1. University of Oklahoma/CIMMS & NOAA/NWS/NCEP/SPC
NUCAPS Demonstration in the 2017 HWT Satellite Proving Ground Experiment
Michael Bowlan
University of Oklahoma/CIMMS & NOAA/NWS/NCEP/SPC
2018 AMS JPSS Short Course
January 6, 2018
Hello Everyone, I am Michael Bowlan, the satellite liaison to the Storm Prediction Center and Hazardous Weather Testbed in Norman, OK. I am going to be giving a presentation about some of the early results and feedback from the 2017 demonstration of NUCAPS in the Hazardous Weather Testbed Satellite Proving Ground Experiment. I will show a few cases where forecasters used NUCAPS in their operations and show some of the feedback and suggestions offered by the forecasters from the 4 weeks of the experiment.

2. HWT Experimental Warning Program (EWP)
Mission: Improve prediction of severe convective weather at the “warning scale” (0-2 hours).
Norman has a large community of researchers, operational meteorologists, students, industry.
But, we serve all National Weather Service WFOs and CWSUs nationwide.
A vital component to the Research To Operations (R2O) process.
First, a little bit of information about the HWT and the Experimental Warning Program or EWP for those of you that might be a little less familiar with the testbed. The mission of the EWP is to improve the prediction of severe convective weather at the quote warning scale (0-2 hours). We do this by simulating a real operational environment for the forecaster and demonstrate new and experimental products for the forecasters to use in their warning environment. Norman is unique in that a large community of researchers, operational meteorologists, students, and private industry are all located within if not the same building, in the same block of buildings. We aim to serve all of the NWS WFOs and CWSUs nationwide, Most of all, this testbed is a vital component in the R2O2R process where new products can be demonstrated in a real time system by real time forecasters which provide feedback to those developers on ways to improve the product for use by all eventually in NWS operations.

3. 2017 GOES-R/JPSS Experiment
4 weeks (19 June, 26 June, 10 July, 17 July)
3 NWS forecasters, 1 broadcast meteorologist per week
Using AWIPS-II
Forecasters issue experimental mesoscale forecast updates (via a blog) and severe thunderstorm and tornado warnings (WarnGen)
We want forecasters to think about how they are using the experimental products in nowcast and warning decision making.
This year’s GOES-R/JPSS Experiment consisted of 4 weeks from June 19 to the week of July 17. Each week was comprised of 3 NWS forecasters and 1 broadcast meteorologist from around the country. We use AWIPS-II to look at real time weather and issue real time forecast updates (via blog) as well as experimental severe thunderstorm and tornado warnings (via WarnGen). The main focus of the experiment is not necessarily the warnings and getting it right, but we want to know how the forecasters are using the experimental products in their forecast and warnings decision making and providing feedback on those products specifically.

4. Forms of Feedback Blog Daily survey (Mon-Thurs) for all products
End-of-Week Survey (Friday morning)
Debrief Discussions (Tues-Fri mornings)
Real-time discussions
Blog
Mesoscale forecast updates
Reasoning behind warning decisions
Updates to previous warnings/forecasts
Best practices
Ideas for improvement
Any thoughts/feedback, good/bad, about the experimental products
HWT Satellite Proving Ground Blog
Weekly Webinar
We collect many different forms of feedback throughout each week. There is a daily survey which is filled out at the end of each forecast shift (Mon-Th), and end of week survey which is filled out Friday morning. Each morning (Tues – Fri) we have about a 1 hour discussion on thoughts from the previous day’s activity and of course there are many real time discussions going on throughout each day. The biggest and primary form of feedback, especially for the developers is the blog where forecasters make posts about how they are using the products and improvements they would like to see. Each week wrapped up with a weekly webinar presented to the field on cases from that week’s weather events.

5. NOAA Unique Combined Atmospheric Processing System (NUCAPS) in the 2017 Experiment
NUCAPS algorithm generates temperature and moisture profiles using information from instruments aboard the Joint Polar Satellite System (JPSS) Suomi-NPP polar-orbiting satellite.
Products include:
NUCAPS Profile Availability (Time/Location) with quality control flags
NUCAPS Vertical Temperature and Moisture Profiles (SNPP, MetOp-A, and MetOp-B)
Experimental modified NUCAPS Vertical Temperature and Moisture Profiles
Multi-level, gridded plan views of NUCAPS thermodynamic info
Example NUCAPS Coverage with QC flags
Example NUCAPS Sounding
So for NUCAPS this year we had a couple of extra features being evaluated than just the SNPP soundings. We still had the operational SNPP soundings, but also had soundings from MetOp-A and B which made passes in the late morning. So, this gave forecasters a couple extra soundings to look at to begin their shift. We also evaluated an experimental version of NUCAPS based on previous years of feedback, which applies a correction in the boundary layer to hopefully give better representation down low. Gridded plan views and cross sections of various parameters derived from NUCAPS were also examined this year.

6. Example Cases and Feedback
With that I will show a few cases and present some of the thoughts from forecasters on NUCAPS and it’s utility in operations.

7. NUCAPS Used in Fire Case
HRRR Forecast Sounding
NUCAPS Sounding
A noticeable inversion was detected near/just above 700mb.
Compared to HRRR, RAP, and NAM soundings taken at a similar time, guidance was unable to detect this feature.
Decided to investigate a smoke plume seen from KBLX radar
In this case near Billings, MT, NUCAPS sounding pass was around 1930z(seen in the upper left). Several soundings were plotted near Billings and locations to the east, ahead of convection. A noticeable inversion was detected near/just above 700mb. Compared to HRRR (seen in lower right image), RAP, and NAM soundings taken at a similar time, guidance was unable to detect this feature. Additionally, the 18z sounding from Glasgow, MT (much farther north and east than Billings) was compared to the nearest NUCAPS profile. No noticeable differences, that would have a sizable impact on mesoscale forecasting were noticed.

8. NUCAPS Used in Fire Case
“The placement of the fire and smoke plume suggests some accuracy of the NUCAPS capture of the inversion, which is missing from model guidance.”
“Additionally, it has been noticed that as convection has pushed eastward this afternoon, it's intensity has been decreasing, which could be an impact of the inversion.”
To further investigate the accuracy of the NUCAPS profiles taken east of Billings, they used a smoke plume visible at 0.5 and 0.9 degrees on the KBLX radar, near Birney, Montana. This fire is approximately 90 miles from the Radar site, with the 0.5 degree scan intersecting at ~9800 feet AGL. From the nearest NUCAPS sounding to the fire, the strongest point of the inversion was ~8300 ft AGL. The placement of the fire and smoke plume suggests some accuracy of the NUCAPS capture of the inversion, which is missing from model guidance. They chose to investigate the smoke plume, because, given the distance from the Billings Radar, this plume would not likely be visible without an inversion causing the smoke to level off and begin to spread out horizontally. Therefore, the conclusion was that there is confidence that the NUCAPS soundings/profiles were a fair representation of the atmosphere. Additionally, it was noticed that as convection pushed eastward in the afternoon, it's intensity decreased, which could be an impact of the inversion.

9. Using NUCAPS During Storm Interrogation
Another use of these soundings is to quickly find the 0C and -20C levels. Many other tools use model data which can have their own errors, but the NUCAPS is actual observation and can bring more confidence when analyzing storms in the vertical. As storms are beginning to develop this day across the Grand Forks, ND CWA, this forecaster wanted to have the 0C and -20C levels on hand. He quickly went to the latest NUCAPS pass and pulled a sounding within the region. As the forecaster says here: “I found this information very valuable during warning operations. I had the heights of these two levels on hand when I was doing storm interrogation. I used these heights when assessing heights of 50 and 60 dBZ, which helped in my warning issuance.” 
Another use of the NUCAPS soundings is to quickly identify the 0C and -20C levels from an actual observation.
“I found this information very valuable during warning operations.”
“I used these heights when assessing heights of 50 and 60 dBZ, which helped in my warning issuance. ”

10. Operational vs. Experimental Soundings
This case compares the operational and experimental versions of the NUCAPS soundings. First is the 19Z special observed sounding from Aberdeen, SD. Looking at the NUCAPS sounding, it was not quite as dry as the observed sounding above 500mb.  There is also a weak inversion just above 500mb on the observed sounding, and it looked displaced to around 700mb on the NUCAPS sounding.  Of course, its worth noting that the surface conditions on the NUCAPS was about 10 degrees too cool on both the temperature and dewpoint.  Looking at the last image of the experimental NUCAPS sounding with modifications, it looks like the adjustments made were right on target raising the surface temperatures and dewpoint between 8 and 10 degrees.  This provided a much more representative picture of the thermo fields near Aberdeen.  The SPC Mesoanalysis page at this time had nearly 4000J/kg of SBCAPE which was very much in line with the experimental NUCAPS sounding. This was fairly common throughout the experiment. Most thought this was a nice addition to the original NUCAPS sounding.

11. Gridded NUCAPS
Some uses of the gridded NUCAPS that were found most useful was looking at the mid level lapse rates. A lot of forecasters looked at these over the weeks and really found having that actual observation of lapse rates in particular was very helpful in their nowcasting environment. There were some drawbacks to the gridded data which I will mention a little later. You can see here how well the gridded NUCAPS data (on the left) matches up with the RAP model forecast (on the right) for mb and mb lapse rates aside from some resolution issues. This not only helped give the forecaster confidence in the observations and the convective potential on this day, but also helped give a little confidence in the model forecast for the situation as well.
NUCAPS (left) and RAP (right) with a gridded mb Lapse Rates and mb LRs contoured.
“Overall, I think the NUCAPS data provides a good assessment of the mid level conditions in this case.”

12. Gridded NUCAPS NUCAPS RAP 700mb 850mb
This last case looks at mixing ratios to see how they compared with the latest RAP model run.  At 700 mb (top two panels with NUCAPS on left and RAP on right), it looks like both are generally showing a dry tongue stretching up from Tennessee across Missouri.  They are also both in good agreement on the mixing ratios over the forecast area of Sioux Falls, SD.  Overall, very impressed how well NUCAPS matches up with the latest model data at 700mb. Looking a bit lower at 850mb (lower panels), it looks like mixing ratios overall are slightly less than what the RAP is indicating.  This may be an indication we are getting a bit too low in the atmosphere for NUCAPS to handle things. A quote from the forecaster on this case says “Thus, confidence may be a bit better at levels at or above 700mb, but not so good for 850mb or lower.  Overall, as you get closer to the surface, it looks like there is a tendency for NUCAPS to trend towards a drier solution than the models.”
850mb
“Thus, confidence may be a bit better at levels at or above 700mb, but not so good for 850mb or lower.  Overall, as you get closer to the surface, it looks like there is a tendency for NUCAPS to trend towards a drier solution than the models.”

13. Summary and Feedback
Found the availability of NUCAPS soundings very beneficial in the field particularly for offices in the western U.S. and other data void regions.
Some work can be done to increase its utility in the field (better temporal resolution, tailored fields of interest for a forecast, better gridded data viewing...little ragged at times, and training for forecast operations).
It would very useful in the field to be able to get multiple passes during the day to be able to monitor evolution of fields of interest Like mixed layer depth evolution, monitoring inversion (stable layers) which are critical for downslope wind storms, air quality, convective suppression or cap breaking, and stability evolution (assuming better temporal resolution).
Most see the utility in having both the operational NUCAPS soundings available alongside the experimental surface obs adjusted values. Even though the correction may be oversimplified in assuming a perfectly well mixed layer of moisture, it is still useful to see.
Like having both soundings and gridded data to get the full potential of the NUCAPS data.
The gridded NUCAPS were found to be a little clunky at first but currently there are no easy menu options to view data and has to be built using the volume and product browsers. This would need to be streamlined greatly.  The high vertical resolution is also valuable in that it allows you to evaluate plan views of temperature, moisture, possibly even heights of temperature surfaces (example freezing levels, -20, inversion heights) and whatever pressure level is available. 
Latency is still the main issue from using regularly in operations.
The forecasters found that having the NUCAPS soundings was very beneficial, especially in those times in the western US or other areas where data and particularly upper air data are sparse. There are some things that forecasters would like to see to increase its utility in the field, like having more frequent passes to be able to assess evolution of features, having some better gridded data viewing mechanisms and parameters, and having some more tailored fields relevant to a certain forecast situation. Most liked having both the operational and experimental versions side by side to compare and see how well the correction really does. They liked having the gridded data, but like I said earlier, they would like to have some more useful fields to view in the gridded framework. Of course the number one issue heard from every forecaster is the latency, which I know is being worked on, but most think these soundings need to be in front of the forecaster in AWIPS no more than an hour after the passage time to be really useful at least in a convective situation as things change so rapidly.

14. Thank You!
Thank you all for your time and I’ll take any questions now.