During today's debrief of the EWP participants, I was able to ask them some specific questions and get feedback on their experience during the week with the GOES-R Proving Ground products. They provided some new ideas as well as some constructive comments and criticism. Here is a breakdown of the information I captured during the discussion...
REGARDING UWCI AND CTC
- In marginal cases when we're not sure whether storms will go it may be more useful
- Nighttime when no visible is around it is definitely more useful
- In winter nighttime snow events, low top convection close to radar not detected... cooling tops earlier might give clue of snow showers occurring... also upslope snow showers, where mountains inhibit radar reflectivities would be a good thing to see
- Didn't have enough time with the products to determine if CI or CTC more useful
- Was slower than radar constantly
- They realize that the products are meant for better temporal and spatial resolutions, but hard to see it's use when not available
- Just because cloud top is cooling, it may not be the dominant storm... may not be n the favorable location or on right side of boundary... using any algorithm without paying any attention to the environment is not proper use of the data in forecasting operations... need to combine and/or base the products on environmental information
- The cases this week were slam dunk, so CI was pretty sure to happen... not so useful then
- During May 19 case over OUN... initially watched for CI on southern storms... once one went, they monitored radar... may be useful in detecting first storm of the day
- "Not a lot of meteorologists are gonna have confidence on just one case... need to see more than that"
REGARDING OTTC
- When not looking at discrete supercells, it may be more useful in detecting embedded supercells
- Need to establish (and communicate) better the relationship between OT and severe weather potential
- Erosion of OT would be much more useful than a detection... updraft collapse or weakening... OT magnitude trends would be much more useful
- Difference the cloud-top temp of the overshoot with model temp of level to give an idea of height above EL for instance would be interesting
- Detection not important alone, especially when seeing core aloft on radar... However, where radar coverage sparse, might be more useful (eg. out west)
REGARDING PSEUDO-GLM
- "We definitely did see an increase in lightning rates associated with increases in storm intensity and cores aloft."
- Want to see less smoothing and more texture... just seeing blobs... knocking down max values (we are working on this)
- In the one very marginal case it was saying that the strongest storms were where the radar said they were.
- Concern raised over closeness to network effects... seemed like higher detections over weaker storms... Eric Bruning mentioned that the flash detection should even that out regardless
- "I appreciated having all the lightning rather than just having CG"
- Would like to see how GLM relates to storm intensity... eg. isolated supercells, multi-cells
- "The data makes sense, but what would that mean to me? What does it really tell me?"
- Marginal cases not a good start to understand the data... need training case.
- Eric Bruning mentioned to the forecasters that satellite flash extents may expand due to cloud scattering
REGARDING INTEGRATING SENSOR DATA
- "I think that's the future, stand alone sensors are less important than integrated together."
- "I envision a virtual world of sort... some fast update model that physically builds the storm based on combined data... model updates every time the sensor data updates."
- "What I want is a Star Trek holodeck."
OVERALL COMMENTS
- 2 weeks would be better... essentially we had 3 days of IOP... would be better to have a few full days of training/archive rather than 3 hours and say "go"
- "Having a tool we are used to using is good" (ie - AWIPS)
- Displaying satellite products (CI especially) might be better used as icons that can be overlaid rather than images... size of icon might indicate intensity... images require toggling between two images rather than overlaying
- Might be useful to provide us with WES data prior to arrival so we can go through it before arriving... or maybe a 'gotomeeting' for training the week before arrival
Friday, May 21, 2010
Thursday, May 20, 2010
Examining 8-km total lightning over Huntsville
.jpg)
Tonight's IOP focuses on severe weather warning over the Huntsville, AL WFO CWA. Huntsville was chosen because of the availability of total lightning data from the Northern Alabama Lightning Mapping Array (NALMA). This provides us the unique opportunity to examine the 8-km pseudo-GLM total lightning flash extent density product in real-time warning operations. Forecasters have noticed a few interesting features of the data, including it's ability to pick up on trailing stratiform lightning over more intense convective storms. The raised the fact that this would be very useful for aviation type forecasts and warnings, as well as lightning warnings for the general public who may think that they are out of danger from lightning after the main convective region passed. They also noticed that the product located areas of more intense updrafts within line segments that are otherwise indistinguishable on radar reflectivity. These more active regions were found to highlight regions where small intense circulations were seen within the lines, which may lead to short tornadic threats. Forecasters were also able to compare the pseudo-GLM data to NLDN via the situational awareness screens on the back wall of the HWT. It was noticed that, particularly in the trailing stratiform regions, that the 8-km total flashes extent densities showed much more information on the total lightning occurring along the line of convection. Total flash rates were available in Google Earth. As expected, lightning jumps were seen with rapidly increasing radar reflectivity. However, the unavailability of these data within the local AWIPS system hindered the forecasters ability to use it effectively, but they expressed lots of interest in seeing this in the future.
The data are being provided with AWIPS as a smoothed grid so that forecasters are able to overlay the data on radar or satellite imagery. The smoothing has somewhat masked the fact that the data are on an 8x8 km grid, but it does seem to show details that are being used to detect more active regions nonetheless. The data are also provided as a running 2-min average, which may be different that what the GLM will provide. However, this may turn out to be a useful way to process the data once the GLM is launched in order to reduce the amount of data being transferred into every AWIPS system.
.jpg)
Thoughts From Today's Look at UAHuntsville's SATCAST_v2 CI Nowcasting Product
From a look at the real-time runs of SATCASTv2 over the past few days, it appears that there are a few issues in terms of performance with respect to, both, false alarms and missed events.
False Alarms:
It seems that early in the day in
stable environments, the algorithm tends to flag developing boundary layer cumulus clouds, forecasting them to convectively initiate, when all they end up doing is growing up to the level of a lower tropospheric temperature inversion and stop development (see Figures below).
Sounding from Springfield, MO
valid at 1200 UTC, 05/20/2010.
Note the low-level temperature inversion.
GOES-13 visible satellite imagery valid at 1432 and 1445 UTC, respectively. Note the rapid early growth of boundary layer fair weather Cu clouds (Above).
SATCAST_v2 CI Nowcast from 1445 UTC imagery (Above).
In some ways, this is discouraging (because they obviously aren't going to produce rainfall anytime soon), but in other ways it is encouraging to see that the system is sensitive enough to detect growing clouds, even in the earliest stages of growth. In order to reduce the number of false alarms in these instances, we've tried to tune some of the CI interest fields (remember, there are 6 of them used for current GOES, and 5 out of 6 of them must "pass" in order for a cloud object to be flagged for a CI forecast) so that they would become less sensitive to very early, lower level cloud growth.
Misses:
On the other hand, SATCASTv2 tends to have issues in highly unstable, uncapped environments, such as the "soupy" airmass that usually is encountered across the southeast U.S. in the warmer months. Often, in these environments, as was the case today (see Figures below), the algorithm becomes much more of a diagnostic tool than a prognostic tool, flagging cloud objects for future CI, right around the same time the corresponding radar scans detect echoes of 35 dBZ or greater... What we consider a "miss". Or.. It simply misses CI altogether. There are, perhaps, two issues that plague the algorithm in these environments:
1) As soon as surface-based air parcels begin to ascend in the unstable airmass, clouds grow very rapidly and become very efficient at producing rainfall, even with relatively narrow initial updrafts. Therefore, tuning the CI interest fields to become less sensitive to early cloud growth in order to decrease our false alarm rates (as mentioned above) actually severely limits us here.
2) Cumulus clouds in these environment sometimes possess very narrow updrafts that remain smaller than the current 4km resolution IR channel pixels can resolve (sub-scale pixels). So, the small-scale objects that we ARE able to track possess pixels that contain sub-scale growing Cu clouds that appear warmer in the respective IR pixels than the cloud tops actually are. There isn't really much we can do about this problem... It is a known limitation of our current GOES satellite instrument... but higher spatial resolution IR data in the coming years aboard GOES-R should dramatically assist with this issue.
SATCASTv2 Forecast at 1415 UTC
05/20/2010
Radar Base Reflectivity at 1416 UTC
05/20/2010
As a team, (myself... John Walker, Wayne Mackenzie, and John Mecikalski) are all working to embrace these challenges and to come up with solutions to these current limitations in the product. One potentially viable solution we may have is to create two sets of CI Interest fields... one set that is more sensitive to early cumulus cloud growth that can be used in highly unstable, uncapped environments (this would at least help with problem #1 under the "Misses"section)... and another set of fields that is less sensitive, to be applied in more stable environments where mainly boundary layer cloud development only is expected (this might help to decrease our number of false alarms).
Of course, we'd need some sort of intermediary input into the algorithm that would divide our domain into regions of "highly unstable and uncapped" locations and regions that are stable with some sort of low-level inversion to prevent much more than very early stage cumulus growth. Perhaps this can be accomplished with high resolution model data (the HRRR maybe)... but the better route would probably be to make use of vertical atmospheric columnar information generated from geostationary sounder data... the higher the vertical resolution the better (only then could subtle low-level inversions be consistently and accurately detected on the fly). Anyway, as reads the title, these are just a few thoughts and observations from the current experiment.
False Alarms:
It seems that early in the day in
stable environments, the algorithm tends to flag developing boundary layer cumulus clouds, forecasting them to convectively initiate, when all they end up doing is growing up to the level of a lower tropospheric temperature inversion and stop development (see Figures below).
Sounding from Springfield, MO
valid at 1200 UTC, 05/20/2010.
Note the low-level temperature inversion.
GOES-13 visible satellite imagery valid at 1432 and 1445 UTC, respectively. Note the rapid early growth of boundary layer fair weather Cu clouds (Above).
SATCAST_v2 CI Nowcast from 1445 UTC imagery (Above).
In some ways, this is discouraging (because they obviously aren't going to produce rainfall anytime soon), but in other ways it is encouraging to see that the system is sensitive enough to detect growing clouds, even in the earliest stages of growth. In order to reduce the number of false alarms in these instances, we've tried to tune some of the CI interest fields (remember, there are 6 of them used for current GOES, and 5 out of 6 of them must "pass" in order for a cloud object to be flagged for a CI forecast) so that they would become less sensitive to very early, lower level cloud growth.
Misses:
On the other hand, SATCASTv2 tends to have issues in highly unstable, uncapped environments, such as the "soupy" airmass that usually is encountered across the southeast U.S. in the warmer months. Often, in these environments, as was the case today (see Figures below), the algorithm becomes much more of a diagnostic tool than a prognostic tool, flagging cloud objects for future CI, right around the same time the corresponding radar scans detect echoes of 35 dBZ or greater... What we consider a "miss". Or.. It simply misses CI altogether. There are, perhaps, two issues that plague the algorithm in these environments:
1) As soon as surface-based air parcels begin to ascend in the unstable airmass, clouds grow very rapidly and become very efficient at producing rainfall, even with relatively narrow initial updrafts. Therefore, tuning the CI interest fields to become less sensitive to early cloud growth in order to decrease our false alarm rates (as mentioned above) actually severely limits us here.
2) Cumulus clouds in these environment sometimes possess very narrow updrafts that remain smaller than the current 4km resolution IR channel pixels can resolve (sub-scale pixels). So, the small-scale objects that we ARE able to track possess pixels that contain sub-scale growing Cu clouds that appear warmer in the respective IR pixels than the cloud tops actually are. There isn't really much we can do about this problem... It is a known limitation of our current GOES satellite instrument... but higher spatial resolution IR data in the coming years aboard GOES-R should dramatically assist with this issue.
SATCASTv2 Forecast at 1415 UTC
05/20/2010
Radar Base Reflectivity at 1416 UTC
05/20/2010
As a team, (myself... John Walker, Wayne Mackenzie, and John Mecikalski) are all working to embrace these challenges and to come up with solutions to these current limitations in the product. One potentially viable solution we may have is to create two sets of CI Interest fields... one set that is more sensitive to early cumulus cloud growth that can be used in highly unstable, uncapped environments (this would at least help with problem #1 under the "Misses"section)... and another set of fields that is less sensitive, to be applied in more stable environments where mainly boundary layer cloud development only is expected (this might help to decrease our number of false alarms).
Of course, we'd need some sort of intermediary input into the algorithm that would divide our domain into regions of "highly unstable and uncapped" locations and regions that are stable with some sort of low-level inversion to prevent much more than very early stage cumulus growth. Perhaps this can be accomplished with high resolution model data (the HRRR maybe)... but the better route would probably be to make use of vertical atmospheric columnar information generated from geostationary sounder data... the higher the vertical resolution the better (only then could subtle low-level inversions be consistently and accurately detected on the fly). Anyway, as reads the title, these are just a few thoughts and observations from the current experiment.
Total lightning IOP plans
At 1:30 PM, the the North Alabama Lightning Mapping Array (NALMA) showed lightning activity along the northern Mississippi-Alabama border. The 00Z 20 May NSSL-WRF run in support of the NSSL/SPC EFP shows continued evolution of this convection toward central Alabama by 00-02Z this evening.
The lightning threat field in the NSSL-WRF using the McCaul blended vertically integrated ice / graupel flux method shows lightning activity extending north-south through Alabama at 1Z. The predicted flash rates are somewhat less over the far northern part of the domain.
An intensive operations period is planned in the EWP over the northern Alabama domain beginning 23Z, where Pseudo-GLM total lightning data derived from the NALMA will be examined along traditional radar-based warning methodology for insight about the location and trends of the most intense updrafts and enhanced threat for severe weather. If the model forecasts verify, we expect the most intense convection over central Alabama with weaker convection over the northern part of the state.
EWP daily briefing... 5/20/2010
.jpg)
Here is some very interesting dialogue from today's EWP 1pm daily briefing. Kris Bedka and Justin Sieglaff posed questions to the forecasters regarding the UWCI and OTTC products and these were they're responses...
In the case of multiple towers with equal weak reflectivities on radar, would CI detection over one and not the other add value to your forecast?
"It's plausible. If that's the way it play out. It would be hard to objective analyze that all the time."
"It does have utility in determining which storm will be the dominant storm."
"Depends on the environment."
When visible satellite is not available, is it (CI detections) even more useful? (this was simulated in yesterday's IOP for one forecaster)
"Yes, absolutely" (also agreed by the other forecasters in theory)
Does a continued signal in CI over one storm provide increased confidence?
"The opposite may be more useful in determining if the storm way dying. It's the end of an event that gives people the most trouble." (forecasters reiterated the need for object tracking)
General discussion on overshooting-tops...
"Overshooting-tops are much more common than I was led to believe in school."
"We did observe that the reflectivity core aloft did intensify when an overshoot was detected."
"Would be interesting to see how they correlate with the occurrence of BWERs. If we had temporal continuity (a continuous) detection, we would expect to see a BWER."
"Overshooting dissipation would be interesting in determining the occurrence of tornado or large hail at the surface."
"If we had rapidly updating satellite, it would be much more useful."
"Might also be useful to have a OT track product... would help in determining storm (updraft) motion."
In addition, we discussed the inclusion of satellite data into an end-to-end tracking system to better compare to radar signatures objectively. It was also noted that next year, it is expected that the EWP will have a more robust AWIPS system that will be able to ingest a total feed (similar to a WFO), that will be on all year that can have new experimental products running on it constantly. This will provide year-round testing of products, rather than a month long experiment. They also plan on having this ready earlier in the year before next year's experiment so dataflow issues are reduced.
We showed the NSSL-WRF simulated reflectivity and lightning threat output for the weather briefing to help decide location for which we would localize the AWIPS stations. Simulated reflectivity showed a large line of storms moving through the area sometime around 00Z. The lightning threat fields showed that the most most lightning activity was expected along the southern edge of the line of storms as the evening progressed. We are hoping that we can see if the addition of lightning data on this southern edge will help determine which areas are the bigger threat during the decision making process for warning along the entire line. Since the storms aren't expected to form in the Huntsville area until the later-half of the EWP IOP, it was decided that we would focus on S. TX for the first half of the day, and then following the dinner break we would switch to Huntsville, AL domain to get some pseudo-GLM demonstration. Will provide updates as they arrive.
Severe hail probability forecasts hail over OK
Yesterday's severe storms dropped baseball sized hail just south of Norman, OK. The 0-3 hour severe hail probability forecast picked up on the storms that dropped hail over Oklahoma after 0 UTC. On the attached image, severe hail is plotted as blue letter a's from 21-03 UTC... most of which occurred after 0 UTC. The probability forecast made at 2200 UTC was valid until 0100 UTC. The higher probabilities up north represent already tornadic storms occurring over Stillwater, OK that have had a history of producing hail in the past. The intensifying probabilities extending south picked up well the storms that would produce hail in the near future.
Overshooting Top Detections Identify Period of Frequent Severe Weather 5/19/2010
The overshooting top (OT) detection product detected a period of frequent severe weather reports for the 19 May 2010 event. Initially the cloud top brightness temperatures were too warm (> 217.5 K) for the algorithm to detect OT signatures between the time of the first reported tornado at 2028 UTC and 2210 UTC. Our first OT detection occurred at 2210 UTC and these continued throughout the duration of the severe weather event across Oklahoma. Though it may have seemed to the forecasters here at the HWT that the OT product did not have much value over radar based severe storm signals, 31 of the 37 severe storm reports in Oklahoma occurred on or after 2210 UTC. See the graphic below for an accumulated plot of OT detections from 12UTC on the 19th to 12 UTC on the 20th.
NSSL-WRF simulated satellite WV imagery from 5/19/2010
This morning we compared some of the output from the simulated satellite imagery from yesterday's severe weather over Oklahoma. The images attached represent the 2300 UTC band 9 (~6.9 micron) imagery forecasts from both CIRA and UW-CIMSS from yesterday's 0Z run of the NSSL-WRF, as well as the 2302 UTC observed WV image from GOES-13. It should be noted that the forecast imagery and the actual observed imagery do not represent the exact same wavelength (6.9 vs 6.7 micron), so the colors on the enhancement and the extent of cold cloud tops will not be exactly the same. Nonetheless, we were able to make some general comparisons and observations from examining the imagery.
May 19, 2010 2302 UTC observed GOES-E WV imagery
May 19, 2010 2300 UTC NSSL-WRF CIRA simulated GOES-R band 9 imagery
May 19, 2010 2300 UTC NSSL-WRF UW-CIMSS simulated GOES-R band 9 imagery
Overall, the NSSL-WRF did a good job simulating the location and timing of convection over most of the CONUS. Terrain induced convection along the Rockies was also well captured by the model. There were a few places where the NSSL-WRF may have been a little fast (such as southern OK) or a little slow (such as over LA) in developing convection. The simulated satellite imagery correctly estimated the timing, location and extent of the development of storms over northern OK that ended up having many severe storm reports, including multiple tornadoes.
Comparing the two simulated satellite outputs, we can see a difference between the two runs, with the CIRA trending towards 'warmer' images (increase in red hues). The two outputs provided by UW-CIMSS and CIRA have different methods of performing the radiative transfer within the model. It is expected that this is the cause of the differences, and these will be examined in more detail next week when participants directly related to the simulated satellite imagery arrive.
Overall, the NSSL-WRF did a good job simulating the location and timing of convection over most of the CONUS. Terrain induced convection along the Rockies was also well captured by the model. There were a few places where the NSSL-WRF may have been a little fast (such as southern OK) or a little slow (such as over LA) in developing convection. The simulated satellite imagery correctly estimated the timing, location and extent of the development of storms over northern OK that ended up having many severe storm reports, including multiple tornadoes.
Comparing the two simulated satellite outputs, we can see a difference between the two runs, with the CIRA trending towards 'warmer' images (increase in red hues). The two outputs provided by UW-CIMSS and CIRA have different methods of performing the radiative transfer within the model. It is expected that this is the cause of the differences, and these will be examined in more detail next week when participants directly related to the simulated satellite imagery arrive.
Wednesday, May 19, 2010
EWP engaged in IOP on tornadic supercells in W. OK
The EWP is now issuing experimental warnings and severe weather statements for the multiple tornadic supercells in western and central OK using GOES-R Proving Ground and Multi-Radar Multi-Sensor (MRMS) products. Convective initiation has been continuously detected by the UWCI and CTC products in advance of significant reflectivities on radar. Overshooting tops are beginning to be detected (2210 UTC) on the northern-most supercells, however, they may not provide additional information (eg. storm reports) forecasters already have available at this time.
EWP ready to go... 5/19/2010
Just got done with EWP's daily afternoon briefing and we have selected to localize over Norman for the day on both teams due to the obvious severe weather threat over the area this evening.
Some notes from the briefing...
1) Localization selection yesterday automatically selected GOES-West for the satellite imagery, which caused us to lose the RSO imagery for the entire IOP... this was noted to the coordinators and should not happen again.
2) Kris Bedka showed a 24-hour verification in the briefing (and earlier in this blog) that OTs did occur in concurrence with severe weather last night and matched the tracks of the severe reports very well.
3) Justin Sieglaff showed us the verification for the NSSL-WRF simulated satellite imagery from the beginning of today and the forecasts match with the current actual imagery from the same time... so we were able to determine that the NSSL-WRF was accurately developing cloud features within its forecasts.
4) The NSSL-WRF lightning threat forecast was shown to the forecasters for this evening and it helped us identify which storms may have stronger updrafts because of their increased lightning output, which we couldn't necessarily determine from the synthetic satellite or radar output.
Will be providing updates as the day progresses...
Some notes from the briefing...
1) Localization selection yesterday automatically selected GOES-West for the satellite imagery, which caused us to lose the RSO imagery for the entire IOP... this was noted to the coordinators and should not happen again.
2) Kris Bedka showed a 24-hour verification in the briefing (and earlier in this blog) that OTs did occur in concurrence with severe weather last night and matched the tracks of the severe reports very well.
3) Justin Sieglaff showed us the verification for the NSSL-WRF simulated satellite imagery from the beginning of today and the forecasts match with the current actual imagery from the same time... so we were able to determine that the NSSL-WRF was accurately developing cloud features within its forecasts.
4) The NSSL-WRF lightning threat forecast was shown to the forecasters for this evening and it helped us identify which storms may have stronger updrafts because of their increased lightning output, which we couldn't necessarily determine from the synthetic satellite or radar output.
Will be providing updates as the day progresses...
Subscribe to:
Posts (Atom)