In this example, convergence between the two endpoints of the 46km baseline is up an order of magnitude from the METAR based convergence.
But is that where the air is starting to converge in reality? Let’s go to the next frame.

32 
Lowlevel Convergence: Spatial considerations 
• The 88D observed the width of the ascending zone to be 10km (the purple line segment) in this boundary.
The clear air echoes in the boundary fineline are the result of insects and plant debris caught in ascending air. In that case, the width of the convergence line is much less than the station spacing of even the mesonetwork.
• After assuming the winds at each end of the yellow line are also true at either end of the purple line segment.
Some of the stations away from the baseline but close to either side of the boundary suggest the winds are pretty homogeneous until right next to either side of the boundary fineline.
• The convergence increases to 6.86 X 10^{3} s^{1}, 3 orders higher than with METARs alone.

33 
Lowlevel Convergence: To summarize the considerations. 
• Actual convergence in the boundary may be 3 orders of magnitude higher than using METARs alone. See graph below.
• Convergence should be estimated across the width of the boundary.
• Estimating one dimensional convergence assumes a uniform airflow on either side. In other words, a front with no undulations or intersections with other boundaries.
• Convergence will be higher at intersections, mergers, collisions with other sources of ascent.
Don’t forget the inverse as well. Convergence across a boundary will be lower where there is alongboundary divergence.

34 
LowlevelConvergence: Checking this out yourself. 
• The key is to figure out the width of the ascending air zone.
See the example on the right side. The width of the ascending air on this boundary is taken to be the width of the small cumulus line. Using the ‘Distance Bearing’ function in D2D, the boundary width is about 4mi. There are a couple things to consider when attempting to estimate the width of the boundary
• The ascending zone maybe sloped causing an unrealistically large width of the boundary.
• The cumulus clouds mark the upper part of the ascending zone and may not be the same width as the region of convergence down near the surface.
• Then estimate the winds directly on either side of the ascent zone.
This is a challenge here because of the lack of stations. However, observing the wind tendencies at each station as the boundary passes is a possible space/time method of estimating the pre and post boundary winds. In this case, the preboundary winds were westerly at 7kts. The postboundary winds were northerly 15kts.
• Finally take the difference in the wind component normal to the boundary divided by the width.
First, you need to know the boundary orientation. The long blue line from the ‘Distance Bearing’ tool marks the rough orientation of the boundary. The line is oriented 100°. Therefore the line normal to the boundary is oriented from 010°. The component of the pre and post boundary winds should be taken along 010°. Then the difference of those components should be divided by the width of the boundary in meters.

35 
Low Level Convergence: Theoretical effects on boundary depth. 
• Applications to density current boundaries.
As in the shear considerations, these theories were derived from density current boundaries.
• A boundaryrelative headwind (tailwind) increases (decreases) the depth of the ascending air column.
Based on numerical modeling of an outflow boundary, the boundary height increases when a boundary moves into a headwind.
• Boundaries with a headwind move more slowly than those with a tailwind.
• From Liu and Moncrieff, 1996 MWR, 124, pp 2282.
From these theoretical and numerical experiments, the ascending air column will be deeper for boundaries experiencing a significant headwind than for those following into a tailwind. It stands to reason that lowlevel convergence with headwind (tailwind) boundaries is higher (lower).

36 
Lowlevel Convergence: Interim Summary 
Convergence is scale dependent
 Data from METARs and mesonets will underestimate convergence in boundaries by 2 to 3 orders of magnitude.
 Model analysis will underestimate convergence similarly.
All operational model resolutions are too poor to capture the true width of a boundary. Also, stormscale boundaries are typically unobserved by models.
Typical boundaries are 310 km wide.
 Actual convergence will be occuring on these scales.
• Convergence will be higher at boundary intersections with other boundaries, circulations, rolls.
In this case, two dimensional convergence becomes important since winds vary both along and normal to the boundary axis. • The deeper the convergence zone relative to the LFC, the more likely initiation will occur.

37 
Topics: 2D Boundary Considerations – Boundaryorthogonal flow and shear 

38 
The Potential Depth of an Ascent Zone: Boundary relative flow and shear considerations. 
Many times, the actual depth of the mesoscale ascent cannot be measured. Fortunately, recent research results indicate there are observable factors that are related to the depth of a boundary ascent zone.
There’s a growing body of research including theoretical and observational that may promise in forecasting convective initiation on boundaries without having to directly observe the depth of a convergence zone or fineline. Simply knowing the properties of the windfield in the reference frame of the moving boundary may give some important considerations as to how deep a boundary ascent zone maybe which leads ultimately to the propensity for convective initiation.
Factors of influence on estimating the depth of a boundary ascent zone:
 Boundaryrelative lowlevel flow.
 Boundaryrelative steeringlayer flow
 Boundaryorthogonal shear.
Each of these topics will be considered in more detail in this section. There are addendums and quid pro quos in this section as with the last one. These concepts will not explicitly tell you where convection will form. Also, boundaries exhibiting unfavorable values of the above parameters can also initiate convection. Also, we do not know many of the characteristics we should be observing. Some or all of these theories may altered or even discarded as more evidence becomes available.

39 
Boundaryorthogonal steering layer flow 
The key is to maximize residence time of a parcel in the ascent zone to ensure it reaching the LFC.
Related to deep layer shear considerations but this applies to any boundary.
When a boundary does not have any density current characteristics, the shear considerations do not apply in the sense in which they were presented. In some sense shear is still important since boundarynormal shear can be too strong potentially ripping apart an ascending zone of air before it can reach the LFC.
However there is another way to look at the potential depth of an ascent zone. Given an ascending air column reaching toward the LFC, what kind of flow aloft will favor the ascending air column to remain over the boundary for the longest time? And what kind of flow aloft should be considered useful?
Observations by Wilson and Megenhardt, 1997.
 Boundaries in FL were most active if cell motion boundary motion 5 m/s.
 Before storms, the 2 – 4 km mean wind = cell motion.
Several observational studies including this one sampled the mean flow in a 2 4 km layer and found the boundaries were most convectively active when the boundarynormal flow was less than 5 m/s. The 24 km layer was chosen since it correlated best with new storm motion. Boundaryrelative flows greater than 5 m/s exhibited less storm activity and those storms that did form on the boundary would quickly be left behind in less stable air and dissipate.
This study was done during a field project around Cape Kennedy during the summer. Therefore the storms were mostly driven by surfacebased phenomenon and there were small amounts of shear. Whether these findings can apply to stronger synoptically forced situations still remains a question worthy of testing at any of your locations.
Note that not all storms will dissipate if the boundaryrelative 24 km flow is high. Convection can persist or even thrive well after they leave the original boundary if there is still sufficient uncapped instability.

40 
shear/flow: expanding circular outflow 
Let’s assume a southwesterly steeringlayer current and a circular outflow boundary created by a past thunderstorm. We’ll also assume that there’s no shear in the 06 km layer. Also, this is a weakly forced environment where convective initiation is based solely on lowlevel forcing and solar heating. We will also assume that the atmosphere inside the circular outflow boundary is stable with respect to surfacebased convection.
The first frame shows cumulus equally distributed around the expanding outflow boundary.
The second – fourth frames show that the cumulus on the northeast side of the boundary are able to ride with the boundary. These parts of the expanding circular boundary have a low boundaryrelative steeringlayer flow. The other cumulus quickly advect over cool air left by the outflow and dissipate.

41 
shear/flow: moving 2D boundary 
This second loop travels with a twodimensional boundary. Each cumulus cloud is embedded within a different steeringlayer flow. The top travels with the boundary. The middle cumulus travels more rapidly than the front such that its boundaryrelative motion is reartofront. The bottom cumulus cloud travels more slowly than the front such that the boundaryrelative motion is fronttorear.
Supposing that the atmosphere is such that a front is needed to initiate convection, the top cumulus cloud would have the best chance to develop since it has the maximum residence time above the boundary.

42 
Shear considerations 
• These theories apply to density current boundaries.
Density current boundaries include strong fronts, seabreeze boundaries and outflow boundaries.
• Deepest ascent occurs when boundary orthogonal environmental shear balances cold pool vorticity (Rotunno et al. 1988)
• Or when the boundary speed (C) cancels out positive shear magnitude ( U_{L}). C/ U_{L} ~ 1.
When the boundarynormal (or orthogonal) component of the shear is oriented forward away from the cold side of the boundary (positive shear) and it has the same magnitude as the forward speed of the boundary, the conditions are optimal for the strongest, most vertically oriented ascent zone. That is C/ Delta U_{L} ~ 1. This is graphically shown in the upperleft schematic.
What often occurs is shown in the lower right figure where the shear vector is directed front to rear of the boundary (negative shear). In that case, no balance occurs in C/ UL and the ascent zone becomes much shallower.
A question may arise, how can boundary speed C with units of m/s be compared with Delta U_{L} with units of s^{1}? A possible answer lies in the assumed properties of a density current behind the leading edge. The strongest flow behind the boundary exists near the surface. The flow weakens with height until reaching near zero near the top edge. Actually there are numerous shearing eddies along the top edge of the cold pool. Then the velocity difference behind the boundary directed back from the leading edge with a magnitude equal to the forward speed of the boundary. Therefore, Rotunno et al. 1988 proposed that the forward motion of the boundary, C, can be used to compare to the magnitude of the shear (Delta U ) in the layer of air in front of the boundary.
Much has still yet to be understood regarding what shear layer should be used to anticipate convective initiation. None of these theories take into account the thermodynamic profile of the environmental air. Does the depth of the shear layer need to match some thermodynamically based parameter? Is one shear layer important for initiation considerations versus storm maintenance? Or is the shear argument relevant at all? Recent findings suggest there is no relationship between cold pool strength and lowlevel shear in the maintenance of longlived severe squall lines. Refer to this work in review by Evans and Doswell, 00. There has not been enough observational evidence to support or refute the shear theories for convective initiation. However, each NWS office has the tools to evaluate whether boundary orthogonal shear is relevant.

43 
Combining shear, steering layer flow and lowlevel boundaryrelative flow theories 
Here is a conceptual figure showing how likely an outflow boundary from a source thunderstorm will produce new convection for combinations of shear and lowlevel flow.
The depth of the leading edge of the outflow boundary (shaded in blue) corresponds to the depth of the ascending zone. The probabilities of new convection are proportional to the depth of the ascending zone in each scenario. There are no thermodynamic considerations in this schematic. The wind profile is given as a vertical profile of horizontal wind vectors pointing toward which the wind is blowing. The lowlevel flow is depicted by the lowest wind vector while the shear is visualized by the change in wind vectors with height. All the wind vectors are groundrelative. The depth of the shear layer depicted by the wind profile is about 4 km AGL. It is intended that the depth of the outflow is lower than the depth of the shear layer.
This figure is adapted from Moncrieff and Liu, 1999.
The second frame in this loop is a schematic considering the effects of mean layer flow on probabilities for initiation. In these considerations, the boundary depth itself is not known to be affected by shear. Instead the concern is the residence time for developing cumulus. Therefore, a boundary with strong headwinds, may not be able to produce deep convection because the steering layer flow prevents an upright deep ascending column.

44 
Case study from 19990610 KLZK sounding 
The supporting data for this case is displayed here to help familiarize yourself with the start of this exercise. The results of what happened are in the teletraining version of this session.
Here is a brief mini exercise where you issue short term forecasts based on the theories on how boundary depth is related to surface convergence, shear and boundarynormal steering layer flow.
Presented here is a sounding representative of the forecast area to be presented in the next page. Note in this case, there is not much shear present in the sounding but there is a steady southerly flow in the lowest 5 km, moderate CAPE and little CIN.
The next page is a loop of satellite imagery and the goal is to draw low, medium and high probability contours of where you expect the most convection to develop on an outflow boundary to be generated by a thunderstorm nearest to the red pointer.

45 
Case study 19990610 GOES8 vis loop #1 
In this loop, focus your attention on the thunderstorm developing just northwest of the red pointer. The loop also has surface METARS plotted where the upper right parameter is virtual potential temperature instead of pressure (AWIPS will soon have this capability).
In the teletraining version of this session, you will draw probability contours of low, medium, and high chances of new convection around the target storm knowing that a circular outflow boundary is about to develop. Let’s assume the outflow boundary will initially move out at 7 m/s (14 kts)in all directions from its source. Do your thunderstorm probabilities for a period of two hours.

46 
Defining boundary motion and boundary orthogonal shear and flow 
 The boundary motion direction is defined as the direction from where the boundary has moved orthogonal to the boundary orientation.
 Then boundaryorthogonal flow and shear can be visualized.
This is very similar to the coordinate transformations done with stormrelative flow parameters. Remember that for a stormrelative framework, the storm is the origin but there is no coordinate rotation. For a boundary, you can visualize the boundary axis as a an original axis rotated to the orientation of the boundary. Then the origin of the new coordinate would lie on the boundary axis at the intersection of the boundary motion vector (S_{b})
Note that sometimes, boundaryrelative flow and boundaryorthogonal flow is used interchangeably. To reduce confusion, boundaryrelative flow implies the orthogonal component of the flow relative to boundary motion. Boundaryorthogonal flow also implies the same orthogonal component of flow relative to boundary motion.

47 
Potential Depth of an Ascent Zone:
Applying shear/cold pool theories to your area.

This is a 3 slide loop showing how boundaryrelative flow and shear can be visualized on a hodograph.
Slide 1: What is the most potentially active boundary orientation and movement given an environmental: • Shear profile? • Lowlevel wind? • Mean steering wind?
The next several slides are meant to show two ways to visualize the shear, lowlevel wind and steering layer wind in the reference frame of a boundary. These are forecasting aids to help you apply or test the considerations about how these parameters affect convective initiation in your area.
We can use a hodograph to immediately visualize the flow in the boundary coordinate system. This hodograph was taken from 29 June 1998 in Omaha, NE.
Delta V _{04km} is the 0 – 4 km velocity difference or shear.
V_{low }is the lowlevel flow vector.
V_{cell} is the cell motion observed for that day.
We can actually plot the boundary motion vector and the boundary axis at the tip of the vector to show that the hodograph can be plotted in the boundary frame of reference.
Slide 2:
Adding the boundary motion • Visualize Boundary relative flow and boundaryorthogonal shear on a hodograph.
• Plot the boundary motion vector on the hodograph.
To plot the motion vector of the boundary on the hodograph, we chose the vector direction as from which the boundary is moving. The boundary speed is taken as the displacement distance of the boundary normal to the axis of the boundary over a period of time of say 30 minutes. The result is a boundary motion vector called Vfront
• Then draw the frontal axis perpendicular to the vector.
In the frame of reference of the boundary, the orientation of the boundary can serve as a new Xaxis. The red line depicts the orientation of the boundary at the physical point where we estimated the boundary motion vector.
Applying shear/cold pool theories to your area.
Slide 3:
Now the wind and shear in a boundarynormal sense can be visualized.
• Boundaryrelative lowlevel flow is marked by the long yellow vector.
The hodograph shows that a southeastward moving boundary will create a strong headwind in the lowlevels. That should be considered a positive for good lowlevel convergence.
• Boundaryrelative cell motion is marked by the white line.
The white line is actually a white arrow eminating from the boundary to the Storm motion point. The groundrelative cell motion is fairly quick but when taken relative to the boundary, the cell is just barely making it ahead of the boundary.
• Boundaryorthogonal shear (Delta U_{b 04km}) is marked by the cyan line.
The boundary orthogonal shear vector is fairly strong and positive, a positive sign for initiation of convection. Note that the full shear vector was moved from its origin out to the boundary itself. Remember that shear is galilean invariant meaning that it is unaffected by any frame of reference. Also remember that the shear/cold pool theories only consider the boundaryorthogonal component of the shear vector.

48 
Potential Depth of an Ascent Zone:
Applying shear/cold pool theories to your area.
Visualizing boundaryrelative flow on a hypothetical expanding circular boundary.

Alternate way to view boundaryrelative flow.
Now we present a hypothetical circular boundary (like an outflow) conceptual figure where you have all possible boundary orientations. Just like on a hodograph, you may plot the mean wind or cell motion vector, the shear vector and the lowlevel wind vector.
• Determine V_{cell} from mean flow or actual cell motion.
Vcell is the cell motion vector and in this case, it is from the west at 30 (any units).
• Determine which boundary motion will yield the smallest relative cell motion.
Here, boundary motion is defined similarly as with the hodograph situation. Take a direction from the origin out to the cyan circle, the intersection point is the boundary motion vector. Note the boundary motion vector that yields the smallest boundaryrelative V_{cell}. In this case, an eastward moving boundary of 20 (any units) yields the smallest value. The smallest V_{cell} still exceeds the boundary speed by 10.
• Observe actual boundary motion and compare
If you have an actual boundary motion, you can plot the direction of motion on this hodograph. Then you can immediately compare the optimal boundary orientation to the actual boundary orientation and compare. If your boundary was moving out of the northeast, then your V_{cell} would be 35, or rapidly front to rear of the boundary.
You can use this relation to calculate boundaryrelative cell speed. U_{b} = V_{cell} cos(theta) V_{front } where U_{b }= boundaryrelative cell speed V_{cell} = Cell motion or mean steering layer velocity. = angular difference between cell motion direction and boundary orientation. V_{front} = Speed of the boundary.

49 
Potential Depth of an Ascent Zone:
Applying shear/cold pool theories to your area.
Visualizing boundaryorthogonal shear on a hypothetical expanding circular boundary.

Remember, the shear theories have only been applied to density current type of boundaries.
To calculate boundary orthogonal shear…
 Subtract the surface flow vector from a velocity vector 2 4 km AGL from a hodograph or by using Product Maker on model analysis.
This step is for estimating the shear vector of the environment on the most unstable side of a boundary. There is still a question of how deep a layer the shear should be calculated. This would be a good question a local study could answer. As a first application, we choose the 03km shear. At the lower end, you may wish to choose a representative wind in the lowest 100m.
 Use equation 2] to calculate the boundary orthogonal shear (S_{b}).
This equation, U_{ b} = V cos (theta) is similar to the one to calculate boundarynormal flow except for shear, boundary motion is irrelevant. The angle (theta) is still the angle between the direction from which the boundary is moving and the shear vector direction. What if the boundary is not moving? Then the angle for the boundary is taken here as the angle perpendicular to the boundary axis taken from the most stable to the least stable side.
For example, a stationary boundary oriented SW to NE with the most stable air located north of the boundary would have a boundary orientation of 315°. This is just a convention but it does help calculate the boundarynormal shear vector.
 See if V_{front} / U_{ b} ~ 1 results in the most action.
For the example on the right, a shear vector of 180 degrees at 10 kts is displayed as a red arrow at multiple points around the hypothetical circular outflow moving away from a point also at 10 kts. According the Rotunno et al. 1988, the optimal boundary motion vector would be one where the shear vector is pointed in a positive direction (away from the stable side) and the magnitude equals the boundary speed. In this example, an eastwest boundary tangent to the top of the hypothetical circle moving north at 10 kts is the optimal boundary for initiating convection according to theory.
We have applied this hypothetical circular outflow boundary to calculate boundarynormal flow and shear to find out which boundary orientation would be most optimal for initiating convection. In the two past examples, the circular boundary was expanding outward at 10 units/time (e.g., m/s, kts). What happens with a slower or faster boundary speed? Well, for boundarynormal steering layer flow, equation 1 factors in the boundary speed and from that you may calculate the most optimal flow. For boundarynormal shear, only boundary orientation is important so that you may calculate the component of shear normal to the boundary. However when you are looking for the optimal condition where shear balances out boundary speed, V_{front} / Delta U _{b} ~ 1, then boundary speed comes into importance.

50 
A boundaryrelative flow calculator applet 
We have created a Java applet that lets you calculate boundarynormal winds.
For wind direction and speed, simply put the direction into Cell Dir. and speed, place it into Cell Speed. Any wind or cell motion is fine for this calculation. For boundary speed, you can place any observed or estimated number into Boundary Speed. The calculator will show you the boundarynormal motion vector for various boundary orientations.

51 
Potential Depth of an Ascent Zone:
Example case study: Using the Calculator.

On D2D:
 Using the ‘distance bearing’ tool, orient a segment parallel to the target boundary.
Upon clicking the ‘distance bearing’ tool, numerous line segments appear on a plan view map. After making these lines ‘editable’, take one of them and oriented it parallel to the target boundary. The line segment will have distance between the two endpoints and the orientation as a legend.
 Track the boundary motion with the ‘distance speed’ tool and normal to the boundary orientation that you measured in part a.
Now take the ‘distance speed’ tool to track the boundary in a loop. The direction of motion should be perpendicular to that of the boundary orientation and the direction of motion should be from which the boundary has moved.

52 
Potential Depth of an Ascent Zone:
Example of measuring boundary orientation and motion.
GOES8 visible loop 5 hours after sounding time.

Applying the above with an example.
We go back to the same example that we used to estimate the boundary width when calculating lowlevel convergence. The next frame shows a loop of 1km GOES8 visible imagery where the boundary motion can be tracked and its orientation measured. We measure the orientation of the boundary axis to be 100 degrees (blue distance bearing line).
The boundary is moving south, perpendicular to the axis of the boundary. Making sure we use the ‘distance speed’ utility to track the boundary motion perpendicular to its axis, we find it is moving toward 190 degrees at 25 kt. By convention, the boundary is moving from 010 degrees at 25 kt.
Remember that the ‘Distance Speed’ Tool marks the direction toward instead of from which your object is moving.

53 
Potential Depth of an Ascent Zone:
Example of measuring mean wind and shear.
12 UTC Sounding hodograph at KOMA.


What is the environmental cell motion?
 Track incipient TCU on satellite, new cells on radar.
 Estimate mean wind in the cell bearing layer.
The hodograph on the right was taken from a sounding 5 hours before the previous satellite loop. It shows a 06 km mean wind of 280 deg at 22kt. There were no incipient cells to track to see how well they agree with the mean wind. What is the environmental shear?
 Depth of shear layer? There are conflicting answers.
The 03km shear layer is similar to the mean wind. 
54 
Potential Depth of an Ascent Zone:
Applying observed boundary motion and 06 km mean wind to the boundary applet.

To calculate the boundarynormal 06 km mean wind, we input the mean wind of 280° at 12 m/s (note the units change) into the Cell Speed and Direction boxes. The distance speed tool measured boundary speed at 13 m/s and we input that into the Boundary Speed box.
The boundarynormal 06km wind displays around the perimeter of the hypothetical circular outflow whose speed is moving outward at 13 m/s. Note, however, that we measured the boundary moving from 010 degrees. Therefore, the boundarynormal 06km wind is taken where the real boundary motion and axis(thick blue line) lies tangent to the hypothetical circle.
The 06 km mean boundarynormal wind is slightly greater than the 10.9 m/s value listed nearby. The applet marks the boundarynormal flows that are optimal for initiating convection in red. The values for this boundary indicate strongly negative boundarynormal mean wind and therefore, we would assign a low probability of this boundary becoming convectively active in the immediate future.
The GOES8 loop a couple frames back supports this forecast as there are no growing cumulus along the boundary.

55 
Potential Depth of an Ascent Zone:
2hr nowcast Discussion.

 The boundary is moving too quickly to permit incipient cells to remain in the mesoscale updraft.
 The 03 km boundarynormal shear vector is oriented reartofront of the updraft but C/(Delta) U = 13/4 which is greater than 1 indicating the boundary is overpowering the shear.
 Satellite data shows no developing cumulus ahead of the boundary.
 Nowcast at this time would not favor initiation in the next couple hours near this segment of the boundary.

56 
Topics: Threedimensional boundary considerations 

57 
Alongboundary Variations:
What are they?

The last section dealt with horizontally uniform boundaries. However…
All boundaries exhibit horizontal variations in their properties that can focus initiating convection.
Studies of Boundary intersections, mergers, collisions:

 Intersections with boundarylayer rolls.
 Seabreeze/outflow boundary collisions.
 Intersecting outflow boundaries.
 Dryline triple points.
There have been numerous studies since the 1970’s that observed the tendency for convection to initiate on boundary intersections, mergers and collisions. These features have often been used to highlight areas where convection is expected to develop.
There is no airmass with spatially uniform instability. Boundaries with poor potential to initiate convection based on flow and shear considerations can still be prolific thunderstorm producers given low enough CIN and LFC. 
58 
References 
Previous studies of: Intersections with boundarylayer rolls.

 Wilson et al. 1992 MWR, 120, pp 1785,
 Fankhauser et al.1995, MWR, 123, pp 291.

Seabreeze/outflow boundary collisions.

 Kingsmill, 1995, MWR, 123, pp 2913.

Intersecting outflow boundaries.

 Wilson and Shreiber, 1986, MWR, 115, pp 2516
 Browning et al. 1997, WAF, 12, pp 915.

Dryline triple points, intersections.

 Hane et al. 1997, MWR, 125, pp 231.
There are many more papers that exist and some of them are listed in the boundary bibliography on the main boundary homepage. 
59 
Alongboundary variations:
Boundary Intersections, Mergers and collisions.

These definitions of boundary collisions, mergers and intersections came from Wilson and Schreiber, 1986. They were also used again by Koch and Ray, 1997. 
58 
Alongboundary variations:
Boundary Intersections, Mergers and collisions
Probabilities of initiating convection

All boundary interactions (e.g., collisions, mergers and intersections) are prolific sources of convection. The Wilson and Schreiber study observed boundary interactions in a part of Colorado east of the Front Range using research mesonets and a research Doppler radar. The Koch and Ray study observed boundary interactions east of the Appalaichians in North Carolina using standard NWS operational datasets (e.g., WSR88D, GOES8, surface data). Both observational studies focused on summer convection. However the Colorado Summer convection exists in environments with high LFC’s and LCL’s compared to that of North Carolina. Koch and Ray, 1997 attributed the differences in the two environments to explain the higher percentages of convection with intersecting and merging boundaries. 
61 
Alongboundary variations:
Boundary intersections with rolls.


Initiation is favored where roll updrafts intersect another boundary.

 Especially for large rolls or rolls with the largest cumulus street.
Here is a schematic figure which shows how the updraft axis of the rolls becomes superimposed on the ascending zone of the boundary to create enhanced regions of updraft. Conversely, the downdraft axis of the rolls will negate or cancel out the boundary ascending air to limit the potential for convection in these areas.
Most average roll spacing is only about 4km, the scale of cumulus clouds. There is not much room for enhanced cumulus to grow at the intersection points along a boundary. Occasionally, large or enhanced rolls exist with an aspect ratio of greater than 7. These rolls exhibit greater depths of circulations and will be even more likely areas of initiation where they intersect boundaries.

62 
Alongboundary variations
example of rolls interacting with a boundary

Here is a 3panel overlay in the Abilene, TX area with a weak cold front intersecting horizontal convective rolls. Fade in the satellite over the base reflectivity and you see where cumulus enhancement occurs as the rolls intersect the front. Had there been no capping layer above the LCL, thunderstorm probabilities would be highest at the intersection points of the two largest rolls with the cold front.
Also note that the rolls actually are lifted up and north of the cold front for a short distance.

63 
AlongBoundary Variations:
Variations in Stability.


Large variations in stability exist in very short distances.

 CAPE (CIN) can double (half) in value for soundings within ascending (descending) regions of roll circulations.
There may not be much difference in surface dewpoints between the updraft and downdraft axis’ of Horizontal Convective Rolls (HCRs). However the depth of the surface moisture changes rapidly. Therefore Mixed Layer CAPE will change dramatically between the updraft and the downdraft parts of HCRs. Satellite is an excellent tool for judging relative variations in stability by observing:

 Areas of concentrated cumulus in imager data.
 Areas of enhanced moisture/instability in derived product imagery.
For short term forecasting of convection (02 hr), there is no better way to monitor small spatial variations in atmospheric instability than by observing horizontal changes in the behavior of cumulus clouds by GOES satellite. Boundaries that lie or move into preexisting fields of enhanced cumulus are very likely to initiate convection. Likewise, HCRs with enhanced cumulus are the ones to pay close attention to, especially if they intersect with boundaries. Again, boundaries with unfavorable values of boundarynormal shear and flow for initiating convection may do so prolifically if they interact with a preexisting field of enhanced cumulus.
In fact, monitoring rapid changes in cumulus fields is a more important reason to invoke Rapid Scan Operations than waiting until there are mature severe storms.

64 
Alongboundary variations:
Summary.

Note that there is often a time lag between the actual surface intersection, collision or merger and initiating convection.
For intersections, the convection may appear to lag the intersection point in cases where either or both boundaries are closing the gap before the intersection. Look for initiation about 1530 minutes following the intersection for any one point.
For a boundary moving into a field of enhanced cumulus, look for these cumulus to initiate into storms between 10 and 40 minutes after boundary passage.
Collisions favor convective initiation at the point of collision about 15 to 30 minutes after the collision.
Stationary boundaries will tend to develop storms within 10km of their locations.
These rules of thumb were developed by Wilson and Schreiber, 1986 for typical summer convection. Given different CIN, LFC and instability, and strength of a boundary, these rules of thumb may change dramatically.

65 
Summary 
 CAPE > 0, low CIN
 Region of enhanced cumulus
 Deep boundary relative to the LFC height
 Strong lowlevel convergence
 Low boundaryrelative steering layer flow
 Moderate positive shear?

Initiation is discouraged.

 CAPE>0, higher CIN
 No cumulus in area.
 Shallow boundary relative to LFC.
 Weak lowlevel convergence.
 High boundaryrelative steering flow.
 Negative shear.
Remember that these variables may contradict each other. For example, a boundary may have excellent shear and flow parameters but there maybe a cap too strong to allow initiation. Conversely, a boundary with poor shear and flow parameters may exist in an environment where any forced ascent will cause numerous thunderstorms.
As in warning environments, nowcasting convection requires a careful consideration of all parameters, some of which were covered in this section.

66 
July 13, 1999.
An example of horizontal variations in stability, boundary collisions and multicell propagation.
Upper air plots 12 UTC

This case (pages 6469) represents the second exercise where we present the setup data but leave the results to the teletraining session. If you are planning on attending a teletraining session, we encourage you to analyze the data presented a couple days ahead.
Our area of concentration will be in Minnesota and the Eastern Dakotas.
The white box on the 925 mb plot represents the forecast area.
The lowlevels show a jet arcing northeastward through Southern Minnesota and a stationary eastwest frontal boundary north of International Falls.
There is a weak midlevel shortwave trough passing through Wisconsin and Minnesota. The forecast region is behind the trough axis.
A northwesterly jet extends from Montana to Nebraska. The forecast region seems to be in a diffluent region. The direction in the upper air flow is uncertain in the forecast region.

67 
Case study
July 13, 1999.
An example of horizontal variations in stability, boundary collisions and multicell propagation.
Watervapor loop and 350250mb PV

At first glance, there does not seem to be much forcing aloft. The GOES8 WV with the 250350mb potential vorticity shows relatively zonal northwest flow with one weak wave passing by the forecast region by the end of the loop. This supports the upper air data showing a midlevel shortwave trough already past the region of interest. 
68 
19990713: Regional sfc plot and GOES VIS 
We will defer to your analysis here. 
69 
19990713: Regional GOESVIS and RUCII CAPE/CIN 
In this loop, there is adequate CAPE (green contours) over the whole forecast region. However there are large regions of inhibition (blue dashed lines less than 20 J/kg) that indicate a strong cap. The CIN does drop to between 20 and 0 J/kg in the northern half of Minnesota and into southeast North Dakota by 20 UTC.
The next page shows RUCII 20 UTC soundings corresponding to some of the station ID’s in the last frame of this loop.

70 
19990713: 20 UTC RUCII soundings 
Three soundings are chosen along a line from Duluth (DLH), to Brainerd (BRD) to Fargo (FAR). Note that there is adequate CAPE, low LFC’s and minimal CIN. Which one of these soundings would indicate to you a significantly higher threat of convection?
The CAPE and CIN were calculated using a mixed 50mb surface layer.

71 
19990713: 20 UTC RUCII hodographs 
Each of the hodographs show a WNW shear up to 18 m/s in the lowest 3km. There is almost enough shear to support supercellular storms. The mean flow at 46 km ASL is also westnorthwesterly at 15 m/s.
The teletraining version contains some more information and two rounds where you make the nowcast.

72 
19990610: North and S. Carolina case of weakly forced convection. 
This is the beginning of the last exercise in this session. Again, you need to take the teletraining session or refer to it for the actual exercise. Please review the data in advance if you plan on attending the teletraining.
We’ll focus the forecast area (FA) to the right of the black line visible on the 925 mb plot in North and South Carolina.
The lowlevel pattern indicates a large anticyclone northeast of the FA and a easterly wave type trough to the southwest. An uppertropospheric anticyclone is located southwest of the FA. Thus, lowlevel easterly flow is replaced by mid and upperlevel southwesterlies.
The moisture values plotted here are actual dewpoints and not depressions. The FA is rich with lowlevel moisture but at 700mb and above, the airmass is dry.

71 
1999061012Z GSO sounding 
The first frame is the 12 Z raw sounding. The second frame is the modified sounding using observed surface conditions at 18Z from the nearest sfc observation. 
72 
1999061012Z MHX sounding 
The first frame is the 12 Z raw sounding. The second frame is the modified sounding using observed surface conditions at 18Z at the nearest observing site. 
75 
1999061012Z GSO & MHX raw hodograph 
The GSO hodograph indicates a shear vector of 130° 6 m/s in the lowest 3 km and then a reversal of shear to westerly from 3 – 6 km. This is consistent with the largescale features analyzed in the upperair plots.
The MHX hodograph is somewhat different in that the lowlevel flow is westerly and the shear in the lowest one km is northerly. However, ESE shear exists between one and three km followed by a shear reversal above to WSW. The differing wind direction in the lowest km may be a function of the coastal landbreeze. Later in the day, a seabreeze develops likely leading to a differing wind profile.
Estimating storm motion based on these hodographs is very problematic. The 0 – 6 km mean wind at MHX is 004° 2 m/s and at GSO, 164° 3 m/s. From 0 – H_{lfc}(2km)the mean wind is easterly about 6 m/s. The wind remains easterly up to 3 km. Therefore, cumulus clouds reaching the LFC should be embedded in a steady easterly flow and therefore incipient cell motion should be from the east at 6 m/s. However, mature cell motion is more difficult to discern. Following the mean 0 – 6 km wind, mature cells should remain nearly stationary.
We go on to the first satellite loop. The GOES8 is in RSO mode.

76 
19990610: GOES8 vis loop 16151745 UTC 
The midday loop shows continuing areas of convection in the Smokey mountains and off the South Carolina coast; the latter being associated with the easterly wave. The station models in the surface overlay plot are slightly different. The virtual potential temperature is plotted in place of the surface pressure.
For the instructors in this exercise:
1. have the participants analyze surface boundaries and areas of best instability (using different colors). Refer to the teletraining session for the actual exercise. This loop is here to help familiarize yourself with the case.
