I share some of Steve Keighton’s comments on structure/logistics of the EWP. Here are a few more general comments: I think collaboration with the EFP could be very useful, I just did not find it to be the one day that I was not on the full-blown warning shift. I also would find it useful to either a) Not be completely focused on the timeliness of warnings, and more on the analysis of the products, (b) Possibly having certain forecasters focus on certain products at various times in the warning process. That said, it is good to see what products came to the forefront and which ones maybe were not (and maybe should have!), and to truly mimic a warning environment,. Having grown accustomed to 3 D2D screens in a warning environment, I did not use quite as many products as I would have in a typical environment, and was also creating procedures on the fly, which I would not normally do in a warning situation. I would be really interested to see the effects of Dual Polarization has on screen real estate, and if it would affect the experimental products chosen for use in the warming decision/awareness process. I think we all tried to strike a balance between attempting to simulate and get warnings out in “real” time vs. giving each new tool more attention.

Below are a few key points about specific products from my experience. These comments have been shared with the staff at WFO Portland, and will be shared with all Western Region SOOs.

* The OUN WRF was a “hot model”, in that it also appeared to break the cap too quickly. That said, it did clue us in to potential scenarios, and seemed to have a decent grasp on convective mode expected, given convective initiation. It would be nice to have this run in an ensemble mode as well, as it basically served as another tool to compare to EMC WRF, NSSL WRF, and the RUC HRRR models.

* The 3DVAR Multi-Radar Real Time Data Assimilation Products, while fairly new, hold alot of promise. The products that I found useful included the Updraft (instantaneous and track) and the Rotation products. The Updraft product, while sometimes misplaced, was still very valuable. The downdraft products were a bit noisy, not only in multicell situations, but even in supercell cases. These products aided in assessing whether updrafts were strengthening or weakening, and the rotation track also aided in more accurately following storm motion. While I did not rely on these products, and did always confirm what I was seeing with my traditional radar analysis, it did provide a clearer situational awareness picture, and it was fairly easy to create procedures to integrate into the warning decision process.

* Convective Initiation tools, while useful, are still contaminated by cirrus. They did provide some lead time before seeing significant radar echoes, though lead time in rapidly developing cases was not much more. I did find, once radar operations got more intense, that I did ignore these products a bit more than I wanted to, but it was more a workload issue. Also, on the days where convective initiation did not occur, these tools provided a correct NULL result. I did like the CIMSS product a bit better, as it provided CI likely, CI occurring, CI possible vs. a YES/NO from UAH. The UAH algorithm was too sensitive, and the CIMSS product seemed not sensitive enough. This is intended as the CIMSS product is going for a low False Alarm Rate, and UAH is going for high Probability of Detection.

* We got to look at multi-cellular and more borderline cases as well, which was very useful instead of focusing on “supercell” cases. The products didn’t do too bad, though seemed to perform best in more traditional supercellular cases.

* The pseudo-GLM was very useful in that it focused attention on storm intensification, and was able to pick up on flash rates much earlier than the CG network. Though I did not see this for many borderline cases, this would be useful when the forecaster is not sure whether particular areas are electrified or not, particularly when not seeing any CG strikes.

* I relied heavily on the Multi-radar, multi-sensor products when in a warning environment. I wouldn’t say they made the warning decision for me, but they served as a great situational awareness tool as to where to focus my attention. Of particular use were the -10C and -20 C Reflectivity products, but my favorites were the 50 dbz height above/below a user specified level (such as the -20 C level), MESH (Multi-radar estimated hail size, both instantaneous and track). These products were great for estimating hail size when combined with radar tools. These products are available (they are adding the Western domain soon, if not already) on wdssii.nssl.noaa.gov as KML files, or on a neat google map in
development wdssii.nssl.noaa.gov/maps.

Thank you again for allowing me to participate in the EWP this year. This was my first experience in an experimental warning program, and I hope to leverage this experience to spread the word on the promise on new warning decision tools, and to hopefully participate in another Experimental Warning Program in the future.

Kevin Donofrio (General Forecaster, NWS Portland OR – EWP2011 Week 2 Participant)

* Overall, another very valuable experience for me, and an important program to get feedback from forecasters during the development process. Thanks for the opportunity!

* A general suggestion is to either require participants to review materials on the products ahead of time (through recorded presentations), and then have a much shorter period the first day for SMEs to summarize and answer questions, thus allowing more time on the first day to get more familiar with products and procedures during a WES….OR, more ideally, have participants come for a two week period to get especially comfortable with the background on the products, and more time spent really working with the various products, and perhaps more opportunity for concentrated time spend really evaluating one product at a time initially. I realize though this would reduce the number of overall participants, and that’s not good either.  So just something to consider perhaps. I felt that there just wasn’t quite enough time to really get to know and work with each of the products, and we were just getting there on the last day.  A more focused effort on each product before given the flexibility to use any or all to help make warning decisions or raise S.A. would be better, since I felt I was trying to make sure I frequently tried to look at all of them, while at the same time trying to keep up with warning decisions using traditional products/methods too.

* Another general suggestion is to consider different interactions with the EFP side than what we did.  Felt it was generally a waste of time to work with the CI group that one day I did that, and would prefer to work with the severe wx desk and get a sense for the various models and high res ensembles they were evaluating. We got to evaluate the OUN WRF, but that’s just one potential solution.

* I was most impressed with the promise of the 3DVAR products in terms of a little more of a complete picture of storm structure/intensity (4D really with tracks of max updraft and rotation very helpful for trends to stand out).  These were particularly helpful in combo with traditional radar products, especially with 5 minute updates, but latency is the one concern. Not likely to make a warning decision based solely on these, but certainly adds confidence to decisions being considered with traditional (usually base radar) products.  Obviously with the future potential to advect this in the future from RUC-based forecasts will add another dimension to it’s utility.  Specific comments on the various products available were made via the survey forms, but here will just mention the need to improve the downdraft product (focus on lower levels), perhaps updraft too (various levels), an updraft helicity product, and then access in AWIPS to the 2D winds would be a great benefit.

* I was a little discouraged with the CI products, but do see the potential, but would like to have spent more time with these in different environments. My impression is that these will be most helpful for otherwise clear conditions with CI expected on a boundary like the dry-line, and much less useful in moist “airmass” type environments, and obviously when there is cirrus or present (these last two scenarios are fairly common here in the mid-Atlantic region). A little concerned about two different efforts/groups developing different CI products, and each seemed to have some advantages, in my limited experience that one one week I was more impressed with the UAH version, (with more detections), but need the multi-tier output I think like the UW-CIMSS version.

* OUN WRF (as other high res convective resolving models have shown) certainly has a lot to offer in terms of helping to anticipate storm mode and to some degree evolution, but not necessarily the locations or timing (timing probably worst).  Still need to have a collection or ensemble of these to get a better feel for the ranges of timing and locations, but looking at the details of a single model and it’s trends can still provide some helpful info for overall S.A., and ultimately leads to quicker decisions if you are ready to anticipate certain structures/evolutions. Despite some errors in timing and placement on that last active day we were there (May 19), the signals were good enough to help prepare for some upscale evolutions during the evening, and did suggest some early convection initiation (which some other models did not have).

* Finally, love the new wdsii-nssl map web page for the MRMS products (even though we weren’t specifically evaluating those). I used these frequently in AWIPS, with some of the same advantages of the 3DVAR products such as trends, but the rapid update for these since they are multi-radar is a uniqe advantage.  I’m more easily to introduce these products to my staff using the web page (http://wdsii.nssl.noaa.gov/maps), but would also like to work with ER SSD to get these into AWIPS via LDM (when I get a chance!).

* I just realized I totally left out any comments on the PGLM data!!  IN part I think since I really did not get a chance to spend much time evaluating it. In one case, one of the key ground sensors actually dropped out and it gave the impression that the flash density dropped when it really didn’t. Also, for this event the NLDN CG data was intermittent so not a good chance to compare the two.  Again, I think with this data set it would be especially important to focus solely on this product for most of an afternoon/evening, or during a DRT case.  Still don’t have a real good feel for how the total lightning relates to the CG data we are used to tracking, and relationship to severe weather.  I think it would be very important to almost spend at least one day on this one product to get some worthwhile feedback. Sorry I don’t have more on this.

Steve Keighton (Science and Operations Officer, NWS Blacksburg VA – EWP2011 Week 2 Participant)