Wednesday, January 10, 2018

January 12-13 Significant Ohio Valley Winter Storm

A significant winter storm impacting the Ohio Valley is expected to lay down accumulating snowfall, potentially in excess in 12", as well as potentially crippling ice accretion.

This post will cover current model guidance, which appears to be forming a consensus as of the January 10 00z model suite.


Pivotal Weather - GFS
We begin with the GFS model's outlook for snow accumulation. The 12z GFS appears to anticipate a stripe of accumulating snow falling from western Tennessee and Kentucky into southern Indiana, much of Ohio and portions of West Virginia. Particularly heavy accumulations are forecasted for northwestern Pennsylvania and western New York, especially near the Great Lakes. Per this model forecast, accumulations look to be in the 2-4" range in western Tennessee and Kentucky, increasing to a broad 3-6" range for Indiana, especially in the southeast portion of the state. Ohio may receive anywhere from 6-12" of snow, with the higher end of this range most likely in the northeast part of the state.

Pivotal Weather - NAM
Next is the NAM's snowfall forecast. As is typical for this model, snowfall accumulations are enhanced, likely beyond what will actually occur. Western Tennessee and Kentucky are projected to receive 4-8" of snow in this solution, a notable bump from the GFS' guidance. Southeast Indiana and a good swath of Ohio are outlooked for a 12-18" snowstorm as per the NAM model, again a substantial increase from the GFS solution. I do not expect the NAM's solution to work out, partially because of its inherent bias of over-doing snowfall accumulations but also the possibility (if not the likelihood) that at least a portion of this projected snow is actually sleet for some areas. This sleet-forecasted-as-snow problem is present in all models for the winter season (and can vary depending on the website you use to get snowfall accumulation graphics), but is particularly maximized in the NAM model. As such, I would take the NAM snowfall graphic with a grain of salt at this point and refer to the GFS and CMC models, the latter of which is shown below.

Pivotal Weather - CMC
The CMC model seems to take a middle ground between the GFS and NAM in a broader sense. On one hand, the CMC anticipates a similar snowfall accumulation story for Indiana/Ohio/Pennsylvania/New York as the GFS, but also sides with the NAM in the idea that snowfall accumulations may be heavier in Tennessee and Kentucky. Adding to the variance is how the CMC's accumulations for Kentucky and Tennessee appear to be slightly west of other guidance. I personally would side with the GFS for accumulations in Kentucky and Tennessee - there will be a very warm air mass in place as this storm gains strength, and I'm not confident in the storm being able to cool off the temperature profiles enough to lay down over half a foot of snow in those areas. I do believe a broad 8-14" range of snow for southeast Indiana, much of Ohio, northwest Pennsylvania and western New York is preferable for now, with higher amounts likely in places nearer to the Great Lakes for Pennsylvania and New York.

Freezing Rain

Pivotal Weather - NAM

Pivotal Weather - GFS
I don't want to discuss freezing rain too much in large part because freezing rain is often a "nowcast" situation, or can only be accurately forecasted immediately prior to the event. I will, however, show the NAM and GFS forecasts for freezing rain to give an idea of where accumulating ice is most likely from this storm. I want to emphasize that, with the above graphics, you should *not* focus on ice accumulations, but *should* focus on the forecasted placement of ice accretion.

The NAM and GFS agree that an ice storm is possible for western portions of Tennessee and Kentucky into southeastern Indiana, as well as south and east Ohio. Further ice accretion is also a possibility for central Pennsylvania, as well as parts of central and eastern New York. The remarkable similarity between projected ice accretion and projected snowfall accumulation makes me more suspicious of the snowfall forecasts, particularly for Kentucky, Tennessee and southeast Indiana, as model guidance could be falsely interpreting freezing rain and/or sleet as snow. As I said, this will likely become a "nowcast" situation for that reason.

To Summarize:

- A significant winter storm is expected this weekend in Kentucky, Tennessee, Indiana, Ohio, Pennsylvania and New York.
- Snowfall amounts of 2-6" are possible in Kentucky and Tennessee, while a range of 6-12" is possible for southeast Indiana into western and central Ohio. Northeast Ohio and western New York look to receive 8-14" of snow by the time this event is finished.
- An ice storm does look possible from this system. While I am not confident in forecasting accumulations, it looks as if central Tennessee, western Kentucky, southeast Indiana, central Ohio, central Pennsylvania and central/eastern New York will be at risk for non-negligible ice accretion.
- The location and intensity of freezing rain will most likely only be accurately determined immediately prior to the event, if not determined during the event (i.e. "nowcasting").


January 23-27 Potentially Significant Winter Storm

A potentially significant winter storm looks to impact the United States around the January 23-27 period.

Tropical Tidbits
The GFS model forecasts a strong surface low crossing the central and northern parts of Japan on Wednesday, January 17th, producing strong lower-level winds offshore the eastern coast of the country as a result. As I've discussed in the past, the East Asian Rule can be applied here. Recall that the East Asian Rule is a leading indicator of sorts for weather in the United States, by a matter of 6-10 days. In this case, the presence of a storm system over Japan on January 17th suggests a storm threat in the United States sometime between January 23rd and January 27th.

This is a little far out to be anticipating a storm system in Japan, so let's see if any other model guidance agrees.
Tropical Tidbits
The Canadian's global weather model, abbreviated often as the CMC model (Canadian Meteorological Centre), agrees to some extent with this phenomenon. The model anticipates a trough to cross Japan at roughly the same time period as the GFS, albeit at a weaker magnitude. Such variations in strength are to be expected, and it's entirely plausible that the "potentially significant" tag will need to be dropped if/when this storm comes into view, both for Japan and then the U.S. For now, however, the key thing is that both models do show a storm system impacting Japan during this time period, a good signal for a storm in the U.S. around January 23-27.

Tropical Tidbits
The GFS ensembles, valid for the same time as the CMC model image, also agree on the presence of a storm system impacting Japan around January 17th. The above image shows the mean MSLP from the individual ensemble members through the black contour lines, and shows individual ensemble member lows in the red numbers. Higher pressure values are shown in the blue numbers. The color represents the normalized spread among members, and as the graphic indicates there is considerable spread / uncertainty around the idea of a storm hitting Japan. However, with the mean MSLP contours indicating the presence of the storm, it seems to confirm the ideas of the GFS and CMC above, that indicate a storm system is expected to hit Japan on January 17th.

So, we can now say it appears more likely than not that a storm system will hit Japan on January 17th. This would correlate to a storm system in the U.S. around January 23rd-27th. But we can go further. We can pinpoint where the storm may end up tracking, to at least some degree.

We'll begin by using the Madden-Julian Oscillation (MJO).
Dr. Roundy:
The screenshot above shows a Hovmoller forecast (from the bright green line upward) of outgoing long wave radiation (OLR) anomalies through early February. Outgoing long wave radiation is used to identify areas of enhanced or suppressed thunderstorm activity in the tropics - here, it analyzes such activity or lack thereof between the 7.5 degree North line of latitude and the 7.5 degree South line of latitude. Cooler colors indicate stormier weather, while warmer colors indicate the absence of thunderstorms. Given that the Madden-Julian Oscillation is identified by the placement of enhanced thunderstorm activity along the Equator, this helps us identify where the MJO may be.

In the above image we will focus on the solid red line, which outlines where the Madden-Julian Oscillation appears to be tracking in terms of longitude (on the x-axis) as based on OLR anomalies. We see that the MJO is currently shown to be maximized just west of the 90 degrees East line of longitude, as shown by the solid red contours. As we move forward in time to roughly January 21, however, that MJO "pulse" (which just so happens to be progressing with a Kelvin Wave, as the thin pink lines show) weakens and peters out, while another "pulse" fires up around the 135 degrees East line of longitude. Around January 25th, in the middle of this potential storm's expected timeframe, the deepest negative OLR anomalies look to be placed around the 140 degrees East line of longitude.

We've established that the MJO is expected to be active during this potential winter storm, and we know where this enhanced thunderstorm activity is expected to be, but we don't yet know what phase of the MJO this corresponds to. If you'll read on, however, you'll see that we can also figure that out!

The above image is a composite graphic from the Japanese Meteorological Agency (JMA), showing 200-millibar velocity potential anomalies across the globe during the month of January when the MJO is in Phase 5. The bottom panel shows OLR anomalies during the month of January when the MJO is in Phase 5. Put our earlier discussion and this image together, and it looks like the MJO is projected to be in Phase 5 when this winter storm is forecasted to impact the United States. We figured this out by seeing how the minima of the negative OLR anomalies were centered in the region around the longitude lines of 130 degrees East and 140 degrees East. We can now see the typical precipitation effects of when the MJO is in Phase 5 during the month of January.

The above image shows just that. It shows typical precipitation anomalies over the United States during the months of December-January-February (centered on January) when the MJO is in any of the phases above. Looking at the graphic for Phase 5, it seems that wetter than normal conditions tend to show themselves during the winter months in the Midwest, Ohio Valley, Central Plains and Southern Plains when the MJO enters Phase 5. Additionally, drier than normal conditions appear to present themselves in portions of the Southeast, as well as the northern Plains. Much drier than normal conditions are typical along portions of the West Coast when the MJO reaches Phase 5. We can derive a couple of things from those anomalies.

For one, the dry patch in the Southeast tells me that the (in)famous Southeast Ridge may make an appearance. This is not what winter weather fans along the East Coast like to see, as it discourages storms from running up the coast and forming into "Nor'easters". This is exactly what winter weather fans in the Midwest, Great Lakes and Ohio Valley like to see, however, as that Southeast Ridge can often help storm systems move northeast and bring heavy snows to the aforementioned regions. In this case, while we of course cannot say for certain yet, the presence of a Southeast Ridge would certainly boost the chances for this potential winter storm impacting the Midwest, Great Lakes and Ohio Valley with heavy snow as opposed to the East Coast. 

So far, we have established that a winter storm is quite possible in the January 23-27 timeframe for the United States, as model guidance seems to agree on a storm system impacting Japan on January 17th (albeit with varying magnitudes). We've also used forecasted OLR anomalies to show how the Madden-Julian Oscillation looks to be in Phase 5 around the time this winter storm would impact the United States, and then used precipitation composites to show how this possible winter storm has a better chance of hitting the Central U.S. with heavy snow than the East U.S. 

But we can go further with this.

Shown above are forecasts for four major teleconnections: in clockwise order, the NAO (North Atlantic Oscillation); the EPO (East Pacific Oscillation); the WPO (West Pacific Oscillation); and the PNA (Pacific-North American index). Glancing over these, only two of them look to be strong enough to actually influence the weather pattern by January 22nd, with the EPO not registering a signal and the NAO descending further into barely-positive territory at that time. The WPO is forecasted to be positive on the 22nd, but given how far upstream it is I generally don't account for it in my forecasts. The PNA, however, is forecasted to be moderately negative right before the timeframe for this potential winter storm opens up, and that provides us with another indicator to use. Let's apply it to precipitation anomalies.

Shown above are four panels showing precipitation anomalies' correlation with the PNA index. Brown colors indicate precipitation anomalies that have a negative correlation with the PNA index, while green colors indicate precipitation anomalies that have a positive correlation with the index. For example, in the brown-colored areas, if the PNA were positive, this says that precipitation anomalies would be below normal (negative, in effect). Similarly, if the PNA were negative, those same areas could then expect above normal precipitation. 

The top-left panel shows this correlation with the PNA index for the month of January, and that panel is what we'll focus on. The most important feature on that panel is the swath of brown colors encompassing the Midwest, Ohio Valley, slivers of the Gulf Coast and portions of the East Coast. The darkest brown shades are located in the Ohio Valley, almost into the Mid-Atlantic. With the PNA forecasted to be negative when the storm threat rolls around, this panel tells us that above-normal precipitation is favored in the Midwest, Ohio Valley and Mid-Atlantic in such a pattern. This is slightly different from the MJO Phase 5 composite in that it doesn't have that Southeast Ridge component apparently visible, but what counts is that both the MJO Phase 5 image and this PNA image indicate that the synoptic weather pattern will be ripe for wetter than normal conditions in the Midwest, Ohio Valley, Great Lakes, and perhaps the Mid-Atlantic and sections of the Northeast, as well as parts of the Plains.

To wrap up, let's assuage the discrepancy I highlighted in the last paragraph, how the MJO phase 5 typically brings about a Southeast Ridge but a negative PNA in January does not appear to do so.

The above image, from North Carolina State University, shows a 'textbook' negative PNA weather pattern. This is seen by the below-average heights in the Western U.S., indicative of stormy weather that could then push east in the form of a storm system (perhaps multiple, but that's a case-by-case basis). On the flip side, a 'textbook' negative PNA will tend to pump up a ridge in the Eastern U.S., as a consequence of the deep negative anomalies in the Western U.S.

But wait a second, didn't that PNA precipitation image show that there wasn't really a ridge in the East or the Southeast? Yes, but it must be remembered that the atmosphere is not rigid. A storm system will not turn due north in the above graphic once it hits the oranges and reds of that ridge. Instead, as I learned quickly (through failed forecasts), storm systems mesh with their surrounding features. Storm systems will press into the ridge and ride above it - typically not pushing it off to the side but also not treating it as a brick wall and having to change direction. We apply this here by recognizing that a storm would likely drop south and into the Southern Plains, because geopotential heights are so low in the West and thus are suppressing the jet stream south, but the storm would then move northeast as it starts bumping up against that ridge. This would result in a path through the Midwest and the Ohio Valley, perhaps exiting through the Northeast or the Mid-Atlantic. This would lead to the above-normal precipitation anomalies for those regions as seen in the MJO Phase 5 graphic and the PNA graphic, but also favor the idea that a ridge *would* be present somewhere in the East US that would allow this storm system to take such a track. Keeping with such a theme, I expect this potential winter storm to take a similar path.

To Summarize:

- A potential winter storm looks to impact the United States in the January 23 - January 27 timeframe.
- Multiple atmospheric oscillations appear to be favoring the central Plains, Midwest, Ohio Valley, Great Lakes, and perhaps portions of the Mid-Atlantic and Northeast for the heaviest precipitation that would be associated with this storm.
- Any snowfall associated in this storm would likely impact those same areas, though depending on the orientation of the ridge in the East this may not be the case for the Mid-Atlantic.
- Substantial uncertainty remains over all parts of this potential winter storm, including if the storm itself will even transpire.
- However, based on the evidence presented above, I do feel confident in the chance for a winter storm in the country sometime between January 23 and January 27, with precipitation chances potentially favoring the mid-section of the country, from the central Plains through the Midwest and Ohio Valley.