Use of low-level lapse rates in forecasting convective initiation

Convective initiation continues to be one of the most difficult forecast challenges in forecasting severe local storms.  "Will the cap break this afternoon? And if so, when and where?" This is a question that all convective forecasters face on a routine basis during the warm season.  An old,  and popular convective parameter that the former AFOS ADAP short-fuse composite used was the "cap" parameter.  This is a calculation that uses parcel theory to come up with the greatest temperature difference in the lower troposphere between the environment and the lifted parcel... when the lifted parcel is cooler than ambient conditions.  The greater this temperature difference, the stronger the inhibition to deep, moist convection (DMC).  A "cap" so derived, however, ignores several important physical characteristics of the atmosphere such as the relative depth and horizontal displacement of the elevated mixed layer which "is" the cap.

As computer algorithms became more sophisticated with a myriad of parcel theory-based parameters being computed, Convective Inhibition (CIN) grew in popularity among convective forecasters/nowcasters.  CIN is the integrated calculation of the temperature difference between a lifted parcel and its environment.  It is the same calculation as CAPE, only for negative energy (lifted parcel cooler than ambient conditions). Both represent the area under a curve on a thermodynamic diagram and quanitify the entire positive or negative bouyant energy distribution in the vertical.  

Parcel theory has its limitations.  This is especially so when dealing with the convective initiation problem.  How big (or small?) is a theoretical bubble of air supposed to be (and, in fact, is it a bubble at all!), such that it can be lifted to its LFC and continue to grow to become a thunderstorm?  The moisture and thermal properties of the convective boundary layer obviously can not be assumed by observations from a fixed station 2 meters off the ground.  Large assumptions are made in determining a mixed-layer parcel for lifting, not the least of which is horizontal and vertical temperature and moisture distribution.  Given the lack of observations above 2 meters AGL on an hourly basis, we mix observational data with model information to determine a 50mb, 100mb, or 150mb mean mixing ratio and potential temperature to lift.  There are gross uncertanties in doing this; uncertainties that grow in magnitude when you begin calculating integrations from such an assumed lifted parcel (CAPE, CIN, etc). 

The CIN calculation, using parcel theory, can be a great tool in assessing the amount of energy needed to be overcome through forced ascent (synoptic scale ascent, boundary layer convergence, orographic lift, or all of the above).  The observational network, however, does not currently exist to compute reliable values of CIN, and even the latest high resolution modeling efforts struggle with boundary layer thermodynamics. 

Given these uncertainties in the calculation of CIN using parcel theory, and the poor performance of CIN as an indicator for potential DMC initiation in recent events around Southwest Kansas, we are investigating other potential tools for this problem.  Recent research work by Jon Davies has shown that lower tropospheric static stability can be a very useful tool in diagnosing environments suitable for tornadogenesis through enhanced updraft stretching.  The same environmental static stability through a 2 or 3km AGL depth can be a very important signal for the initiation of such storms.  To get surface-based DMC initiation, you need a source of lower tropospheric lift (enhanced mass convergence or mechanical lift), a thermodynamically unstable atmosphere, and adequate low level moisture.  A lower tropospheric atmosphere with low static stability is more vulnerable to convective initiation given sustained, sufficient lift... versus an enviornment with higher static stability. 

Given this basic physical reasoning, we are experimenting with the 0-2.5km AGL lapse rate parameter computed from the AWIPS Local Analysis & Prediction System (LAPS) hourly analysis. We display this parameter in plan view and highlight areas of lower static stability in the crucial layers of the lower atmosphere for convective initiation. In this way, we attempt to use this parameter as a proxy for CIN but in an easier to read presentation more like the old CAP parameter from the AFOS ADAP and without as many of the problems encountered with CIN or CAP.  Examples follow herewith:

Case Example:  20 June 2005 (Northeastern Colorado)

There is a signal of moisture convergence in three areas.  The most coherent area appears to be associated with a surface trough over northeastern Colorado. 0-2.5km AGL low level lapse rate shows lapse rates approaching dry adiabatic through this layer over almost all of eastern Colorado, extending south into the far western TX-OK Panhandles (indicative of a lower atmosphere more vulnerable to convective initiation).  As you go east, lapse rates decrease to ~8 to 8.5 C/km over much of southwestern KS and western OK, decreasing further into central KS and OK, which implies a more statically stable lower atmosphere.

Surface-based convective inhibition (CIN) shows a large area of ~25 J/kg or less of inhibition over almost all of the high plains extending into portions of central KS and OK.  Interestingly, this CIN analysis shows a slightly increased area of inhibition in portions of northeastern Colorado where the 0-2.5km AGL lapse rate suggests a nearly dry adiabatic profile.  SPC's mesoanalysis of Mixed-layer CAPE and CIN.  It also shows an axis of lowest CIN farther east of the highest 0-2.5km AGL lapse rates.  CIN analysis almost always shows a minimum where CAPE is maximized (which, in this case, 4500+ J/kg is being analyzed across western KS into north-central NE.

1km visible satellite showing deep, moist convection erupting up and down the Front Range of the Rockies, as well as along a weak, old Pacific cold front from the Nebraska Panhandle southwestward into northeastern Colorado.  Note that over the area where objective analysis CIN was showing almost no CIN, the atmosphere is convection free with only fair weather cumulus developing in patchy areas.
Dodge City radar shows a marginally severe thunderstorm north of Limon, CO.

Final note:

As of the time of this writing (6/20/2005), the usage of 0-2.5km AGL lapse rate as a forecast tool, in combination with Chart #1, is very experimental.  The physical reasoning, however, is basic and meteorologically sound in identifying regions more susceptible to convective initiation, versus using parcel theory and its inherent limitations and subsequent analysis errors as mentioned above. Furthermore, we will test to see if hour to hour continuity of this parameter allows for the short range spatial and temporal prediction of initial DMC.

We hope to find that our skill in forecasting convective initiation will improve using low level lapse rates with moisture convergence and instability (CAPE/Theta-E) versus using the CIN analysis.

written by Mike Umscheid, NWS-DDC with review by Jim Johnson, NWS-DDC is the U.S. government's official web portal to all federal, state and local government web resources and services.