Scott Lindstrom and Justin Sieglaff from UW-CIMSS have developed a website for use in this year's GOES-R Proving Ground Spring Experiment to compare the simulated satellite imagery produced from the NSSL-WRF and the observed GOES-13 imagery in real-time. The website provides us with a quick comparison between the model's expected convective development and what is occurring in actuality. This helps us greatly in determining the model's performance on the fly for that day's 00Z run. This site has been and will continue to be used in forecaster briefings in the EFP and EWP to help us decide where convection will develop during the day.
http://cimss.ssec.wisc.edu/goes_r/proving-ground/nssl_abi/nssl_wrf_goes.html
Tuesday, May 25, 2010
Discussing SATCAST at EFP Briefing
After the active convective day on Monday May 24, SATCAST results were presented to at today's 3pm EFP briefing. Several examples were shown were SATCAST performed well and some weaknesses of the algorithm. Some of the dryline activity over TX was not nowcasted by the algorithm because of the strong growth rates of the convection over the GOES temporal resolution of 15 minutes. There was some good discussion about GOES-R 30 second rapid scan and whether that high temporal resolution is useful for the SATCAST algorithm.
EWP potential innovations
Today is not quite as exciting as yesterday, no watches all the way from Canada to Mexico. Some severe weather is expected over the TX panhandle as well as over WI. However, since WI is almost completely covered in cirrus and TX panhandle is relatively clear, we decided to monitor that area for the day in favor of the GOES-R products, localizing over Pueblo, CO and Amarillo, TX.
Archive cases are still building and would be difficult to go through at this point so we decided to run today's IOP like a normal WFO shift, where we will monitor the areas of interest and issue nowcasts for the first half until warnings are needed. This will allow for monitoring of the UWCI and CTC products in their intended roles. This is a wholly new idea for the EWP, as previously it had been focused primarily on warning operations only. The forecasters seem excited to do something similar to what they are expected to do during a normal shift. Depending on today's experience, we may have started a new trend for the EWP...
Archive cases are still building and would be difficult to go through at this point so we decided to run today's IOP like a normal WFO shift, where we will monitor the areas of interest and issue nowcasts for the first half until warnings are needed. This will allow for monitoring of the UWCI and CTC products in their intended roles. This is a wholly new idea for the EWP, as previously it had been focused primarily on warning operations only. The forecasters seem excited to do something similar to what they are expected to do during a normal shift. Depending on today's experience, we may have started a new trend for the EWP...
EWP daily briefing... debrief
During the first part of today's afternoon briefing for the EWP, we discussed the use of the UWCI and OTTC products from the 2130-2300 UTC time period. Below is the main points brought up in the discussion...
OVERALL
Products are provided as images within AWIPS. Image combine dims visible and makes hard to see. Would be nice to have something different... would be better to load a contoured fields... "I want to have as many things loaded on one screen as possible and I can turn them on and off... I had to use 3 panels to display the products and that's not good." (We did this for NAWIPS at SPC last year and is now on the todo list for AWIPS this year)
UWCI and CTC
CI vs CTC... bias towards one?... Liked CTC better... picked up about 15 minutes before radar and better discriminated areas of interest... showed more signal than CI.
Are you ok with more false alarms?... "As long as forecasters are trained, false alarms are not a problem. Even with CI itself... make it less stringent. I can tell if cirrus is over top of it. I know why you're doing it, but maybe you don't have to go as far."
"Is the anyway to have the CI continue to track a vertically developing cloud to see if the development continues?" (This is under development within WDSSII)
Forecaster mentioned that he used the UW-CIMSS CI phone application this morning... He had good impressions and would like to see that sort of display in AWIPS (similar to the "OVERALL" discussion)
OTTC
"I could make out that it was an OT before it could... about 30 minutes before. I guess this was because of the coarse grid, but by the time an OT was detected I already had a hook echo with a tornado on the ground."
We showed the 24-hour detections of overshoots and thermal couplets. The thermal couplets were much more confined to severe areas and the overshoots were much more widespread and matched with areas of CG activity, but both seemed to lag occurrence of severe weather.
OVERALL
Products are provided as images within AWIPS. Image combine dims visible and makes hard to see. Would be nice to have something different... would be better to load a contoured fields... "I want to have as many things loaded on one screen as possible and I can turn them on and off... I had to use 3 panels to display the products and that's not good." (We did this for NAWIPS at SPC last year and is now on the todo list for AWIPS this year)
UWCI and CTC
CI vs CTC... bias towards one?... Liked CTC better... picked up about 15 minutes before radar and better discriminated areas of interest... showed more signal than CI.
Are you ok with more false alarms?... "As long as forecasters are trained, false alarms are not a problem. Even with CI itself... make it less stringent. I can tell if cirrus is over top of it. I know why you're doing it, but maybe you don't have to go as far."
"Is the anyway to have the CI continue to track a vertically developing cloud to see if the development continues?" (This is under development within WDSSII)
Forecaster mentioned that he used the UW-CIMSS CI phone application this morning... He had good impressions and would like to see that sort of display in AWIPS (similar to the "OVERALL" discussion)
OTTC
"I could make out that it was an OT before it could... about 30 minutes before. I guess this was because of the coarse grid, but by the time an OT was detected I already had a hook echo with a tornado on the ground."
We showed the 24-hour detections of overshoots and thermal couplets. The thermal couplets were much more confined to severe areas and the overshoots were much more widespread and matched with areas of CG activity, but both seemed to lag occurrence of severe weather.
UWCI and OTTC first forecaster interactions
A new set of NWS WFO personnel from St. Louis, Des Moines, Buffalo, and New York were giving GOES-R training with respect to future satellite-based decision support for convective initiation, convective overshooting-tops, and thermal couplets (enhanced-V) detection. All the forecasters were very receptive to the training and asked excellent questions with regard to temporal latency and how products might help with situational awareness. After training was conducted each forecaster was given access to AWIPS workstation and allowed to set up environment as they would at their office. Four panel displays were constructed consisting of UWCI cloud top cooling rate (CTC)/convective initiation signal plotted over GOES imager visible imagery and vertically composited radar. Forecasters in general felt that the CTC rate provided more lead time and better quantitative magnitude based on verbal communication. In some instances no reflectivity was present on radar for up to 15 minutes after CTC indication, in other cases the product acted more as a diagnostic. Chris and I will look through evaluation forms (filled out after AWIPS interaction last night) and use those to help guide today's training. Here is a summary of yesterday's 24 hours (12 UTC - 12 UTC) UW CI/CTC compared to SPC reports and accumulated NLDN.
Initial evaluation of Overshooting-top/Thermal Couplets (OTTC) also occurred but forecasters spent most of their time first familiarizing themselves with UWCI products.
Initial evaluation of Overshooting-top/Thermal Couplets (OTTC) also occurred but forecasters spent most of their time first familiarizing themselves with UWCI products.
Monday, May 24, 2010
EWP examining potentially tornadic storms during IOP
The EWP is performing an informal IOP to familiarize the forecasters with the experimental data they will be using for the rest of the week following the afternoon training session. Today's focus for the first half of the evening is for the UWCI and OTTC products over N. TX into W. KS where a threat for tornadoes is expected into the late evening and overnight. Forecasters are examining the CTC product as well to determine the relative strength of the storm updrafts, which is made possible because the CTC product does not shut off until the storm has glaciated for two scans. Preliminary interactions with the forecasters have been promising regardless of some continued issues with the OTTC data. Forecasters have regularly been seeing lead times with the UWCI and CTC products of 15 minutes over radar, as well as their own subjective determinations of when convective initiation has occurred based on satellite or radar. They have also been finding that the CTC product has been providing them with some valuable information regarding the locations of the most vigorous updrafts along a line of storms. Wayne Feltz will be providing a more detailed update on his interactions with the forecasters.
UAH SATCAST_v2 Example Near New Orleans, LA
SATCAST-v2 Output from
1745 UTC GOES-13 Imagery
1745 UTC GOES-13 Imagery
SATCAST-v2 Output from 1815 UTC
GOES-13 Imagery
GOES-13 Imagery
The SATCAST_v2 algorithm forecasted the first convection of the day along the North Shore of Lake Pontchartrain with just under 30 minutes lead time for the cell in western St. Tammany Parish (see top-left image, above) and between 15-30 minutes for the cell along the St. Tammany Parish / Hancock County border (see top-left image and lower-right image, above). Because it was the first CI in the area for today, the 1744 UTC radar image, below, showed a clear scan across the entire region. However, these storms, and more, became apparent on radar only after a half hour or more elapsed. Within the next hour, the NWS office in New Orleans issued Special Weather Statements for this storm cluster, and they eventually had to issue the first Severe Thunderstorm Warning on the slow moving western cell around 2007 UTC.
Radar data valid at
1744 UTC.
Radar data valid at
1813 UTC.
Radar data valid at
1832 UTC.
EWP daily briefing... Model-based product development
Today was the orientation briefing for the NWS forecasters participating in the EWP this week. As part of the briefing we present a short weather discussion where we will ultimately select the region where we will be focused on for the day. Today during the briefing I once again showed the forecasters the NSSL-WRF simulated satellite imagery and lightning threat forecasts and explained how they work. One forecaster asked about generating products from these model data, in particular the 10.7-3.9 channel difference for low cloud and fog. I mentioned that we hope to be developing products from the simulated model output in the future as a proxy for the real products to be available on GOES-R rather than relying on current instruments which may not be able to provide us with the spectral, spatial or temporal we need to produce these products. In addition, Wayne Feltz mentioned that the eventual plan would be to have some sort of real-time rapid updating model forecast imagery to simulate observed satellite imagery and products for use in the Proving Ground.
The plan for the day is to operate over the WFO Dodge City, KS and Amarillo, TX CWA regions and focus on the GOES-R products first while storms are hopefully still initiating to get some feedback from the UWCI and OTTC products while they are most applicable. Unfortunately, at the time of the training, which took up the first half of the EWP day, severe thunderstorms were already ongoing with associated tornado and severe thunderstorm watches all the way from the borders of Canada to Mexico. (see images below) Therefore, getting some quality CI events may be difficult due to expanding cirrus from the already mature convection... updates will be provided as the day progresses.
SPC watches and associated radar reflectivity by 2000 UTC on 24 May 2010
The plan for the day is to operate over the WFO Dodge City, KS and Amarillo, TX CWA regions and focus on the GOES-R products first while storms are hopefully still initiating to get some feedback from the UWCI and OTTC products while they are most applicable. Unfortunately, at the time of the training, which took up the first half of the EWP day, severe thunderstorms were already ongoing with associated tornado and severe thunderstorm watches all the way from the borders of Canada to Mexico. (see images below) Therefore, getting some quality CI events may be difficult due to expanding cirrus from the already mature convection... updates will be provided as the day progresses.
SPC watches and associated radar reflectivity by 2000 UTC on 24 May 2010
SATCAST Convective Initiation Examples Over The Memphis Area
Today there is significant convection forming over the Memphis area. Since Memphis is an important airline hub for Delta and Fedex, knowledge of convective initiation is important for planning of airport operations.
Below are some images from the UAHuntsville SATCAST over the Memphis area at 1515 UTC. The red indicates the likely area of convective initiation.
Notice the red area in southwest Tennessee. The radar image below at 1516 UTC shows a complex of storms in west central Tennessee, but no convection in the area of the SATCAST nowcast.
Other radar images from 1526 UTC and 1531 UTC respectively are below.
It's important to note that some of the convection was missed over the TN/AR border due to the temporal difference between the images is 30 minutes due to a full disk scan (full disk scan on current GOES takes ~26 minutes to complete which occured at 1445 UTC and every 3 hours) and a housekeeping procedure at 1532 UTC. Because the convection was rapidly developing, 30 minutes was too long to provide sufficient lead time for some of the storms. Notice how quickly the storms grow over 30 minutes in the two 10.7 micron images below (the image times are 1445 UTC and 1515 UTC, respectively):
With GOES-R, CONUS scans will be available every 5 minutes which will be able to provide better lead-times for the convective initiation algorithm as opposed to the current 15-30 minute temporal resolution current available with GOES.
As the morning progressed, the aerial coverage of the convection increased, and SATCAST was able to depict this increase. Below are some examples of SATCAST depicting the increase at 1602UTC.
And the radar image from this region is below at time 1615 UTC:
It is also important to note that some of the smaller cells in the image above are not detected by the SATCAST algorithm. This is likely caused by the small spatial scale of the individual storms being less than the 4km spatial resolution of GOES in the Infrared region. GOES-R will provide an infrared spatial resolution of 2km which will allow for the smaller convective storms to be identified.
Week 2 begins... NSSL-WRF updates
It's Monday, which means week two of the Spring Experiment has officially begun. Bill Campbell from NWS HQ, Louis Grasso from CIRA, Wayne Mackenzie and John Walker from UAH, and Wayne Feltz and Jason Otkin from UW-CIMSS are here to participate within the Spring Experiment activities and interact with the forecasters this week. This morning was spent making some changes to the NSSL-WRF simulated satellite imagery. UW-CIMSS was having some issues with the 17Z forecast time period of their run, so their output now begins at 18 UTC. This did not affect the use of the data within the EFP much since most of the forecasts begin at 18 UTC. The times have successfully been extended out to the 03Z forecast period for all bands from UW-CIMSS as well. CIRA extended it's forecast times from 12-00Z to 12-06Z for bands 8 and 13 over the weekend. This change has been implemented in our ingest scripts and the images are now available within any SPC and HWT N-AWIPS system.
Subscribe to:
Posts (Atom)