Category: Daily Summaries

Daily summary of operations–includes information on which projects ran operations and problems that may have occurred.

Overview: Week 3, Day 4

Posted in Daily Summaries, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

The final operations day of the testbed began with our last daily debrief session, and we talked a bit about the applications sounding information with respect to the future GeoXO mission. The forecast for today was…meh? We had storms in the forecast, but the environment during our operations was marginal for severe weather overall. One office was localized to Billings, MT (BYZ) to capture developing thunderstorms closer to the jet streak, while a second office was localized to Topeka, KS (TOP) close to a remnant MCV that was kicking off convection already by 16 Z. Both offices got two mock-DSS events again today. If thunderstorms were looking unlikely to form in the Billings CWA by ~20Z we planned to move them to the Miami, FL office. Thank you Florida for being our safety net for thunderstorms.

Two MDs were issued by SPC during the afternoon for our forecast areas, with one resulting in a severe thunderstorm watch.

 

 

With Jonny’s help I made a ‘test RGB’ from the Synthetic GXI imagery. There’s no physical basis to the RGB (yet!), but it was a proof of concept that we could do it. An example and the recipe is shown below. What features can we pull from this new imagery relative to water vapor at low levels and through the entire column? Maybe we can figure it out in future testbeds!

 

 

During operations, I showed forecasters another example of SZA imagery over California with the impacts of the marine layer with low clouds and fog. Forecasters also compared this imagery against the Nighttime Microphysics to see how long the ‘gap’ was between the two, and what features you could identify with each product. Additionally I found a case from a fire from sunrise through the lens of the Day Fire RGB.

 

Kevin

Tags: None

Overview: Week 3, Day 3

Posted in Daily Summaries, Lightning Stoplight, LightningCast, OCTANE Speed Sandwich, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

Day 3 started with our debrief from Tuesday, along with a focus group activity for GeoXO lightning mapper capabilities. I showed forecasters SZA imagery near sunrise during a period of glare from the GOES-East perspective near sunset. Overall forecasters felt the cloud tops were more ‘washed out’ by the brighter visible channels, but mentioned the important feature from the Day Cloud Phase Distinction is often the cloud phase before the overspreading anvil cloud begins.

Traditional Day Cloud Phase Distinction RGB

SZA Day Cloud Phase Distinction RGB

Wednesday targeted the Northern Plains again, with the cold front being are only source of lift and deep layer shear that we can pull from this week. We decided to localize the forecasters to NWS Grand Forks, ND (FGF) and Aberdeen, SD (ABR), and each had two DSS events so they could leverage the LightningCast SuperDashboards.

1PM

Storms were slow to develop early in the forecaster period with a bit of cloud cover making the Synthetic GXI imagery less usable, so the product developers took both offices on a ‘tour of Europe’ showing off the Water Vapor Transmittance product from the Meteosat-12 FCI. This led us back to the states where we talked more about colormaps for WVT and what forecasters preferred. I made a display as part of that discussion. Upper left is WVT with ‘flipped’ colormap, upper right is WVT with the origional colormap, lower left is WVT with a red-green colormap made by a forecaster in the previous week, and the lower right is PWATs from the most recent HRRR run (18Z).

2 PM

Forecasters in both offices focused mostly on DSS tasks early in the forecast period, along with comparing OCTANE’s CONUS and MESO products. Additional discussion centered around the Lightning Stoplight tool and debating the ‘optimal’ color table to convey the intended actions from those viewing the product.
As thunderstorms approached the mock-IDSS events forecasters interrogated LightningCast data and discussed its applications, along with the desire to have variable ranges for the dashboard web display. We did run into an interesting case where a thunderstorm initiated over a narrow band of cirrus clouds, which may have impacted signals from OCTANE and LightningCast.

To end the day, I spammed AWIPS looking for sunrise imagery to show off SZA (thanks Justin for turning on the feed so early!)

Fog and low clouds from the marine layer along the California coast from the Day Cloud Phase Distinction RGB (SZA left/traditional right)

Valley fog over the Appalachian Mountains in the morning from the Day Snow Fog RGB (SZA left/traditional right).

Kevin
Tags: None

Overview: Week 3, Day 2

Posted in Daily Summaries, LightningCast, SPG2026, SZA Imagery on by Kevin Thiel.

10 AM

Tuesday began with our morning discussion from Monday’s operation/orientation day. Forecasters had plenty of storms to observe yesterday, which allowed everyone to get acquainted with OCTANE. Forecasters talked about the applications of OCTANE data, and one said it can be a ‘super tool’ for mesoanalysis but also needs ‘super users’. For the Synthetic GXI WVT product, forecasters also showed early support for the ‘flipped’ color table but wanted to see it in action more this week before deciding. First impressions of the Lightning Stoplight and SZA Imagery were also positive, with forecasters finding the Stoplight tool easy to understand and message to partners.

I showed an example of some SZA imagery on CIRA-SLIDER from this morning of river valley fog in the Northeast US. When comparing with traditional imagery and the Nighttime Microphysics RGB, forecasters reacted positively to the ‘brighter’ SZA imagery and expressed the difficulties of transitioning from day to night when observing fog from their respective home CWAs.

SZA Day Cloud Phase Distinction RGB

 

Regular Day Cloud Phase Distinction RGB

Nighttime Microphysics RGB

After talking about the GeoXO GXI in our focus group, our forecast discussion for the day sent us back to the Cheyenne, WY (CYS) and Rapid City, SD (UNR) NWS offices to monitor developing storms along a cold front. Forecasters were also assigned mock-IDSS events in Fort Laramie, WY and Sturgis, SD.

1 PM

Similar to Monday, shortly after operations began SPC issued an MD for our area followed by a Severe Thunderstorm Watch.

 

In the CYS office (still without a radar), discussion focused on their mock-IDSS event and the timing of lightning activity within the event’s range ring. With thunderstorms developing upstream of the site, timing dominated the conversation and the forecasters talked about how the LightningCast dashboard helped identify trends, especially if the partner needed more time to execute their safety plans.

The forecasters also looked at OCTANE, and noted that the the MESO Cloud-Top Cooling products showed a decreasing intensity trend while the CONUS CTC product appeared relatively steady.

3PM

The UNR office frequently viewed OCTANE products today, with several severe thunderstorm warnings issued today. Their mock-IDSS site appeared to ‘miss’ the storms today, but we’ll see what they say tomorrow!

I plan to show this to the forecasters tomorrow to get their thoughts. Sorry for spamming the blog with SZA imagery.

-Kevin

Tags: None

Overview: Week 3, Day 1

Posted in Daily Summaries, LightningCast, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

Week 3 kicks off today and the ridge that stymied us in week two is….still there! Thankfully we’ve recovered a good bit of moisture across the plains and southeast for convection to work with. We began today with product introductions and an orientation session, followed by our forecast discussion to decide where we would target today. The forecast was pretty clean cut, with two options of elevated convection in NE Colorado/SE Wyoming and additional convection from a remnant MCS across Mississippi/Arkansas. Based on timing of CI, instability in the high plains, and the size of the MesoAnywhere domain, we decided to localized to the NWS offices of Cheyenne, WY (CYS) and Boulder, CO (BOU).

 

1 PM
SPC issued an MD just before operations began for our area, citing thunderstorm development and a high chance for a watch to be issued later. And by 2pm we had a watch!

In the CYS office, forecasters looked at synthetic GXI imagery and identifying the moisture advection northward into their CWA later in the forecast period. This also led to the discovery that a few of the procedures had ‘inverted’ color maps for WVT. While not intended, we did compare how they looked from each color map. Which is better? You decide!

Original colormap. Dry air/clouds are white, moisture is dark.

‘Flipped’ color map. Dry air/clouds are black, moisture is light.

First impressions of SZA from this morning’s imagery was positive with forecasters in both the CYS and BOU offices. A forecaster with experience in AK said they regularly adjust brightness of visible imagery on the fly, especially in lower light seasons. A forecaster with more experience in the central plains said they rarely adjust the brightness. Some of the areas we examined included over Colorado and Montana. Forecasters mentioned using the DCPD RGB frequently, and having more time with that imagery would be welcome. Additionally, one mentioned liking that the colors stayed more consistent throughout the day.

Forecasters in the CYS office compared LightningCast v1 and v2 from GOES-East and GOES-West along with the timing of the first lightning flash. V2 introduced more heterogenety in the bands, while V1 appeared fairly uniofrm and more widespread. Also LightningCast v2 seemed to be ahead of v1 for thunderstorm growth over the Denver metro.

3 PM
Both offices issued severe thunderstorm warnings for severe wind and hail hazards. Looking ahead, we have a steady stream of SLGTs this week, so we should stay fairly active. Also wanted to noted the CYS radar was out today, so some questions about using Sat data (especially things like OCTANE) for warnings are of interest for the morning discussion tomorrow.
-Kevin
Tags: None

Overview: Week 2, Day 4

Posted in Daily Summaries, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

10AM

Thursday began with another group discussion, focusing on the following topics:

  • When to give the ‘all clear’ using the Lightning Stoplight tool
  • The limitations of satellite imagery and OCTANE-CONUS products on days with lots of convective anvil debris overtop developing convection, obscuring signals of stronger updrafts.
  • Playing with the color bars for the WVT product and finding what ‘works’ for showing different moisture boundaries like dry lines and sea breezes.
We played another round of ‘When does the satellite imagery become useful/useless?’ with the SZA imagery. This time we viewed a sunrise over the Mid-Atlantic with the 0.64 µm (ABI Channel 2) visible band, and forecasters stated when the imagery became useful. Here’s the results…
  • SZA Imagery: 0956 Z (x3), 1006 Z
  • Traditional Imagery: 1026 Z (x4)
SZA Imagery

Traditional Imagery

After a full week of MRGL and SLGT events, we were rewarded with…another SLGT! Surface moisture and instability was difficult to come by over nearly the full CONUS, and we relied on moisture transport from the Pacific front and sub tropical jet. Oddly enough this was positioned behind the dry line, which led to some large inverted-V forecast soundings and downburst potential from elevated thunderstorms. The 1630Z SPC convective outlook had a Slight Risk in the Central and Southern Plains, driven by a 15% CIG1 risk in SW KS and the TX panhandle. We decided to localize to the Goodland, KS (GLD) and Dodge City, KS (DDC) NWS offices today (Note: Later GLD office moved to Amarillo).

1 PM

Early on forecasters sent DSS messages to their mock events and waited for convection to develop. In the GLD office some convection tried to form in far-eastern CO, but dissipated and was unable to sustain itself. OCTANE-CONUS CTC showed low cooling values (<10C) and LightningCast was showing lower probabilities. Conversation moved to the Synthetic GXI imagery, and observing the moisture from the mid-level jet across the Southern and Central Plains. The group compared the 5.15µm Synthetic GXI band with the ABI 7.34 µm (Channel 10) IR band designed for ‘Low Level’ water vapor.

3PM

Decided to move the GLD office to Amarillo (AMA) after observing minimal instability in their CWA, and a more agitated cu field on the western edge of the AMA CWA. SPC also issued an MD and subsequent severe thunderstorm watch for the DDC and AMA forecast areas, which reinforced our thinking that convection would be focused further south during our forecast period.
4PM
After some discussion with the DDC forecasters and some developers, we made vis-IR sandwich procedure that mimicked the ‘oops all Day Cloud Phase Distinction RGB’ procedure. The motivation was that once convection gets going, forecaster may liked to transition away from the DCPD RGB and to imagery with more direct visible and infrared information.

5PM

At the very end of the day, we talked about the SZA imagery some more and its applications for OCONUS users in places like Alaska. While roaming the CONUS comparing SZA and traditional imagery, a developer pointed out a large blowing dust event in ND, which was evident even in the DCPD RGB, along with the Channel-2 visible imagery.

That’s a wrap for week 2!

-Kevin
Tags: None

Overview: Week 2, Day 3

Posted in Daily Summaries, OCTANE Speed Sandwich, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

10 AM

Wednesday presented a ‘7-10 split’ of severe potential, with thunderstorms expected in the Great Basin and Mid-Atlantic regions. Most of this was driven by a Slight Risk from the 13 Z SPC Convective outlook, driven by 15% CIG1 wind probabilities. Timing and moisture were a little less certain out west, while convection was more certain in the east but had limited deep instability and subsequent severe weather potential. After some discussion, we localized to the Pittsburgh, PA (PBZ) and Pocatello, ID (PIH) offices.

During the morning discussion we played the same comparison game between the SZA and traditional imagery, but this time for the upper great lakes region around sunset on Tuesday, using archived imagery from CIRA-SLIDER. Forecasters were asked when the imagery no longer became useful to them, and here’s their reported times when looking at the same loop between 21 Z and 02 Z.

  • Traditional: 2156 Z, 0001 Z, 0001 Z, 0021 Z
  • SZA: 0111 Z, 0111 Z, 0111 Z, 0111 Z

Here’s the loops so you can play along at home! (Sorry these are really big GIFs, the blog didn’t like the when I converted them to MP4 format)

Traditional Day Cloud Phase Distinction RGB

SZZ Day Cloud Phase Distinction RGB

1 PM

Forecasters in the PBZ office were thrown into a situation where convection was already developing in the western side of their CWA, and they began issuing warnings, investigating OCTANE, and sending DSS messages to their mock-events.

In the PIH office storms were slow to get going, so we looked at some imagery from the FCI and the synthetic GXI imagery from the HRRR. One target we casually watched throughout the day was a dry line in the TX panhandle, and we compared them to moisture boundaries in Europe and north Africa.

3 PM

PIH was issuing warnings by this time, and frequently relied on radar imagery and knowledge of the environment when making warning decisions. When looking at the OCTANE CTC product and other satellite imagery, the presentation of most storms were not impressive.
In the PBZ office, the forecasters noted that the OCTANE CTC product was limited in its application due to dense cirrus and anvil debris obscuring cooling signals in the ABI infrared bands. In the OCTANE Speed Sandwich however, the forecasters found the speed values and textures from visible imagery easier to view new and intensifying updrafts. This resulted in some discussion about how to ‘train’ for this.

By the end of the day, a number of severe wind reports appeared across the PIH CWA from their mesonet, while in PBZ a single hail report was featured. More reports may come in tonight or tomorrow?

-Kevin

Tags: None

Overview: Week 2, Day 2

Posted in Daily Summaries, OCTANE Speed Sandwich, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

10 AM

Tuesday began with a group discussion and first impressions for the five experimental products. Topics included how forecasters applied LightningCast and the Lightning Stoplight, OCTANE Cloud-Top Cooling for monitoring updraft strength, and the value of SZA imagery near sunrise to observe low clouds and fog. I showed an example of the SZA Day Cloud Phase Distinction RGB on CIRA-SLIDER, with clouds and storms across Florida around sunrise. I asked the forecasters to identify what time the imagery became ‘usable’ for them, and compared it against the traditional imagery of the same scene. Here were the results:

  • Traditional imagery: 1131 Z, 1141 Z, 1151 Z, 1206 Z
  • SZA imagery: 1051 Z (x3), 1101 Z

I’m hoping to run a similar experiment with SZA imagery near sunset tomorrow. Here’s the imagery I showed today. When do you think each imagery becomes ‘usable’?

Traditional Day Cloud Phase Distinction RGB

SZA Day Cloud Phase Distinction RGB

We jumped into our forecast discussion, which asked the question ‘Can we avoid going back to Florida again?’ Unfortunately that answer was NO, so we localized to NWS Melbourne, FL and NWS Miami, FL.

 

 

1 PM

Shortly after ops started SPC issued an MD for the Florida peninsula.

Forecasters in the MFL office talked about providing DSS with LightningCast and Lightning Stoplight, and noticed in the OCTANE MesoAnywhere product how cirrus motions aloft cloud interfere with low cloud motions. MesoAnywhere is on the left, ABI MESO imagery is on the right.

Forecasters in the MLB office issued a handful of warnings, and we had a discussion about the OCTANE-CONUS Cloud-Top Cooling product. While the OCTANE MESO products showed more information, forecasters still found the OCTANE-CONUS products useful if a MESO wasn’t availalble in this case.

Late in the day we viewed the frontal boundary in KS where storms might initiate this evening through the Synthetic GXI products, namely the 5.15 µm and WVT products. We compared them with the ABI split window moisture field (10.3-12.2 bands) and the ALPW from JPSS. In talking with another developer, we got the idea to create a ‘sandwich’ product that meshes the WVT and ABI Split Window products.

-Kevin

Tags: None

Overview: Week 2, Day 1

Posted in Daily Summaries, Lightning Stoplight, LightningCast, OCTANE Speed Sandwich, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

10 AM

Week 2 of the Satellite Experiment kicked off this morning, with a fresh batch of forecasters and a recharged group of developers. Convection was difficult to find unless you were along the Gulf or Atlantic coasts, and with marginal deep layer shear the potential for severe weather was also also categorized as Marginal from SPC in its 13Z outlook. After our orientation session, the forecast discussion today was focused on western Louisiana, Florida, and the Carolinas. Ultimately we decided to localize to the NWS offices in Jacksonville, FL (JAX) and Melbourne, FL (MLB) with the expectation that upstream convection and land-sea breeze interactions could initiate and sustain convection throughout our operations window (18-22 Z).

1 PM
Operations kicked off with the developers providing hands-on demonstrations and walk-throughs with the forecasters. With a slower severe weather pattern on the first day, we encouraged everyone to think of today as a ‘training day’. Forecasters asked several questions about the synthetic GXI imagery, such as the heights sampled by the 0.91 and 5.15 µm bands, the impacts of snow cover, and how the synthetic products compare to the real 0.91 µm imagery from FCI. We also talked about making RGB products from the synthetic imagery, and realized that the imagery specialists at CIRA had taken the liberty of remapping the products to the ABI 2km grids (kudos!). I’m hoping to create some RGBs this week that we can play with and improve on. What would forecasters like to see in an RGB? Moisture by height? Low level boundaries?
2 PM

We got an MD from SPC! Not a great chance for a watch, but good to know some severe risk exists in our CWAs.

4 PM
The rest of the operations day was pretty slow, so I spent some time working (arguing?) with AWIPS to make an RGB for the Synthetic GXI data. It’s a work in progress…

Found an interesting case of dissipating and initiating convection form the perspective of LightningCast v2 and Lightning Stoplight v2 further west near the Alabama coast. You can watch the probabilities drop and the stoplight transitions from red to green for the dissipating storm, while probabilities increase shortly before the first lightning flashes. Lots of anvil debris to obscure the signal of initiating convection in this example, but there appeared to be at least some lead time by LightningCast before the first lightning flashes appeared from the Lightning Stoplight.

A nice thunderstorm went up east of Jacksonville, with a notable cooling signature from the OCTANE-CONUS Cloud-Top Cooling product. I compared that with the OCTANE-MESO product, and may encourage forecasters to do the same later this week.

To wrap up, I decided to explore all the flavors of the Day Cloud Phase Distinction RGB with a procedure I called ‘oops-all-dcpd’.

  • Upper Left: GOES-19 CONUS
  • Upper Right: OCTANE MesoAnywhere (from GOES-19 CONUS)
  • Lower Left: SZA (from GOES-19 CONUS)
  • Lower Right: GOES-19 MESO
Kevin
Tags: None

Overview: Week 1, Day 4

Posted in Daily Summaries, Lightning Stoplight, LightningCast, OCTANE Speed Sandwich, SPG2026, Synthetic GXI, SZA Imagery on by Kevin Thiel.

10 AM

Day 4 kicked off with another debrief session, with considerable focus on LightningCast and Lightning Stoplight for decision support. Near the end of the operations period yesterday we got the Lightning Stoplight tool into the HWT’s AWIPS-II instances (thanks to Justin, Jonny, Roger, Kris, and the rest of the Stoplight team for the hard work to get that done!). We talked about what products in AWIPS-II you could display with it, and some forecasters preferred to have the lightning information (GLM, ENTLN) while others didn’t. I’m hoping another day with DSS will better inform forecasters of these products, and using OCTANE and Synthetic GXI imagery in marginal/messy environments.

Picking our WFO locations was a bit easier today, but it was unfortunately from how limited the areas for convection were. A cold front had washed out much of the moisture and instability down to the Gulf Coast, leaving us with marginal convection from the 13 Z SPC outlook. You know it’s a meager setup when we had to request an ABI MESO domain for our area, and we were glad the SAB approved. We settled for Lake Charles, LA (LCH) and returned to San Antonio (EWX) in hopes of getting some semi-discrete convection in each. Both offices were again given two DSS events and encouraged to use the LightningCast Super Dashboards.

1PM

Operations kicked off and forecasters spent a good bit of time playing with LightningCast and the Lightning Stoplight tool together, and providing feedback to both sets of developers. Working with forecasters and developers, we made a few ‘flavors’ of the Stoplight tool in AWIPS by modifying the colorbar. Here’s the descriptions from the animation below:

  • Top left is default
  • Top right colors the flashes in the last minute as dark red
  • Bottom left changes the green to blue
  • Bottom right colors the flashes in the last minute as pink followed by a red, orange, yellow scheme

Forecasters had a positive response to the top right panel, and said that coloring the 0-1 minute flashes from GLM and ENTLN dark red in some ways removed the need to overlay those products in AWIPS, simplifying their display. I saved this display off as a procedure and hope to show future weeks and get their feedback too.

3 PM

The Synthetic GXI folks decided to show the forecasters WVT data from the METEOSAT-12 FCI, since it has the 0.91 µm band that GXI plans to add. Features such as fronts and low level moisture boundaries were identified and discussed.

4PM

We found an interesting case of CI near the Lake Charles, LA CWA, with anvil debris from upstream convection obscuring the convection from visible and infrared satellite imagery. The OCTANE cloud top cooling product had a hard time showing cooling until the updraft produced its first lightning. LightningCast v2 and v1 were also compared for their performance in this challenging scenario.

5PM
To end the day I went back and made another animation of the SZA product. This time I got sunrise over the CONUS (thanks Justin!) and showed some early morning convection across central TX, with a different cloud types and layers.

Even in the MRGL, we still had a full day!

-Kevin

Tags: None

Overview: Week 1, Day 3

Posted in Daily Summaries, Lightning Stoplight, LightningCast, SPG2026, SZA Imagery on by Kevin Thiel.

10AM

Day three kicked off with a loaded discussion, with OCTANE, Synthetic GXI Imagery, and the Lightning Stoplight Tool taking center stage. One interesting topic came from the presentation of Lightning Stoplight, and how to interpret regions with no data next to pixels in the 0-10 minute range (red). Would a site manager interpret that area as ‘safe’? This also led into a discussion on how forecasters would present these data to partners, with most in the group stating that they wouldn’t feel comfortable giving the Lightning Stoplight or LightningCast tools to EMs without considerable on-site guidance or training.

The forecast discussion was split four ways between Texas, Mississippi, West Virginia, and Colorado. We ended up going to the San Antonio, TX (EWX) office for a supercell and large hail threat, and the Charleston, WV (RLX) office for widely scattered convection with a marginal wind and hail threat. Forecast groups were also now tasked with two mock-DSS events, forcing them to monitor multiple events on the LightningCast superdashboard product. The Lightning Stoplight tool was not in AWIPS until 21 Z, but forecasters were advised to use the web display from NASA SPoRT.

1PM

I started the day by going west to Hawaii to collect SZA imagery at sunrise from GOES-West and sharing these with the forecasters. How much more ‘usable’ information can SZA provide compared the traditional imagery, and how much sooner if possible?

4 PM

After several discussion regarding DSS, LightningCast, and how to interpret the Synthetic GeoXO Imagery, we were able to get the Lightning Stoplight product into our AWIPS-II! Forecasters stated in the morning discussions they want to immediately union these data with LightningCast, and some procedures were made to help them out at the beginning of ops tomorrow. I created an animation of the Lightning Stoplight that includes GLM, ENTLN flashes, and ABI Clean-IR imagery. I set the opacity of the Lightning Stoplight to 80% so the colors stand out, but you can still see some detail underneath. We’ll see how well the forecasters like this information and if they make any new displays. This animation shows the dissipation of a thunderstorm on the western side of the scene, and you can see the loss of GLM FED data, while the Lightning Stoplight product changes colors and shrinks in size to indicate a decreasing lightning hazard.


The day ended with a quick look at sea-breeze thunderstorms in eastern Cuba for the SZA imagery. Its interesting to see the low clouds and how long they stay present in the SZA’s modified DCPD RGB. We’ll see how the forecasters like this in the morning.

 

Kevin

Tags: None