When I predict severe weather, it usually is a process absent of the details: capping & cloud cover, initiation, instability, moisture & low-level dynamics, upper-level dynamics & speed shear, and low-level spin. Beyond three days, some can't be predicted (surface and lower-troposphere parameters) and others (instability, upper-level dynamics, & low pressure areas) are predicted via the global models (GFS, CMC, ECMWF, UKMET, NAVGEM, etc.) but with uncertainties. They are certain enough for the SPC to outline 15, and sometimes 30, percent risk areas on their Day 4-8 Outlooks.
None of these parameters, however, are predicted by the climate models (Euro weeklies & CFS) that I will use in forming my next entry but I can probably gauge how many systems will pass by the NE or Mid-Atl. states in a given season through pattern recognition (much like a winter forecast. So, you can't do much with a seasonal outlook.
You also can't get the mesoscale details nailed (or are more certain) until three days out because you need the short-range models:
- 16-km SREF - two models ea. w/ 12 initial conditions - out to 87 hr from 3z, 9z, 15z, & 21z
- 12-km NAM/ NMM/ ARW out to 84 hours from 0z, 6z, 12z, & 18z
- Hi-Res --------- " ---------- out to 60 hours ------------- " ------------ (hi-res = 3-4-km) and
- 13-km RAP & 3-km HRRR out to 18 hrs from every hour of the day
- There's also the RGEM (Canadian) model out to 48 hrs which covers most of the N & S-Ctrl US
They are nested within the globals and subsequently look at the details between the fronts and pressure systems the globals detect. Therefore, the short-range suite of models can best predict the severe weather parameters. However, they are uncertain enough beyond roughly 60-70% of each run to warrant a phasing in of these models with the globals. This is also why the SPC's detailed probabilities only go out on day 1 but the
models are improving, which may - in a few years - result in the SPC using (some form of) detailed probabilities on Day 2 outlooks.
UPDATE 2/14/2020: SPC recently started doing those detailed probabilities on Day 2... replacing "max risk by hazard".
Enough background.... Let's delve into the parameters:
AM Capping
A cap is a layer of warm air that exists around 3/4-1 mi
above our heads that keeps severe weather at bay until the peak heating
of the day.
It is best measured on a morning sounding using the temp. profile, which in turn produces the LSI/ CSI (lid/ cap strength index). 2-4°C is breakable. Any lower, the storms fire too soon. Any higher, the storms might not form at all.
There's also CIN: Integral (adding), with respect to height, of the
negative bouyancy (T(parcel) < T(env)), which is measured in J/kg
(energy per unit mass). -75 J/kg is the goal for CIN at the peak heating
of the day. The two heights are usually from the LCL (see
Moisture below) to the LFC (level of free convection, where T(parcel) first equals T(envt).
I discuss how we break the cap later in the entry.
Moisture
You
cannot have a storm without water vapor in the air. You'll know it is
moist if the air is sticky. We measure that several ways: dew pt (sfc to
3/4 mi) then, for upper-air levels... relative humidity (RH) and/ or
dew-pt-depression (T-T(d)). We also use LCL height, which is the height
at which RH is lowest.
To find this height, extend the sfc temp up vertically at the rate of 9.8°C of cooling per 1 km (5.5°F per 1,000
ft) of ascent and the dew pt up vertically at the rate of 2°C of cooling
per 1 km (0.33°F per 1000 ft) of ascent until they meet.
Dew Pts. at 12z (about 1-2 hrs after sunrise ET) should be 50-60°F
and rise to 55-60°F+ before the storms. LCL heights of 2-3 km MSL
([above] mean sea level) is good for any severe weather.
Cloud Cover/ AM-PM Sunshine
Cloud cover
usually contributes to increasing magnitude of CIN, but if they clear by 11a-12p LT, the cap can be overcome. However, sunshine throughout the morning and
afternoon is ideal. The sun mixes any incoming
moisture down to the sfc (increasing the dew pts) and will sometimes
increase the sfc temps to beyond or at the convective temperature. Any lack of moisture and/ or heat at the sunrise
observation is solved.
LCL = ~0.5 km. The yellow line is roughly a parcel, which is negatively bouyant up to 4 km. The turquoise area between the yellow line & red temp profile is the CIN. And the convective temp (blue dashed line) is approx. 22-23°C.
Instability
Instability is interpreted & measured just like CIN: using a sounding (afternoon specifically) but we look from ~1 km to 9-12+ km. At those heights, the parcel is positively bouyant: warmer that its surrounding
environment at or past the
LCL, and it should stay that way for the next 4-8 mi of
ascent (lower in the winter and higher in the summer). This allows a
thunderstorm to become sustained once it forms.
This is
measured in CAPE
(1100 J/kg is good for Ern US severe storms) usually using the avg temp & dew pt in the lowest 30-, 50, 60-, or 100-mb,
called mixed-layer (ml) CAPE -OR-
Lifted Index (meas'd at
500 mb or ~18,000 ft MSL) = T(parcel) - T(envt), which should be at
least -2°C but between -4°C to -8°C. Values of -9 or lower are
considered extreme.
The parcel is now mostly right of (warmer than) the red temp. profile, making it positively bouyant with the turquoise area now representing the CAPE. Sfc. temp. now equals convective temp, CIN is almost entirely gone & LI = -4°C (~5.5 km).
In the cases of elevated convection
(strong cap but storms still possible), lift the parcel from 850 mb to
500 mb and calculate the lifted index again (this is called the
Showalter Index).
Lapse Rates
The
difference between two temperatures (T) as you ascend along the profile
divided by the height (z) between them is called a lapse rate:
[T(UL)-T(LL)]/[z(UL)-z(LL)]. I already gave an example of a lapse rate
when I discussed LCL height: -9.8°C/km is called the
dry adiabatic lapse rate and on a sounding, it goes up and to the left. The 2°C/km dew pt ascent rate is called the saturation mixing ratio and is not a lapse rate. In fact, lapse rates primarily refer to temperature. A parcel from the LCL will ascend via another lapse rate, called the
moist adiabatic lapse rate, which is initially 5.5°C/km but it steepens to the dry adiabatic lapse rate as you ascend.
A low-level lapse rate (sfc-to-roughly-850-mb) of 6°C/km is good but 8-10 is ideal for severe weather. A mid-to-upper-level lapse rate (850-to-300-mb) of 6°C/km is good, but 7-8 is ideal. A mid-level lapse rate (700-to-500-mb) of at least 7°C/km is pretty much required for an outbreak (lots of reports over a large area).
In the above sounding, the lapse rates (in °C/km) are, respectfully: 8.4, 7.2 & 6.4 respectively. For what it's worth, the highest lapse rate between 850 & 300 mb (with at least 150 mb of ascent) is 7.7°C/km.
Initiation
We've allowed for the AM cap to mostly erode. We just need to get a parcel from the LCL to the LFC in order to tap into the instability. Momentum of the parcel from the rise to the LCL may not be enough and it shouldn't (at least before 2p LT). We need a
source of lifting. This can happen with the aid of an old storm's outflow boundary or by a dry line, trough or front. It is boosted by synoptic considerations above 17,000 ft to around 30k-40k+ ft, particularly the jet stream.
(Deep-Layer) Speed Shear
It is the difference in the speed with respect to height, usually measured from the surface to ~50% of the storm depth (effective shear) but 0-6-km, 0-8-km and 0-500-mb shear are also measured/ useful. The wind, due to friction, usually increases with height. It's a vector subtraction, where u is the E-W component of the wind (u=|
v|cos(angle)) and v is the N-S component of the wind (v = |
v|sin(angle)):
SHEAR = SQRT([u(upper-level) - u(lower level)]^2 + [v(upper-level) - v(lower-level)]^2)
DIRECTION = tan^(-1) of [v(UL)-v(LL)]/[u(UL-u(LL)].
Please note that the angle direction is not based on the same "scale" as the equations above: convert METAR/ Radiosonde from "direction from (N is 0°)" to "direction toward (E is 0°) by taking the wind and subtracting 270° and back to the original "scale" after calculating the reverse tangent.
This is a hodograph: wind vectors at different altitudes plotted on a polar diagram. The wind usually increases with height and takes on a more western component so the faster winds at higher altitudes are further from the center of the diagram.
The turquoise line is approximately the effective shear vector (45% of the storm depth) and you subtract the slower wind (left) from the faster wind (right) after converting both winds to the u- & v-components (proper angles).
The preference is that the shear be at
least 20 mph to allow the storm's updraft
to separate from the downdraft and sustain it for more than the 30-60 mins. of an ordinary TS's mature phase.
Storm Mode
Compare the direction of the deep-layer shear vector to the boundary initiating a storm. The more parallel they are, the mode will be linear. The more perpendicular they are, the mode will be more discrete.
Sometimes the mode will be cluster if the
CAPE*SHEAR does not exceed 20,000 (m/s)^3. The underlined equation results in the Craven-Brooks Number (Jeff Craven & Harold Brooks of NOAA, 2004). We'll modify it as CAPE*SHEAR/(1000 (m/s)^3). 10-20 is good but 20-30 is better.
Tornado Requirements
CAPE: 1500 J/kg is better but 1100 J/kg is needed in the warm season
Other Instability: LI < SI and LI at least -4°C, preferably -5 to -6 or lower
LCL Height: 0.65-1.25 km
is ideal, but from 0.35-0.65 & 1.25-1.8 km is ok
Deep Shear: 30-40 kt (35-46 mph) or at least 15 m/s is needed. But 18-20 m/s or higher is ideal
Modes Favored: discrete is best
0-1-km Shear: 7.5 m/s (15 kt, 18 mph) minimum but 10-15 m/s or more is best
And now, we add directional shear using storm-relative helicity (storm motion vector is
c):
Integral - from sfc to either 500 m, 1 km, top of the PBL (just below the cap) or 3 km - of the storm-relative wind |(
v -
c)| * |vorticity| * cosine of the angle between (
v -
c) & vorticity). The higher the SR helicity, the better the spin. On the hodograph below, it is the area of a piece of pie - between the two dark blue lines and between the storm motion vector & hodograph curve (red):
The 0-1-km helicity is roughly the area of the triangle from green R to the 0-1-km shear vector (turquoise line), but it also extends to the red curve of the wind. So, it is more of a summation (integral) of very small triangles.
Values of the SRH as follows (1km, PBL(top), 500m & 3km respectively) are 100-150 (m/s)^2, 250 (m/s)^2, 75% of the PBL(top), and 300-400+ (m/s)^2 are best. Surface-to-PBL(top) SR Helicity is also called effective SRH.
Combo Parameters
Craven-Brooks is one important multi-parameter function. We have Energy-Helicity Index, Supercell Composite and Significant Tornado Parameter)
0-1-km*
EHI (*modified) = sbCAPE (or 0-30-mb/ 0-50-mb mlCAPE) * 0-1-km SR Helicity / [160,000 (m/s)^4]
Values of 0.5-1 or more are good for storms to rotate
0-3-km*
EHI (*modified) = mlCAPE (0-50-mb, 0-60-mb or 0-100-mb) * effective or 0-3-km SR Helicity / [160,000 (m/s)^4]
Values should be greater than the 0-1-km modified version. If those values are 1 or more, storms will likely rotate
SCP = [muCAPE (most-usable) / (1000 J/kg)] * [effec. SR Helicity / 50 (m/s)^2] * [effec Shear / (20 m/s)], w/ the last term set to 0 if Effec Shr is less than 10 m/s or 1 if Effec. Shr is 20 m/s or more
One (1) means that the storms are likely to become supercells with a two (2) or more being ideal
Modified
STP = [mlCAPE / (1500 J/kg)] * [effec SR Helicity / 150 (m/s)^2] * [effec. Shear / 12 m/s)] * [(2 km - mlLCL height) / 1 km] w/ the effec. Shear term capped at 1.5 and the LCL term capped at 1.0.
Values of 1 or more, combined with SCPs of 1-2 or more indicate tornadoes likely. If the values are 0.5 & 1, weak tornadoes are somewhat likely. EHI is probably more useful for determining if squall line tornadoes are likely.
Conclusion
By looking at soundings, hodographs & their indices combined with an analysis of the spatial guidance (model "error"), we get a pretty clear picture of what is likely to happen if severe weather forms. In the case above - Philadelphia on March 1st, 2017 - not much happened in PA. Was it the 6.4°C/km lapse rate? Was it the AM cloud cover and resultant -200 J/kg of CIN at noon? Most will cite the CIN. Sometimes a storm will form and it can't make a tornado despite all the basic requirements confirmed by model consensus (a powerful confirmation I might add). There's a lot that goes into a severe weather forecast that we don't know. And continued research is needed to fill in the blanks.
My next entry - the seasonal severe weather outlook - will be in approximately 10-20 days. Have a great week.
