PHS Blog Post AMA 6/13

PHS today was interesting given some of the local convective parameters. The main one that caught my attention was the lesser 0-6km bulk shear that it was giving just ahead of an incoming storm in the Oklahoma Panhandle. The PHS was around 20-30 knots whereas other guidance and SPC Mesoanalysis was showing 50+ knots. I did look at other parameters such as MUCAPE and SBCAPE for example and they were in the ballpark of other guidance including SPC Mesoanalysis.  If the PHS was correct then the storm mode would feature more multicell with some organization vs more of a discrete supercell mode so it was going to be interesting to see what would occur. 

What wound up occurring was a persistent long track supercell along with other convection which was anticipated given the PHS wind shear.  Overall, I was impressed with the PHS wind shear forecast as that tailored away any thinking that there may be additional supercells as additional updrafts would interfere with the intensification of other cells. The PHS however was around 1-2 hours slow with convection moving into the AMA CWA, however the composite reflectivity forecast did show more of a cluster of storms developing. The PHS was also showing a corridor of stronger 0-3 and 0-1 SRH which in reality the intense cell did move through and actually wound up producing a few tornadoes. The ProbTor was also showing tornado potential as high as 40% as it was moving through this corridor.  Overall, I was impressed with the PHS today and think that it could create good DSS messaging opportunities along with being skillful in Mesoanalysis.

PHS showing the lesser values of 0-6 bulk shear in contoured form

SPC Mesoanalysis depicting the 0-6km bulk shear at the same time

How the radar wound up turning out at 2130Z with the multiple cells

– Tor Nader  

DFW and TAE HWT Feedback for 6/13

General question overall, can the units be displayed in fraction form instead of with a (-1 or2)?  Such as m/s instead of m*s-1?


Can temperature units be in F or even C instead of K?

Is 0-1 KM bulk shear AGL?

Could the bulk shear be in knots instead of m/s?

Can the bulk shear vectors and mean cloud wind vector (or 850-300mb wind) also be plotted?  This can help with determining storm mode. 

The PHS had storm coverage well matched with radar when simulation started.  However it was about 3 rows of counties too far north than what actually was happening.  This led me to believe that all the other data with PHS was also shifted north for what was actually happening.

Could mixed layer CAPE/CINH be added to the stability menu?  I think SPC’s mesoanalysis page defaults mixed layer to 100mb.  We use mixed layer CAPE quite frequently at our office.

Having DCAPE as a viewable product would be very helpful to diagnose a severe wind threat from supercells.

Noticed the PHS surfaced based CAPE/CINH matched quite well with mesoanalysis except for the CAPE/CINH just north of FL.  Both images are from 21z.


Helped maintain awareness about the overall threat for lightning, especially on the outskirts of the storm.


Noticed the ML CAPE was much further east than what happened.  Of course this was from the morning overpass, not the afternoon one.

Did notice that the ML CAPE was lower than the surface based and MU CAPE.  Was going to look at a NUCAPS sounding to determine why that may have been the case.  However was not able to do that due to no data aside from the morning run. ML CAPE is on the left, MU CAPE is on the right below.

Prob Severe

Is there a way to highlight which storm Prob Severe is showing the trends for?  When I am looping the radar with multiple storms on it (example image below), I quickly lose track of which storm I had the trend graph displayed for.  Really like the ability to look at the trends graphically through AWIPS instead of having to go to the webpage to do so.

Noticed the readout only displayed above the pointer, not below.  The right image is what happened when I tried to sample the Prob Severe data.

Could there be an option to select the different fields we want Prob Severe/Hail/Tor/Wind to show?  There is a lot of data there, good data, but at the same time too much to look at during storm interrogation.

Does ProbTor calculate QLCS tornado probability?  This would help build confidence for issuing these kind of tornado warnings because the environment can change so rapidly from one radar scan to the next.


This helped me really see the updraft divergence of the cells.  This was in deep layer shear environment of 20 kts.  Neat to see the divergence weakening on the middle bottom cell in the upper left pane weaken as the cloud top temperatures warmed for that cell, confirming that the updraft was weakening.

– Rainman

6/13/23 EWP Blog Post – SHV

Active CWA assignment today. Warning issuance followed by the log form entry was somewhat cumbersome and time consuming but I understand the reasoning. 

Great interaction with the subject matter experts and their insight and comments were much appreciated. 

Good discussion on the potential uses of the OCTANE data. My CWA partner mentioned constructing real time hodographs that could be sampled. This is a good idea. The mental exercise to visualize the wind profiles from the height / direction/ and magnitude data is taxing to me and I think would be too time consuming in operations. 

Developer stepped through the awips imaging method to highlight specific layer winds and I found this very helpful. A layered wind four panel procedure would be helpful. Also good discussion on reaching for storm scale signals in the wind fields and whether wind speed data bins could be higher resolution than radar at high elevations and distance from radar. An intense left moving storm during this exercise window across NW LA into far SE OK showed strong anvil level divergence that persisted for an extended period.

Screen capture of the ice machine and intense three body scatter spike that persisted for close to an hour at least. Probsevere was very good especially once the tracking element was assigned to just the primary updraft. I issued several warnings during the exercise window and utilized probsevere on each one. The time trend window is great and I hope it gets to the operational version soon.

The PHS idea of better utilizing sampled profiles seems good to me. I would prefer it not be solely validated by WRF model output. I like the idea of showing where the fusion data senses departure from model initial conditions similar to shown on  this website:

NUCAPS data today  was from the previous run with the 10-12 hour forecast near the exercise valid time. I compared the 11 hour NUCAPS forecast to the zero hour RAP analysis for MUCIN across the exercise domain. There was a large discrepancy between the two. Due to convection that had spread across this region earlier in the day I discounted the NUCAPS data. In this instance the NUCAPS forecast would have been viewed as suspect and likely would not have been used.

– jbm

Assessing Potential for Additional Severe Convection in ABQ June 12, 2023

Our group role played as ABQ on June 12, 2023, where severe convection was already underway in the northern third of the CWA during the early to mid afternoon. Rather than interrogating ongoing storm strength and hazards, we were curious as to whether or not additional severe convection was possible south of the current convection. In essence, we were performing tasks of a mesoanalyst.

The synoptic environment was composed of upper troughing over the Great Basin region, with a stout, unseasonably strong mid to upper level jet at its base overspreading New Mexico, and a shortwave trough pushing out of the Four Corners into southeastern Colorado as well as northern New Mexico. GOES-16 mid level water vapor imagery showed the location of the bulk of synoptic forcing through steadily cooling brightness temperatures (not associated with deep convection). Overlaying NAM 500mb geopotential heights helped illustrate where strongest height falls associated with the shortwave trough were located. In this case, strongest height falls/cooling brightness temperatures were located within southern Colorado into far northern New Mexico, and was driving the ongoing severe convection in northern ABQ. However looking south and upstream within central New Mexico, there were still cooling brightness temperatures within water vapor imagery, just not as robust relative to the north. This implied there was still some synoptic forcing for ascent to overspread areas like San Miguel, Guadalupe and De Baca counties, the counties we were most interested in for additional initiation.

Another display we referenced to assess low level moisture content was GOES “Clean LWIR 10.3um – Dirty LWIR 12.3um” Split Window Difference sandwiched with visible imagery to show current areas of Cu development/maintenance. This showed low level moisture pooling up nicely along higher elevation of the Sangre de Christo Mountains southward to the higher elevations along the western flanks of the southern High Plains, along with agitated Cu at the terminus of these moisture gradients. Additionally, local areas where there were easterly pushes of the moisture gradients suggested dryline dynamics were at play, potentially providing additional forcing for ascent.

After assessing the current state of the environment, we referred to PHS Layer Composite Reflectivity, which is simply just simulated reflectivity offered by the WRF that runs within PHS environment that incorporates latest hyperspectral sounder information on thermodynamics via polar orbiting satellites while merging it with the motion and high spatiotemporal resolution of GOES ABI . The 18Z model run indeed sparked additional convection off of the higher terrain within Guadalupe County.

So in the eyes of this model, the potential was still there. It was also nice to compare current radar (in this case MRMS Composite Reflectivity) in the same display as PHS reflectivity to see how the model was performing in this aspect thus far.

But let’s not take this for face value as this is modeled convection. We desired other datasets to help provide insights on the thermodynamic environment that may play a role in allowing additional convection, or lack thereof. With this, we referenced other datasets within PHS:

This display shows MRMS Composite Reflectivity and surface observations overlaid with PHS thermodynamic parameters in the top and right halves, and shear parameters in the lower left. Focusing on the two right panels, PHS suggested moderate instability (~1500 – 2500 J/kg) lee of the southern Sangre de Christo Mountains into the southern High Plains of northeastern New Mexico, with perhaps slightly lower instability within Guadalupe and De Baca counties. More interestingly was the stark increase in LCL heights within these counties compared to the environment just to its north. This raised a yellow flag, leading us to to think thermodynamics may not be as favorable as the area with current convection. Perhaps relatively drier air may be located in this area of interest compared to that currently driving convection? We referred to other satellite-derived datasets like NUCAPS soundings, Gridded NUCAPS and NUCAPS-Forecast (not shown) to assess upstream mid level moisture (namely 700 mb RH) advecting into the area, of which reveals a well developed and strong EML with notably strong inversion nested around 700 mb. We also referenced PHS CIN (not shown) which exposed a relatively higher area of capping in this same area where LCLs were higher.

It was nice having the high spatial and temporal resolution offered by PHS when performing mesoanalysis, and in my opinion had an advantage over objective analysis (e.g. SPC Mesoanalysis), with its higher resolution and ability to step forward in time on an hourly basis.

Here is the full loop of this PHS display between 18 UTC to 23 UTC:

Within our area of interest, the considerations of relatively higher CIN, strong EML advecting over this area, relatively weaker forcing compared to points north, we had some concerns that sustained convection would develop, but we still didn’t discount the possibility should enough subtle forcing via synoptic (DCVA/height falls) and mesoscale (dryline) lift contribute together to overcome locally higher capping likely in place. Ultimately, we gave around a 20-30% chance of additional severe convection within Guadalupe and De Baca counties in the next 1-3 hours starting at 21Z.

Just before today’s HWT window closed, OCTANE hinted at attempts of deep convection within central San Miguel County via the warming colors collocated with agitated Cu turned orphaned anvils behind exiting deep convection. This was advantageous as it caught our eye perhaps faster than Day Cloud Phase Distinction RGB indicated for attempts at deep convection.

Ultimately, deep convection did not occur within Guadalupe and De Baca counties.

– OSMBLSN     

Comparing PHS Data with SPC Mesoscale Analysis and MRMS Data

The first day was slightly overwhelming with all the different products to test out this week. I decided to focus on one of the products I was unfamiliar with and that was PHS data. I compared PHS SBCAPE/MUCAPE values from the 12.14z model run to the latest SPC Mesoscale Analysis data. At first glance, the PHS SBCAPE/MUCAPE values depicted were generally higher over southern Colorado and northern New Mexico than the SPC data. However, I did notice a small bullseye of CAPE minimums in south-central Colorado and was curious about why that was depicted (See Figure 1). The SPC data was not as detailed with the gradients and had no mention of the minimum in south-central Colorado (See Figure 2). Thankfully, one of the developers was in the room and mentioned that the model might be depicting storms in that location. I loaded up the composite reflectivity from PHS (See Figure 3) and sure enough, PHS was suggesting storms in south-central Colorado.

Figure 1: Top image shows PHS SBCAPE at 20z with a minimum in south-central Colorado.

Figure 2: The bottom image is the SPC Mesoscale Analysis at 20z.

The composite reflectivity had the right idea with convection near that location compared to MRMS data. It was slightly further north with the maximum dBz values, but for an 8-9 hour forecast, it depicted the convection fairly well IMO.  The PHS did much better with the convection over the northern portions of PUB CWA as it showed a decent line of storms at 22z which matched up well with MRMS.  For obvious reasons, this model could certainly be helpful in providing higher confidence in convective storms for various NWS partners and DSS events.

Figure 3: top loop showing the PHS layer composite reflectivity.  This is the 12.14z model initialization.

Figure 4: MRMS loop of the composite reflectivity for the hour of 20:00z to 21:00z.

– Podium

Day 1 Review of Products & Operational Applications


  • Distinguish different air masses when they are characterized by different advection regimes
    • Stable cloud streets in magenta (Figure 1)
    • Agitated Cu and deep, moist convection in yellow/green (Figure 1)
  • Locate boundaries
    • Identify areas of convergence based on the depicted flow directions
  • Suggestion: it would be great to be able to sample the cardinal direction (i.e. NW, SE) in AWIPS in addition to the direction already given in degrees

Figure 1


  • Attempts at convective initiation failed over Guadalupe County and De Baca County… but why? Isobaric analysis revealed supportive dynamics for ascent south of the already-occurring convection.
    • 700 mb relative humidity and 850 mb humidity data (Figure 2) reveal a sharp moisture gradient across northeastern New Mexico
      • SPC 850 mb Td mesoanalysis verifies this (Figure 3)
    • Dry air was already moving across both counties from the west and was progged to continue over the next few hours, thus hindering convective initiation

Figure 2

Figure 3

– Vort Max

6/12 Observations from ABQ CWA


Have knots abbreviated as “KTS” instead of “KN”

Change direction of movement from to lettered degrees instead of a numerical value

This product does help me become more aware, in realtime of the winds vertically.  However, not entirely sure how to adapt this into my forecasting routine yet.


Great to see the better sampling/readout of the data compared to what it was 2-3 years ago

The parcel count is a great way to help me gauge the validity of the data (confidence in how accurate it is)

Overlaying a NUCAPS forecast field with the parcel count allows me to sample what the field has and what the parcel count is so I can have more of less confidence in the data

No less than six hours for operational needs.  9 hours might be the perfect duration.  I usually don’t use hourly data much beyond 9 hours due to my lack of confidence in the reliability of the data.

Hourly data is very nice.  Half hour spacing is a bit too much data for that duration.
-Something to consider is could there be half hour spacing in the data for the first three hours?  However I don’t know if that would be enough of a benefit to use.

I would use the 700-500mb lapse rates, ML CAPE, temp and dew point, and parcel count most often.


Very easy to look at; very smooth data

Didn’t have much time to look at the data


Noticed on a well defined supercell the lightning probability is not over 80% despite lightning ongoing.  Does the program cap at 75-80%?
-Have GLM ligthning data displayed too

Otherwise this does help draw my eye to where lightning could occur in the near future.

Prob Severe

Didn’t have time to look at it much. Did notice the storm in eastern CWA had higher values for version 3 than for version 2

– Rainman

HGX Convection impacting the Car and Truck Show in Burton, TX

Initial setup for the HGX vicinity showed several cells to the north and east of the DSS site, propagating southeastward. The strongest cell, pictured below, had a PSv3 of 73%, while PSv2 remained as 56%.

We issued a warning for a northern cell moving into the CWA into Madison county, based on a -70C cloud top brightness temp and PSv3 total prob over 70% (had been climbing from the 50s fairly steadily). But the cloud top shear noted by Octane was not strong (~20-25 kts), so the warning was very borderline. Just a couple scans later, it lost most of its texture on the vis imagery and lost its shear in the Octane direction product. Cloud tops warmed a bit as well.

00H NUCAPS-Forecast (NF) is showing moderate CAPE now (1st image below), which may help explain the messy sub-severe multicell clusters, but the forecast valid at 02z this evening shows a resurgence in the CWA (2nd image below).

Looking to our NW, one of the stronger cells is outside our CWA, but the Octane direction is showing good cloud top diffluence.

The PHS SCP forecast valid at 20z looks to be around 2-4 over our area, although this doesn’t match well with the SPC meso page SCP, which focuses high values W of our CWA.

The 21z PHS MUCAPE (15z run) looks like it has insane values of 6000-7000 J/kg near the coast and just offshore. This is much higher than the SPC meso page, showing 3000-4000 J/kg at most.

Looking at the optimal application of LightningCast, it seems that the point-based meteogram would work best for CI and in situ developing convection, versus storms propagating into the area. In our case here, at the DSS site, the point-based LC probs are low, suggesting little concern. But we can see from the GLM FED data that there are mature cells with lightning just to the NE that will probably move near the site in the next hour, which certainly poses a safety concern.

The NUCAPS sounding near Victoria (far SW CWA) showed a lot of CAPE and DCAPE, but a rather dry profile. This is confirmed by WV imagery over much of far S TX.

22z PHS Composite Reflectivity (above image) compared to 22z MRMS Composite Reflectivity (below image ) depicting PHS struggling on timing as the cluster of thunderstorms propagate from northwest to southeast.

Late in the event, this cell is showing slow strengthening on PS — currently both v3 and v2 have 40% total.

– Edgar and Harvey Specter

MAF Convection

Convection has been slow to initiate in W TX, but we’re finally seeing some vertical extension and greater texture in the CU field on the NE side of the mountains. LightningCast is now showing higher probabilities in this area, after several scans of sub-50% contours.

We are seeing good destabilization. The PHS MUCAPE compares quite well to the SPC mesoanalysis.

LightningCast probabilities continue to climb, and we’re finally seeing a few flashes on GLM imagery.

Our LightningCast meteograms are finally showing some higher probs in the next hour near our DSS site in Fort Stockton.

We’re now seeing more defined features in the Octane imagery, including cloud top divergence signals and cooler brightness temps, down to -63C. But our ProbSevere markers remain low.

Looking at the NUCAPS-Forecast initialization at 19z, the mid level lapse rates are lower in areas NE of the higher terrain, where convection has been a bit slower to strengthen, as compared to the storms to our SW over Mexico, where MLLRs are higher.

The new TPW imagery from NUCAPS shows well the delineation of the moist air to the E and drier air to the W.

The PHS SHiP pattern looks good, but its magnitudes appear way off, compared to the SPC meso page and how SHiP is calculated.

Hail probs are exceeding 80% now on ProbSevere, although our GLM lightning is rather low! Interesting how assumed large hail presence doesn’t necessarily equate to classic charge separation that would prompt high electrification.

21z 700-500mb comparison of SPC Meso (top) vs 21z NUCAPS-Forecast (bottom)
    – Some differences in lapse rates between the two, with NUCAPS about .5-1°C/km  lower than SPC Meso page. 

ProbSevere, Octane, and IR imagery supports warning issuances for Reeves and Pecos counties.

With the warning now out for Reeves and Jeff Davis counties, this storm continues to exhibit high hail probabilities on PSv3.

Up to 120 kts of ST divergence with the warned cell (at about 50 kft).

Interesting how ProbSevere identifies objects. There appears to be 3 separate updrafts in Pecos county, however PS is seeing one object.

– Edgar and StormofCentury

Examination of LightningCast and OCTANE in Idaho and Oregon

Near the Idaho/Oregon border, we observed splitting cells on the afternoon of Wednesday, June 7. We believe we were able to see the first hints of this process get underway with the help of the OCTANE product. Subtle gradients were observed on the left side of the main storm prior to the ultimate cell merger. 

In the image on page 2, we see a nice comparison using LightningCast between an environment characterized by cirrus contamination vs an environment that was not contaminated. On the eastern side of the CWA that was characterized by cirrus contamination, LightningCast probabilities peaked at 50% on a storm that ultimately produced lightning. In contrast, two thunderstorms to the west (developing in a clear, cloud-free environment) both saw LightningCast values exceed 80% on two storms that ultimately produced lightning.

Above, we see perhaps the first instances of a splitting cell in progress, and OCTANE helped give us a first indication of the split (upper left in particular)

Above: LightningCast Meteogram as of 2053z, depicting a small trend upwards, albeit extremely minimal, within the next 60 minutes for some lightning impacting the American Aquarium at Hayden Beverage in Boise (DSS location).

Above: 21z PHS shows development of convection in Owyhee and Twin Falls counties (southern BOI CWA) depicted by a red box; however, the 4 Panel Situational Awareness is indicating that doesn’t seem to be the case, with mere cirrus cloud development. Perhaps the timing is off?

Above: On the other hand, PHS is doing a fairly good job depicting the convection just east of the DSS site. Currently (almost 22z), a few decent cells depicted by ProbSeverev3 (right around 15-18% probability, where MESH values are just over 0.50”)

Above: 2123z: LightningCast indicating just above 50% probability of lightning within the next hour yet the Meteogram is well below 20% for the GOES East reading. It is suspected that this is because the AWIPS readout is Plax-Corrected, whereas the Grafana Meteogram may not be?

Just after 22z, Grafana Meteogram (above) shows GOES East  ProbSevere Lightning to be 67%. This latency further aids in the belief that the Meteogram is not Plax-Corrected.

As of 2220z, downward trend displayed with regards to the Grafana Meteogram (topped at 70%, now showing 54%). This is indicative of the cells east of Boise (DSS area) initially showing signs of weakening…HOWEVER…

Things are starting to get juicy at 2224z with ProbSevere v3 showing the cell south of Idaho City has increased to 36% probability (below).

– Bolt and Harvey Specter