The Cloud Top Cooling (CTC) product was utilized by an SPC forecaster in monitoring for severe weather development across northern Utah into western Wyoming on Wednesday, May 28 2014. This region is characterized by especially poor radar coverage due to mountain beam-beam blocking and coarsely spaced radars (Fig. 1). For this reason, satellite-based products are especially useful to SPC forecasters in the western third of the the United States. In this particular case, the CTC product was helpful in highlighting when rapid convective development was beginning, as well as where the most rapid development was occurring (Fig. 2). Severe winds were eventually reported with these storms. The CTC product was mentioned in SPC Mesoscale Convective Discussion (MCD) 748 (Fig. 3): “THE GOES-R CLOUD TOP COOLING PRODUCT INDICATED COOLING TOPS OVER SWRN TO WEST-CENTRAL WY AND ONGOING CLOUD TOP COOLING IN NERN UT.”
Fig. 1: United States radar coverage map.
Fig. 2: 140528/1715 – 2245 UTC GOES-West visible imagery, Cloud Top Cooling (color fill), storm reports.
GOES-14 provided the Storm Prediction Center (SPC) with 1-minute Super Rapid Scan Operations for GOES-R (SRSOR) imagery May 8-25. This gave SPC forecasters an opportunity to use and comment on a capability that will be available with the next generation geostationary satellite system, GOES-R. The GOES-R Advanced Baseline Imager (ABI) will be capable of scanning one 1000×1000 km sector every 30 seconds, or two sectors every one minute. For more 1-minute imagery information and examples, please visit the following page: http://cimss.ssec.wisc.edu/goes/srsor2014/GOES-14_SRSOR.html
The forecasters at SPC who had the opportunity really valued viewing this high temporal resolution imagery, and were sorry to see it go. Adding to the benefits of this imagery is the fact that the latency of each image is typically 2-4 minutes after the time stamp, something the forecasters value greatly. It seemed as though every day when I would peak into operations, at least one forecaster would have the 1-minute imagery looping on a workstation. I gathered informative, detailed feedback from SPC forecasters regarding features and processes they are able to see in this imagery that are otherwise not evident or more difficult to discern in regular 5-30 minute imagery.
One example of an SPC forecaster using the 1-minute visible imagery in his forecast decision-making came during the early evening hours of May 21. Convection was already ongoing across the western third of Texas by 2200 UTC. However, one of the mesoscale forecasters observed newly developing convection in the Texas panhandle as well as in SW Texas via the rapid updating, 1-minute imagery.This indicated to him that the severe threat would continue over the next couple of hours. The observation was noted in SPC MCD #0647: GOES 14 ONE-MINUTE IMAGERY SHOWS CONTINUED UPDRAFT GENERATION WITHIN A MORE MATURE CLUSTER JUST E OF AMA…AND ADDITIONAL TSTM DEVELOPMENT W OF MAF…SUGGESTIVE OF A CONTINUED SVR HAIL/WIND THREAT FOR AT LEAST THE NEXT 1-2 HRS.
Below is a loop of the 1-minute visible imagery over the region for the hour before this MCD was written (Fig 1), as well as the MCD graphic (Fig 2).
SPC forecasters have found the 1-min imagery to be quite useful when monitoring mature convective evolution, as is seen in the forecaster comments below:
“Post-storm initiation, the high-resolution data allowed for careful analysis of overshooting and collapsing tops, the character of the storm anvils (ie. health of the storm) and the identification of convectively generated outflows.”
“I found it very useful in… Assessing the trends of convection via evolution of cloud character in the storms themselves (overshooting tops, subjective visual assessment of storm-top divergence and flanking-line development, etc.)”
The use of the UW-CIMSS Cloud Top Cooling (CTC) product in the Storm Prediction Center (SPC) has been highlighted in a several previous posts on this blog. Please view previous posts for a brief description of this product, along with its utility in operational environments. This post demonstrates yet another example of this product being utilized in the decision-making process of SPC forecasters.
On May 10, SPC highlighted a slight risk for severe weather across eastern Kansas into much of Nebraska. The threat included significant hail, damaging winds, and even a few tornadoes. One of the biggest challenges facing SPC forecasters is not only being able to identify where convective initiation is going to occur (or is concurring) in the near future, but also which of the developing storms will be the strongest and most likely ones to produce severe weather (information beyond just CI). The CTC product can be used for this purpose, as it tells the forecaster where you have convection developing, but also quantifies that vertical growth.
Looking at the animation below, convection began initiating in south-central Kansas on the 10th shortly after 1800 UTC along a dryline/warm front (Fig 1). At 1830 UTC, the CTC product indicated weak cooling, and by 1915 UTC very strong cooling (<-20K/15 min) was detected, indicating that convection was beginning to develop rapidly. The first severe weather with this storm was reported at 2021 UTC (1″ hail), over one hour after the first very strong CTC detection. Storms continued to exhibit rapid cooling over the next several hours as they initiated to the southwest and northeast along the boundary. These storms produced widespread severe hail, wind and even a couple of tornadoes across Kansas and into Missouri (Fig 2).
Figure 2: SPC severe storm reports across Kansas and Missouri
The SPC forecaster on shift at the mesoscale desk was monitoring the CTC product during this event (FIg 3). In a Mesoscale Discussion (MD) issued at about 0100 UTC, the forecaster mentions the CTC product: “…WITH THE GOES-R CLOUD TOP COOLING PRODUCT INDICATING THE GREATEST COOLING ATTENDANT TO THE STORMS IN THESE THREE KS COUNTIES.” This is an example of how the CTC product gave the forecaster additional confidence to where the strongest storms were developing within the watch box. Full MD can be found at: http://www.spc.noaa.gov/products/md/md0555.html
Convective initiation has once again lined up pretty well with the instability depicted on the theta-e difference graphic in the nearcast. I have now seen this in a couple of cases. The nearcast forecast brings some instability north into Oklahoma later this evening. It will be interesting to see how this lines up with the convection at that time, as the HRRR model has shown some decent development across portions of North Central Texas and Southwestern Oklahoma.
Late Monday evening more thunderstorms developed over portions of southeastern Virginia. The environment was marginal; ~600 J/kg of MUCAPE and ~35 kts of effective bulk shear. By 0216 UTC 6 May 2014, the developing storm near Sussex county, VA had moderate MRMS MESH values (~0.85 inches). The moderate MESH and environmental fields yielded a 31% ProbSevere probability (Figure 1). Four minutes later, at 0220 UTC, the GOES scan from 0215 UTC captured both strong normalized vertical growth rate and strong glaciation rates associated with this storm, which caused the ProbSevere output to jump to 79%–demonstrating the value of satellite growth rates! The probabilities continued to increase as the radar storm intensity increased. The NWS issued a severe thunderstorm warning for this storm at 0241 UTC (21 minutes after the first ProbSevere value in excess of 50%) and first severe hail report was received at 0238 UTC (18 minutes after the first ProbSevere > 50%).
Figure 1. Animation of NOAA/CIMSS ProbSevere and MRMS composite reflectivity valid 0216 – 0242 6 May 2014.
Additionally there were storms further east from the storm described above. These storms also exhibited ProbSevere values in excess of 50%. These storms had similar GOES inferred vertical growth rates and glaciation rates and similar MRMS MESH values, however these storms were not warned nor were associated with severe reports. This example demonstrates key points made in the NOAA/CIMSS ProbSevere training–the model skill is comparable, yet lower than an experienced forecaster, but often provides additional lead-time. One also wonders given the late evening timing of these additional storms if severe hail occurred but was never reported.
The NOAA/CIMSS ProbSevere Model is a statistical model that combines, NWP environmental conditions, satellite tracking and growth rates, and radar tracking and storm intensity to predict a developing storm will first produce severe weather in the near future (0-60 min). The NOAA/CIMSS ProbSevere model is being demonstrated to National Weather Service forecasters at 2014 Big Spring Experiment at the Hazardous Weather Testbed. NWS forecasters are encouraged to evaluate the ProbSevere model output as a means to increase confidence and lead-time in issuing initial severe warnings on developing storms. Today the weather is rather quiet across much of the country. A strong storm developed in southwestern Virgina and the evolution of the ProbSevere model performance on this storm is highlighted below. Of note, this storm achieves high probabilities and as of 2030 UTC, has been the only storm over CONUS today to achieve appreciable ProbSevere values.
Figure 1. The NOAA/CIMSS ProbSevere at 1950 UTC 5 May 2014–the first time the model produced probabilities above 50%. Radar data is not yet seen in this display as a domain switch was in-progress within HWT.
The NWS office in Blacksburg, VA issued a severe thunderstorm warning at 1952 UTC. The following two figures show the continued increase of NOAA/CIMSS ProbSevere probabilities.
Figure 2. NOAA/CIMSS ProbSevere at 1954 UTC 5 May 2014–probabilities are now 80%+.
Figure 3. NOAA/CIMSS ProbSevere at 2006 UTC 5 May 2014 now exceeds 90%.
This storm went on to produce severe hail (1.00″) at 2009 UTC–19 minutes after the initial ProbSevere > 50%. While this storm was not located in the a CWA of focus for this afternoon, forecasters are using this storm to gain confidence in the ProbSevere model and ask questions about some of the technical details of the model.
On Sunday April 27th, 2014, a moderate risk for severe thunderstorms was forecast by the SPC across eastern Kansas and Oklahoma, most of Missouri and Arkansas, as well as portions of Texas, Louisiana, Mississippi, and Tennessee. Before 12Z on that morning, convection developed along the dryline/Pacific cold front across central Kansas and Oklahoma. These thunderstorms became elevated as they moved east northeast towards Kansas City and western Missouri due to mid to lower tropospheric warm air advection. The early morning convection added uncertainty to the forecast as there was a question whether the atmosphere would recover and destabilize for convective initiation (CI) before the dryline and Pacific cold front moved through across eastern KS and western MO behind the morning convection.
Visible satellite image from GOES-13, METARs, NWS Convective Watches and Warnings, and manually analized surface features valid 1815 UTC 27 April 2014.
In the image above at 1815 UTC, although there was clearing right ahead of the dryline, extensive cloud cover and cool temperatures were present in the wake of the morning convection across eastern Kansas and western Missouri. Ninety minutes later, at 1945 UTC, temperatures started to rebound into the lower to mid 70s across southeastern Kansas and a cumulus field developed ahead of the dryline across eastern Oklahoma.
Visible satellite image from GOES-13, METARs, NWS Convective Watches and Warnings, and manually analized surface features valid 1945 UTC 27 April 2014.
One of the GOES-R future capability products, Probability of Convective Initiation, provides probabilistic 0-1 h forecasts of cloud objects achieving convective initiation (35 dBZ or > radar echo). Inputs into the algorithm are convective cloud properties from either GOES-13/GOES-15 and 20 Rapid Refresh model output fields. This fused product is excellent at providing guidance to the mesoanalyst or warning forecaster on which portion of a cumulus field will develop into convection. In the image below at 1945 UTC, there were low probabilities of CI within the newly developed cumulus field across northeastern Oklahoma.
Visible satellite image from GOES-13, GOES-R Probability of Convective Initiation [%], and NWS COnvective Watches and Warnings valid 1945 UTC 27 April 2014.
At 2045 UTC, 60 minutes later, the cumulus field developed northward into southeastern Kansas and probabilities increased to 60% within that cumulus field. This indicates a higher likelihood of CI and is starting to provide confidence to the forecaster that the atmosphere is destabilizing enough for the potential of CI.
Visible satellite image from GOES-13, GOES-R Probability of Convective Initiation [%], and NWS COnvective Watches and Warnings valid 2045 UTC 27 April 2014.
Another GOES-R future capability is the Convective Cloud-Top Cooling (CTC) product. This algorithm uses GOES-13/GOES-15 imager data and cloud phase information to provide situational awareness on which convective cloud objects are quickly growing vertically (i.e., a proxy for initial updraft strength). A stronger CTC rate is directly correlated with updraft strength and larger hail when compared to the WSR-88D maximum expected hail size algorithm. At 2125 UTC, 40 minutes after the elevated CI probabilities, two thunderstorms were developing across southeastern Kansas and northeastern Oklahoma. This is evident by the strong CTC detections in those areas.
Visible satellite image from GOES-13, GOES-R Convective Cloud-Top Cooling [K per 15 minutes], and NWS COnvective Watches and Warnings valid 2045 UTC 27 April 2014.
Fifteen minutes later, the first severe thunderstorm warning was issued on the thunderstorm with the northern CTC detection in southeastern Kansas. These two thunderstorms eventually developed into the supercells that produced the two EF2 tornadoes that affected Baxter Springs, KS (southern CTC detection) and Bates/Linn County, KS (northern CTC detection). Although the two products shown are GOES-R future capabilities, they are used today with GOES-13/GOES-15 data to demonstrate the algorithms and provide observational information of the convective environment that is not currently available with other datasets. When these two products are used together with observations forecasters are already comfortable with, situational awareness of the convective environment is elevated and more accurate short-term forecasts of convection may possibly be provided to partners.
~ Chad Gravelle, NWS Operations Proving Ground Satellite Liaison
Today marks the first day of the 2014 Hazardous Weather Testbed (HWT) Spring Experiment, which will run through the first week in June (Experimental Warning Program (EWP) Big Spring Experiment will be off for Memorial Day week). The HWT provides the GOES-R Proving Ground with an opportunity to demonstrate baseline and future capabilities products associated with the next generation GOES-R geostationary satellite system that have the potential to improve short-range hazardous weather nowcasting and forecasting. The availability of GOES-R products will demonstrate, pre-launch, a portion of the full observing capability of the GOES-R system, subject to the constraints of existing data sources to emulate the satellite sensors.
GOES-R products being evaluated this year include: Synthetic Satellite Imagery, NearCast System, Convective Initiation, Probability of Severe Model, Overshooting Top Detection, PGLM Total Lightning products, Tracking Tool and Lightning Jump Algorithm. These products will be demonstrated by National Weather Service forecasters and broadcast meteorologists within a simulated warning operations environment using a real-time AWIPS-II framework within the HWT EWP. When and if appropriate and if time permits, several of these products may be demonstrated informally in the HWT Experimental Forecast Program (EFP) during the mid-afternoon when the EFP severe storm forecasts are being updated.
As in previous years, participants, in addition to myself and visiting scientists, will make frequent posts to the following blog regarding the days activities, product performance, product feedback, etc.: http://www.goesrhwt.blogspot.com/
The storms that ravaged the southern United States this past week not only produced deadly severe weather, but also incredible flooding. Figure 1 shows parts of the Florida Panhandle and southern Alabama received in excess of 10 inches of rain on Tuesday, April 29 alone!
Figure 1: April 29 12Z to April 30 12Z precipitation analysis. More negative values indicate stronger OT’s
A previous blog post introduced the Overshooting Top Detection product and explained its utility in severe weather situations. Overshooting tops are also indicators of where heavy rainfall may be occurring. Furthermore, the constant presence of overshooting tops over a particular location over an extended period of time may indicate a prolonged period of heavy rainfall, which could lead to flooding.
The animation in Figure 2 shows GOES-East IR imagery with overshooting top detection’s overlaid from the afternoon of the April 29 into the early morning hours of the April 30. During much of this period, GOES-East was in Rapid Scan Mode, meaning images were often available every 5-10 minutes (instead of 15). Notice the persistence of overshooting tops centered over the Mobile area throughout the period, where copious amounts of rainfall were recorded. By about 09Z, a downward trend in overshooting top detection’s had begun as the storm system shifted eastward and weakened. The Overshooting Top Detection product provides a day/night capability for forecasters to easily identify where within a convective system the strongest updrafts are occurring, and where severe weather and/or heavy rainfall may be occurring given other meteorological factors.
Figure 2: April 29 22Z – April 30 10Z GOES-East IR with Overshooting Top Magnitude overlaid.
Figure 3 shows this same system during the early morning hours of April 30 at much higher resolution. This is a 375 m IR image taken with the Suomi NPP VIIRS instrument. Notice the visibility of features that aren’t easily seen in current GOES IR imagery such as gravity waves and overshooting tops.