PGLM products in AWIPS

Forecasters are now examining the PGLM products within their AWIPS workstations during warning operations with several storms over SW OK. The forecasters have created a 4-panel display that contains the 3 current PGLM products we are providing within the EWP, as well as the traditional radar reflectivity and NLDN detections (see image above). The forecasters currently have access to an instantaneous flash extent density (top right corner), a 60 minute sum (bottom left corner) and a 60 minute max track (bottom right corner). We are also working on providing the forecasters with a flash initiation density product, which will hopefully be available next week.

First CI of the day...

So unlike yesterday, activity has not waited and is currently ongoing and the EWP has began warning operations over the OUN CWA where a tornado watch has been issued at 1803 UTC. Forecasters were monitoring the GOES-R CI products throughout the morning following the issuance of their initial AFD. A line of cumulus formed along a dryline extending through western OK down into TX. At 1702 UTC, UWCI flagged a towering cumulus cloud for "pre-CI growth" (see image below) near the TX/OK border. SATCAST did not flag this storm and at this time radar reflectivity had already exceeded 35 dBZ for this particular storm (see image below). It is likely that the storm initiated between the 15 minute scan period from 1645 UTC and 1702 UTC. UWCI continued to flag this storm during the 1715 UTC scan for "CI ongoing". Also at 1715 UTC, SATCAST flagged a storm just north of the initial storm which continued to form a 35 dBZ echo at 1754 UTC, providing a 39 minute lead time on radar. The southern initial storm had it's first CG flash at 1800 UTC, providing about an hour lead time for the first occurrence of CG for the UWCI.

4-panel CI display (top) and radar reflectivity (bottom) for 1702 and 1707 UTC respectively.
4-panel CI display for 1715 UTC
Unfortunately, as it always seems, we lose satellite data for a 30 minute gap during initiation period, so the EWP forecasters have moved into warning operations using radar data as storms continue to develop northward along the dryline. Hopefully at some point this afternoon/evening we will get a chance to examine the overshooting-top and thermal couplet products in warning operations, as well as the PGLM as the storms move a little further east into the OKLMA domain.

Simulated satellite imagery

The Convective Initiation group compared simulated ABI and current GOES water vapor imagery to examine the dry line over Texas and western Oklahoma and to evaluate the performance of the NSSL-WRF through early morning. The presence of lee waves downstream of the mountains in extreme southwest Texas was noted on the imagery. Overall, the NSSL-WRF accurately depicted most of the large-scale features, with the usual differences in the locations of individual cloud and convective features. The simulated imagery indicates that convective initiation will occur later in the day along the dry line from south-central Kansas to north Texas. The dry line was also evident in the mid-level (band 9) water vapor imagery prior to convective initiation.

-Jason Otkin

Oklahoma CI 18 May 2011

Forecaster comments reposted from EWP blog...

"2330Z Update: UAH CI indicating possible development across Garvin and McClain counties. There does appear to be an HCR in this location but we believe that the edge of thin cirrus clouds are causing algorithm to detect cloud cooling that is not real. Looks like we are going to move operational area to eastern Colorado where there is a lone supercell. kbrown

2230Z Update: Got a hit off of the UAH CI over southern Comanche county where there are weak echoes aloft (15-20dbz). These echoes are associated with a wildfire plume, however.

2140Z: Brief UAH CI detected across north-central Oklahoma in cloud streets (1832Z), but no subsequent echoes were detected.

Although there has been several hours of cu/tcu formation near and east of dryline, CI algorithms have not detected CI across OUN domain. From 20Z to 2115Z this is actually a good thing since no echoes have developed (good case of low false alarm). Before the CIMSS CI became contaminated with Ice Cloud Mask, there were a few hits for CI on leading edge of incoming cirrus across northwest Oklahoma between 19Z to 20Z. No echoes were subsequently detected.

Isolated CI hits from the UAH algorithm did accurately depict some elevated echoes over the OK/AR border but no lightning occurred. Leadtime for echoes was 15 to 30 minutes."

kbrown

A Visit By Don Berchoff

Reposted from EWP blog...

"As we waited agonizingly for convection to initiate in our Central Oklahoma Domain (and as of this writing, we are still without storms), we were paid a visit by Don Berchoff, Director of the NWS OS&T."



"Our discussion with Mr. Berchoff included:

- the need to continue improvements in Convective Initiation (both sensor and model based)
- using multi-sensor blending for more holistic storm interrogation
- his thoughts on the coming of AWIPS2 and the utility for extensibility
- the need for improved analysis and visualization tools in light of the 12 fold increase in data flow

“Fortunately” the weather “cooperated” with us and everyone in the room was able to engage in the discussion."

-K. Manross :: Week 2 EWP Weekly Coordinator

Potentially missed CI target... but with a silver lining

As we have seen over the past couple years during the Spring Experiment, the bane of the satellite community's existence has struck again. It's 2100 UTC and that means we have another 30 minute gap in satellite data right during initiation period... so any updates in the imagery and the CI products will have to wait until after 2115 UTC, and our last image was at 2045 UTC. During that period the cirrus is likely to move over the area, inhibiting any of the CI products from detecting the clouds underneath... We will continue to monitor the area, however it seems like we may have just missed our target.

However, there is a silver lining to this event. Forecasters have been watching the CI products constantly over the past couple hours and have noticed no false alarms over the area of developing cu field for either of the CI products. There were a few false alarms with the cirrus overrunning the surface causing false cooling with the UWCI, as well as a couple false alarms with some of the stratus area in eastern CO with the SATCAST (UAH CI) product. The forecasters mentioned that the fact that these CI products were showing no signals over the cu field of interest provided them with increased confidence that CI was NOT occurring. Increased confidence is a good thing, whether it leads to exciting weather or not.

Monitoring CI over OK...

EWP NWS forecasters have begun real-time operations within the HWT. We are focusing on the UWCI, UAH CI, and Nearcast products over the domain waiting for convection to develop along a dryline extending from central OK down across the Red River. An area of cumulus clouds have begun to form along the dryline and we are waiting for further development. An area of thin cirrus (our worst enemy) is moving in rapidly from NM and will eventually impede upon our ability to detect CI from the satellite based products (see image below). We are hoping that CI will occur prior to the cirrus entering the area. Below is an example of what we are looking at, with the UWCI and ice cloud mask overlaid on visible satellite imagery. A false alarm has occurred due to the thin cirrus being misclassified by the cloud typing algorithm and created a false cooling event as it passed over land.


The EFP is also monitoring CI on the CI desk and have been looking at the Nearcast product extensively over the area. They have noticed that the Nearcast has been showing a band of moisture and increased instability potential aligning itself with the dryline and then being advected quickly east. They are concerned that if this area of moisture and instability moves too far east before CI occurs that the potential for any severe convection will be limited. I explained to them that the RUC winds may not be truly representative of what will really happen and to monitor the new forecast that came in at the top of the hour to see if this trend continues.

EWP daily briefing... 18 May

During today's EWP forecaster daily briefing, we discussed the previous day's activities over DC and then over eastern CO later on in the evening around 2345 UTC. When asked about how the CI products performed during the early event, the forecasters offered the following comments...

"It was complex and challenging to use in this sort of widespread event. We primarily used the UAH CI product yesterday. It was generally pretty noisy in this kind of pattern, but I did see some clusters and bands with some mixed success. In some cases it popped up where there was clearly some convection ongoing, but there where definitely cases where it did show some good signals. I would like to see it more in some cases like this that are difficult."

"I was watching it today over the same area and it gave 45 minutes to 1 hour lead time over lightning."

For today we have decided to operate over western OK to capture some clean slate CI as well as give us a chance to examine some of the other experimental products, such as the PGLM, which are limited to specific domains... one of which being over Oklahoma.

SATCAST CI Nowcasting: Translation & Evolution

One substantial improvement to increase the utility of the SATCAST CI algorithm (the GOES-R Proxy CI Algorithm) will be including details on where a given object, forecasted to become a new convective storm, is moving. For a forecaster, a present limitation of this CI nowcast product is that there remains uncertainty on where a nowcasted cumulus cloud object will in fact be located when it produces ~35 dBZ intensity rainfall (or perhaps lightning, if the environment is conducive). The example shown illustrates how adding propagation information will help in the CI nowcasting process, especially since the cloud object evolves substantially as CI occurs and thunderstorm development proceeds.

Shown is an example between 1432 and 1645 UTC 17 May 2011. The synoptic setting was a well-defined upper low, with strong cyclonic rotation and cold air aloft, leading to scattered generally low-topped, shallow convective storms. RUC surface-based CAPE values were 500-700 Jkg^-1. Convective clouds were propagating northwestward from central North Carolina to south-central Virginia. Circled is a northwest-southeast oriented cumulus cloud line in the vicinity of Raleigh, NC at 1432 UTC (Fig. A), with an added propagation vector included. Figures B (1445 UTC) and C (1602 UTC) show where the developing cumulus clouds and new storms, respectively, had moved since 1432 UTC. CI occurred by 1515 UTC; the 1606 UTC radar image is also shown. The cumulonimbus cloud with an obvious anvil is seen extending into far south-central Virginia at 1645 UTC (circled in Fig. D). In Fig. D, the full translation vector is shown beginning where the first object in Fig. A was seen. Only a few lightning flashes were observed by 1645 UTC associated with this storm. In this example, the translation vector points in an unusual direction, with storm propagation toward the northwest. This translation vector will also help forecasters maintain identification of a developing cumulus cloud, which will change substantially (in terms of size and shape) as it evolves, leading to less uncertainty in a CI nowcast.

Future enhancements to SATCAST will include translation vectors per nowcasted objects designed to help forecasters more precisely predict the location where heavy (~35 or greater dBZ) rainfall will occur. In many cases, the distance (time) between a highlighted CI object can be several 10's of kilometers (30-60 min) from where the rainfall/thunderstorm is actually observed to occur.

CI over VA

Posted by EWP forecaster and taken from EWP blog...


"Widespread cumulus fields forming across central eastern VA and UAHCI products capturing this much better than CIMSS in this case, which is a very moist, weakly unstable, but uncapped environment (almost tropical). While most new CI IDs are very scattered to isold in nature, now beginning to see some banding or clustering, which actually matches very well with at least one 4km WRF (12Z run for SPC…see second image above), and will be watching to see if the IDing of this banding of CI zones within otherwise wideapread cu field helps to identify where stronger storms could soon be forming in this kind of environment."

Steve Keighton