Wednesday, May 11, 2011

I know it when I see it and other discussions

We had many discussions over the last two days. One was regarding CI definitions. Of the variety of opinions we heard, a storm was defined by:
1. Whether lightning occurred,
2. a coherent, continuous thunderstorm that eventually reached a significant low level reflectivity threshold (40-45 dBz) within 30 minutes,
3. any combination of 1 or 2, which also produced severe weather (e.g. it was just a storm, but a severe storm)

These variations on the theme are exactly what we were considering for the experiment, the experimental algorithms from the model, and the forecast verification we had played with prior to the experiment.

Another conversation involved what forecasters would use from the experimental suite of variables. The variables would need to be robust, easy to interpret (e.g. quick to interpret and understand), and clear. This is a tough sell from the research side of things, but it is totally understandable from a forecaster perspective. Forecasters have limited time in an environment of data overload in which to extract (or mine) information from the various models. They have very specific goals too, from nowcasting (e.g. 1-2 hours; especially on days like today where models miss a significant component of what is currently happening), to forecasting (6-24 hours), to long lead forecasting 1-8 days.

We also spent some time discussing the Tuesday short wave trough in terms of satellite data and radiosonde data.  I argued that many people believe that satellite data and its assimilation is much more important now (Data volume, coverage, and quality control) than radiosondes. It was mentioned that radiosondes are very important on the mesoscale especially in the 0-24 hour possibly 48 hour forecasts. Still more opinions were expressed that some forecasters have questioned the need for twice a day soundings. Opinions in the HWT ranged from soundings are important, to soundings should be launched more often, to sounding should be launched more often at different times. It is plausible that some of our NWP difficulty may be due to launching soundings at transition times of the boundary layer.

I am of the opinion that if model suites are launched 4 times per day that soundings should be launched at least 4 times per day, especially now where cycling data assimilation is common practice. This would return our field to the 1950's era where 4 times per day soundings were launched at 3,9, 15, and 21 UTC.

Lastly, we discussed the issue of what happens when a portion of the forecast domain is totally out to lunch? Like today where the NM convection was not represented. I think I will talk about that tomorrow once we verify our forecast in the OK area from today. Stay Tuned.

It's complicated

As expected, it was quite a challenge to pick domains for days 2 and 3. Day 2 was characterized by 3 potential areas of CI: Ohio to South Carolina, Minnesota and Iowa, and Texas. We were trying to determine how to deal with pre-existing convection: whether it was in our domain already or would be in our domain during our assumed CI time. As a result, we determined that the Ohio to South Carolina domain was not going to be as clean-slate as Texas or Minnesota. So we voted out SC.

We were left with Texas (presumed dryline CI) and Minnesota (presumed warm front/occlusion zone). Texas was voted in first but we ended up making the MN forecast in the afternoon. Data for this day did not flow freely, so we used whatever was available (NSSL-WRF, operational models, etc).

The complication for TX was an un-initialized short wave trough emanating from the subtropical jet across Mexico and moving northward. This feature was contributing to a north to south band of precipitation  and eventually triggered a storm in central and eastern OK, well to the east of our domain. The NSSL WRF did not produce the short wave trough and thus evolved eastern TX much differently than what actually occurred despite having the subtropical jet in that area.  So we were gutsy in picking this domain despite this short wave passing through our area. We were still thinking that the dryline could fire later on but once we completed our spatial confidence forecast (a bunch of 30 percents and one 10 percent) and our timing confidence (~+/- 90 minutes) it was apparent we were not very confident.

This was an acceptable challenge as we slowly began to assemble our spatial forecast, settling on a 3 hour period in which we restrict ourselves to worrying only about new, fresh convection by spatially identifying regions within our domain where convection is already present. This way we don't have to worry about secondary convection directly related to pre-existing convection. We also decided that every forecaster would enter a spot on the map where they thought the first storms would develop (within 25 miles of their point). This makes the forecast fun and competitive and gets everyone thinking not just about a general forecast but about the scenario (or scenarios if there are multiple in the domain).

The next stop on this days adventure was MN/IA/Dakotas. This was challenging for multiple reasons:
1. The short wave trough moving north into OK/KS and its associated short wave ridge moving north northeast
2. the dryline and cold front to the west of MN/IA,
3. the cold upper low in the Dakotas moving east north east.

The focus was clear and the domain was to be RWF. This time we used a bigger domain in acknowledgement of the complex scenario that could unfold. You had the model initiating convection along the warm front, along the cold front in NE on a secondary moisture surge associated with the short wave trough, and a persistent signal of CI over Lake Superior (which we ignored).

We ended up drawing a rather large slight risk extending down into IA and NE from the main lobe in MN with a moderate area extending from south central MN into northern IA. After viewing multiple new products including simulated satellite imagery (water vapor and band differencing from the NSSL WRF and the Nearcast moisture and equivalent potential temperature difference, it was decided that CI was probably with everyone going above 50 percent confidence.

In Minnesota we did quite well, both by showing a gap near Omaha where the moist surge was expected but did not materialize until after our 0-3 UTC time period. Once the moisture arrived ... CI. In MN CI began just prior to 23 UTC encompassing some of our moderate risk even down into IA, yet these "Storms" in IA were part of the CI episode but would not be objectively classified as storms from a reflectivity and lifetime perspective, but they did produce lightning.

The verification for Texas was quite bad. Convection formed to the east early, and to the west much later than anticipated associated with a southern moisture surge into NM from the upper level low migrating into the area nearly 11 hours after our forecast period start.

As it turns out, we awoke this morning to a moderate risk area in OK, but the NM convection was totally missed by the majority of model guidance! The dryline was in Texas still but now this convection was moving toward our CDS centerpoint and we hoped that the convection would move east. A review of the ensemble indicated some members had some weak signals of this convection, but it became obvious that it was not the same. We did key in on the fact that despite the missed convection in the TX panhandle the models were persistent in secondary initiation despite the now-developing convection in southern TX. We outlooked the area around western OK and parts of TX.

In the afternoon, we looked in more detail at the simulated satellite imagery, nearcast, and the CIRA CI algorithm for an area in and around Indiana. This was by far the most complicated and intellectually stimulating area. We analyzed the ensemble control member for some new variables that we output near the boundary layer top (1.2 km AGL roughly): WDT: the number of time steps in the last hour where w exceeded 0.25 m/s and convergence . We could see some obvious boundaries as observed, with a unique perspective on warm sector open celled convection.

In addition we used the 3 hour probabilities of CI that have been developed specifically for CI since these match our chosen 3 hour time periods. We have noticed significant areal coverage from the ensemble probabilities which heavily weight the pre-existing convection CI points. Thus it has been difficult to assign the actual new CI probabilities since we cant distinguish the probability fields if two close proximity CI events are in the area around where we wish to forecast. That being said, we have found them useful in these messy situations. We await a clean day to see how much a difference that makes.

Monday, May 09, 2011

Day 1 in the 2011 HWT EFP

What a great start to the HWT. There were troubles, and troubleshooters. We had plenty of forecasters and plenty of forecast problems. All in all it was quite a challenge.

The convection initiation (CI) team had some great discussion on the CI definition including all the ways in which CI gets complicated. For example, visually we can identify individual clouds, or cloud areas on satellite. When using radar, we might select areas of high reflectivity that last for say 30 minutes. In the NWP models, we rely on quantitative values at a single grid point at two instances in time.

We also have the issue of whether CI is part of a larger episode (close in space and/or time by other storms) or developing as a direct result of previous convection (ahead of a squall line). In these relative cases, visually identifying new storms might be easily accomplished, but in the model atmosphere (in a grid point centric algorithm) new CI points may be all over the place, say as gravity waves or outflow achieve just enough separation to be classified as new (thus CI) even though it might simple be redevelopment. From a probability standpoint, spatial probabilities of CI may thus be larger around existing convection. Does this enhanced probability, ahead of the line, signal actual new storm development?

Trying to establish an apples to apples comparison between model and human forecasts of such discrete events is a major challenge. We are testing 3 model definitions of CI to see their viability from the perspective of forecasters, and we will also evaluate object based approaches to CI.

Of course we cannot talk about where CI might be without talking about when! When will the first storm form? This gets back to your definition of CI. Should the storm produce lightning to be classified a storm? How about reaching a threshold reflectivity? How about requiring it that it last a certain amount of time? The standard definition of storms relies on its mode (ordinary, multicell, supercell); all having a unique evolution with the placement of the updraft and precipitation fall out. But what about storm intensity (however you define it)?

I should also acknowledge that defining all of this can be quite subjective and is relevant to individual users of a CI forecast. So we are definition dependent, but most people know it when they see it. Lets consider two viewpoints: The severe storm forecaster and an aviation forecaster. The severe storm forecaster wants to know about where and when a storm may form so they can decide the potential threat thus leading to a product (mesoscale discussion for specific hail, wind, tornado threats) provided that storm or CI episode is long lived. The aviation forecaster might be concerned with the sudden appearance of cumulonimbus which could pose an immediate threat to aircraft. But they are also concerned with the resulting coverage of new storms (diverting traffic, shutting down airports, planning new traffic routes or patterns) and the motion, expansion, and decay of existing storms.

And lastly it will be important for us to establish what skill the models and forecasters have with respect to CI. This is not a new area of study, but it is one where lots of complexity, vagaries of definitions, and also a lack of understanding contribute to making this one of the greatest forecast challenges.

As we refine what our forecast will consist of, we will report back on how our forecast product evolved. The more we forecast, the more we learn.