Thursday, January 30, 2014

February 4-6 Potentially Significant Winter Storm - Updated January 30

There remains a good chance of a significant winter storm in the February 4-6 timeframe. There is also a good chance of the storm track shifting north, so let's dive right in.

(The following is copied from the January 28 post on this storm)

Shown above is the 500mb anomaly forecast off the ECMWF model, valid on the morning of January 26th. If you've followed this blog for a while, you know that we can use that area of negative height anomalies over Japan to forecast a storm in the United States. As Joe Renken states, we can extrapolate a storm over East Asia out 6-10 days to arrive at the timeframe for a storm here in the United States. Going that length past January 26th gives us a storm timeframe of February 1-5. That's not the only thing we can take away from this predictor, however. Looping the ECMWF forecast tells me that the storm system will be moving essentially due east as it passes through central Japan. This tells me that we could (keyword here is could) see our February 1-5 storm system also moving on a west-to-east track, not really curving north or dropping south too much. The 12z and 18z GFS models caught on to that idea of a west-to-east storm, but had the storm hitting the US on January 31st to February 1st, which is just outside the timeframe presented here. We'll discuss the models a bit later on in this post.
Adding to the evidence of a possible storm in the early February timeframe is what we saw happen in the Bering Sea on January 18th. The image above shows contoured 500mb height anomalies over the north Pacific on January 18th. We see a storm system moving east from far northeast Russia, which is the westernmost 'X' on the image above. The next day, we saw the storm move nearly due east to the second 'X' in the image, only moving just a bit north in the process. Going back to that link I posted above, there is also something called the Bering Sea Rule, developed by Joe Renken. The Bering Sea Rule states that a storm in the Bering Sea can result in a storm over the United States 17-21 days later. If we extrapolate the January 16th date out using the 17-21 day timeframe, we arrive at a potential storm in the February 2-6 timeframe. This fits well in the February 1-5 timeframe we got from the East Asian correlation, and the nearly west-to-east movement in the Bering Sea tells us the storm in the US may be west-to-east as well, like we saw in the ECMWF forecast. 

The East Asian correlation and the Bering Sea Rule aren't the only things rooting for this storm- the Lezak Recurring Cycle is also in on this storm potential.

The Lezak Recurring Cycle, or LRC, is a tool developed by meteorologist Gary Lezak that, in essence, can enable forecasters to predict the overall weather pattern months in advance. The gist of the LRC involves a cycling weather pattern that develops in October and November of each year; no pattern is the same from year to year. Around mid November, the LRC begins to repeat, meaning we start to see a similar weather pattern in mid November that we saw in early October. This means that the cycling pattern has begun, and it will continue to cycle on a regular (non-changing) 40-60 day interval for the next ~10 months before it dissipates over the following summer.

This season's LRC cycle is on a 57 day cycle length, meaning the weather we saw in December can be expected to "come back" as we enter February. The image above shows the surface condition map on the left and 500mb map on the right from December 14, 2013. If w extrapolate that date out 57 days, we come to the February 7th mark. To make up for natural shifts that occur in the weather pattern, we can reasonably stretch the timeframe out to February 6-8 if we want to. In this case, it wouldn't be the strangest thing for this system to bend the cycle length a bit to fit in with the East Asian + Bering Sea correlations, and I feel that they will be working in sync on this particular storm threat. As the 500mb map above shows, we have high pressure over the West Coast, with only slight ridging in the Southeast and the storm system easily working its way across the Plains. Based on the pattern I'm seeing to start February, I'm thinking a storm track similar to the one this system took in mid-December could very well be in the cards.

Now that we've got the trio of long range predictors backing this storm potential, confidence is rather high on the idea of a storm system happening around February 4-6. Now, we can start to outline a possible track. 

This is the Pacific-North American (PNA) forecast for the next several days, with the forecast shown on the top panel, composed of various ensemble forecast members. We see that the PNA is forecasted to remain negative before, during and after the time period of this potentially significant winter storm. Beyond the storm, it looks like the PNA wants to move back towards neutral or positive territory, but the majority of ensemble members want to stay in a solidly negative state.

The pattern in a negative PNA state is shown above, and we can see how cold and stormy weather invades the West US as a result of persistent high pressure in the Gulf of Alaska. This stormy West then allows high pressure to form in the East, essentially cutting off any chances of a Nor'easter. What the -PNA does allow is the Great Lakes Cutter type storm, where a storm system begins in the Plains and 'cuts' north towards the Great Lakes, laying down potentially significant snowfall in its wake. It's very possible we see somewhat of a cutter storm here- the long range predictor East Asian correlation and Bering Sea Rule argue for a more west-to-east track, which does look like the preferred track at this point, but models may want to shoot the storm north when it hits the Ohio Valley, which also is in the realm of possibility. I'm supporting the more northern track as a result of the Southeast Ridge showing its strength when the system tries to dig towards the Gulf of Mexico.

Now, we have to discuss the models themselves, and where they've gone wrong. Model guidance has really pulled the storm track to the south (though recently they have been correcting north again). I'd like to explain why this south track is unlikely, and why the more northern track is probable.

There's a key model bias that has been (and will continue to) impacting model forecasts for this storm. The model bias references a negative height bias, which is shown when a storm system enters the Plains. In those cases, model guidance lowers heights in the East US too much, leading to a storm track too far to the south. In reality, high pressure systems in the East should be stronger than what they are projected to be on model guidance, leading to a more northern storm track. This system fits the bill for that bias; for the full explanation from the NWS office in Louisville, you can read below:

The model has a slight cold (heights too low) bias, especially for about the eastern third of the country, with respect to the prediction of mid/upper tropospheric geopotential heights and resultant thickness calculations. Often the MRF will dig troughs in the height field too far south across the Great Lakes and Northeast regions, most noticeably after about Day 3. During the cold season (mainly October-April), the MRF will depress the storm track too far south across the Plains states as a response to this cold bias. Forecasters across especially the Northern Plains/Upper Mississippi Valley areas need to be aware of this characteristic and be careful predicting, for example, snowstorms to "go to their south".
How have we already seen this model bias? Take a look at this side-by-side forecast from the 0z GFS model on the left, and the 12z GFS on the right, both forecasting 500mb vorticity values for the morning of February 4th, when the storm is just beginning.
If we look at the 0z GFS on the left, we see the storm system on February 4th is digging into the Plains, but the high pressure in the East US, shown as the arch-like contour lines, isn't strong enough to push the storm system north. This lets the storm system stay to the south and dump snow on the midsection of the US. However, in the 12z GFS, we see what looks like a correction of that height bias we mentioned earlier, as the heights are noticeably higher in the East. This allows the storm to push north and dump snow across the northern US, including the Plains, Midwest, Great Lakes and parts of New England. The negative PNA (whose forecast went even more negative today when compared to yesterday) raises an immediate red flag to the forecasts that want the storm to stay further to the south. The increasing strength of the -PNA projections only highlight that red flag. There are indeed items that could make this storm take the southern track, but I find it more reasonable right now to take that northern track.

We could compare maps all day, but a better idea would be to observe biases from different models and see what we can deduce. Let's take a look at a few charts below.

This image above shows 500 millibar height contours in two different colors, valid at the same time. The blue lines represent observed 500mb heights as of January 29th, while the red lines show 500mb heights from an earlier ECMWF model forecast. This ECMWF model forecast is at Hour 144, and is valid for January 29th; the same time period as the blue lines/observed 500mb heights. Why are we using this? By comparing the observed 500mb heights and ECMWF Hour 144 500mb height forecast (for future reference, I chose Hour 144 because we are about 144 hours away from this potential storm system as of this posting; it is to get the best grip on potential biases), we can find model biases. Just glancing around the image and comparing numbers on a red line to the same number on a blue line tells us that the red lines are far too south. For example, if we see the 552 dam height contour on the red line, and then find the 552 dam height contour on the blue line, we see that the red line is much further south than the blue line. This tells us that the ECMWF model has a bias in this case that forecasts 500mb heights too low. How does this affect our storm? Well, if the forecasted heights are too low, it could mean that the Southeast Ridge will actually be stronger than what is being forecasted, and that would result in a northward track.

This image shows observed 500mb heights in blue and the hour 144 500mb height forecast, but this time the forecast is from the GFS model, not the ECMWF model. Despite the model change, the bias remains the same- comparing the same number on different colored lines results in the GFS forecasting 500mb heights too low. This bias would once again argue for that northward track. The GFS model has been trending north with its latest forecasts, which could mean it is shedding this bias. The UKMET (not shown) also has this heights-too-low bias.

There is also some concern about high pressure coming down from Canada being too strong, which would then force the storm system south. However, model guidance has problems here, too.

This image shows observed mean sea level pressure (MSLP) values in blue, and the ECMWF 144-hour forecasted MSLP values in red. If we look in the Southern Plains, we see a small swath of 1032 millibar MSLP values, indicative of high pressure. But, if we then look for a corresponding 1032mb contour on the red lines from the ECMWF forecast, we see a big swath of 1032mb values displaced a bit to the south. If we keep comparing numbers, the result is the same: it appears the ECMWF forecasts MSLP values to be too strong than what they are when the MSLP values are actually observed. In our case, this bias could mean models are being too aggressive with the Canadian high pressure. Weakening of that Canadian high pressure would then help out chances for that northward track.

If we look at observed MSLP values in blue and 144 hour forecasted MSLP values from the GFS model in red, the bias remains the same. We can clearly see by comparing the location of any pressure contours in the US that the GFS is overdoing high pressure systems when you compare the forecast to what's actually observed. This pressures-too-high bias would really strengthen the argument for that northward track.

Still don't believe these biases (or you just don't get what's going on here)? The two two-panel images below show you computed 500 millibar height biases from the GFS' 0z forecast, which is top image, on the left, and 12z forecast, seen on the top image, on the right. The UKMET (United Kingdom forecast model) 500mb bias' for the 0z forecast (bottom image, left panel) and 12z forecast (bottom image, right panel) are shown.

GFS 500mb model bias.
0z forecast bias on left, 12z forecast bias on right.
Blues mean the model is too low with 500mb height anomalies.
Reds mean the model is too high with 500mb height anomalies.

UKMET 500mb model bias.
0z forecast bias on left, 12z forecast bias on right.
Blues mean the model is too low with 500mb height anomalies.
Reds mean the model is too high with 500mb height anomalies.
The result is quite clear: model guidance is far too low with 500mb heights in the East US. This means that if the bias transfers over to current forecasts (which it very well may do), we can expect a stronger Southeast Ridge in future forecasts, leading to a more northern solution.

As you can see, there's a lot of evidence arguing for these models being too far south in their forecasts. And while I do believe that there's a rather strong case to be made for the models being too south, both due to bias and not getting the incoming negative PNA regime right just yet, there is some evidence that the storm could go south. However, I'm more confident in this northern solution at this time.

But we aren't done yet! Let's talk about how this storm could get to be a monster, and what's prohibiting it as being so on current forecast models.

I. Gulf of Mexico is Open for Business

This is a forecast of the 700mb relative humidity forecast, with 700mb wind barbs and pressure contours superimposed. The point I'm making with this image is not when it was made, but rather what's going on here. Check out that huge moisture fetch from the Gulf of Mexico! Strong winds coming straight from the Gulf will provide this storm with copious amounts of moisture to work with. This moisture will translate into heavy snow, heavy rain, and even sleet and freezing rain. This is a big component of the storm, and thankfully, it is a component that models are not fighting over, unlike what seems like every other part of this potentially significant storm. Needless to say, whoever wins the snowy side of this storm will win big. Not 2 feet of snow, but probably just over a foot or so.

II. Jet Stream Phasing Needed

In order for this storm to really give its all, we will need to see the two branches of the jet stream phase. This means the 'polar jet stream' and 'subtropical jet stream' would need to come together when the storm comes along. Above, I have a GFS forecast showing a lack of phasing on the left, and a forecasting showing phasing occurring on the right. The left image shows the two jet stream branches completely disorganized, obviously separated from each other. That particular GFS forecast resulted in a weak winter storm. However, on the right panel, we see the branches of the jet stream are phased as one, and that particular GFS forecast resulted in a substantially more snowy solution. If we want to see this storm really get going when it hits, we should be rooting for a phased jet stream out west. More recent forecasts aren't as gung-ho on the phasing shown in the right panel above, but they are certainly more open to the idea of combining than the left panel.

Lastly, let's discuss the MJO. 

Another item rooting for this northern storm track is the Madden-Julian Oscillation. The Madden-Julian Oscillation, also known as the MJO, involves the placement of enhanced or suppressed tropical convection year-long. The image on the left shows precipitation anomalies for Phases 1, 2, 3, 5, 6 and 7 for January, February and March. Phases 4 and 8 were left off, as they are not needed in this post. The forecast on the right shows 200mb velocity potential anomalies forecasted by the GFS. We see blues (dashed lines) highlighting active tropical convection, thus an active MJO phase, while oranges and solid lines indicate the lack of tropical convection. We see that there is a swath of blues covered by a dashed line oval at the bottom of the image, near the 180 longitude line. The MJO is predicted to be just west of that line, and if you are to look at the outgoing longwave radiation (OLR) composite for December-January-February MJO phases, you would find that Phases 6 and 7 are defined by enhanced tropical convection (blue colors on the right panel) being placed on or just west of the 180 degree longitude line. Taking a look at Phases 6 and 7 on the precipitation composite on the left tells us the MJO favors a storm track through the Midwest, Great Lakes and Ohio Valley. This would enhance the chances of that northern storm track working out instead of the southern storm track for this storm system.

Here's a glimpse at a recent forecast showing the northern track at this point. To be perfectly honest, it does look rather realistic, with maybe a slight bump up or down in totals in the near future.

Don't forget to like us on Facebook and follow us on Twitter!