Wednesday, January 22, 2014

Long Range Lookout: Model Confusion Signals Pattern Change Ahead

It's time we start up our weekly Long Range Lookout segment, though this particular post will focus more on the medium range. Nevertheless, this discussion will be as important as the rest of them.

8-10 day 500mb height anomaly forecasts from the ECMWF (left), GFS (middle) and CMC models.
Medium range model guidance above shows rather drastic model differences between each guidance system, with multiple areas of the Northern Hemisphere differing in what exactly is going to happen. We see the ECMWF model on the far left prefers to hold a negative Arctic Oscillation (AO) regime in place, as is shown by stagnant high pressure in the Arctic Circle. In this forecast, the polar vortex appears to be broken into two pieces, with one lobe located in northern Eurasia and the more significant piece of the vortex displaced over Greenland, producing a strong positive North Atlantic Oscillation (NAO). The positive phase of the NAO tends to induce zonal flow over the nation, meaning no huge cold or warm outbreaks anywhere in the nation. Despite this, we still see suppressed high pressure over the West Coast producing a positive Pacific-North American (PNA) index, and that leads directly to suppressed low pressure anomalies over the Central and East US. I'm not buying the positive PNA forecast, and the reason for that will be explained a bit later down in this post. I do agree with the negative AO projection, though I think we may see high pressure near the Gulf of Alaska expanded a bit in the future.
For the GFS forecast in the middle, we see a much stronger negative Arctic Oscillation signature, with high pressure forming a bridge across the North Pole. This splits the vortex fully into two pieces; one part ends up in northeast Asia, while the other part is weakened and stretched out between Greenland and the United States. The GFS also shows a positive NAO forecast, which is why I'm a little hesitant accepting that deep low pressure anomaly in the Eastern US. However, strong high pressure blocking in the upper latitudes over Alaska would support such an idea, though the lack of support from the Pacific North American index gives me the feel the GFS is a bit too ambitious by bringing the vortex down towards the United States. Other than that issue, I believe the remainder of the forecast is rather sound. 
The CMC model on the right has a very similar forecast as the ECMWF model, though in this case we see a closed low offshore Baja California, in what appears to be an attempt at a Rex Block (high pressure directly north of low pressure, see image below). In this particular case, though, adding to the issues I find with the ECMWF forecast, the subtropical jet stream would most likely be strengthened by this closed low. Due to the positive NAO forecast in both the ECMWF and CMC forecasts, and due to how we aren't finding an El Nino in east Pacific waters (as the image below shows), I'm not so sure the pattern is conducive for strengthening of the subtropical jet stream after this northwest flow pattern we are currently in dissipates.

Typical Rex Block pattern. Notice the zonal flow to the east of the blocking pattern.
Observed SST anomalies over the El Nino-Southern Oscillation (ENSO) monitoring area. The lack of significant warm waters at the bottom of the image tells us there is no El Nino currently in place. (CPC)
Let's take a look over teleconnection forecasts now, and see what they can tell us about the medium range period.

We'll begin with the Arctic Oscillation teleconnection forecast. As I stated above, the Arctic Oscillation has a negative phase, which means stagnant high pressure over the Arctic, and there is also a positive AO phase, which indicates stagnant low pressure over the Arctic. The -AO argues for a colder weather pattern in the United States, while the +AO argues for a warmer weather pattern.
The first thing we notice when looking at the ensemble forecast for the Arctic Oscillation is the wide spread in ensemble members at the medium and end of the forecast period. This tells us that the weather pattern is about to flip after we get out of this northwest flow pattern, and the discrepancies among the medium range models above confirm this idea. So, how can we pull a forecast from here? We can look at the errors previous forecasts made. Looking at the bottom panel of this image, we see the observed Arctic Oscillation value superimposed on the spread of forecast members from previous days, weeks and months of forecasts. Looking at that image, it seems that the observed Arctic Oscillation usually ended up more positive than the forecasts thought it would be. This has been the trend through much of the winter, until recently. Looking at the observed Arctic Oscillation from December 14th onwards, the trend has been for the observed AO to be more negative than forecasts believed it would be. I do not believe the idea that the Arctic Oscillation will be positive in the medium and long range, so it may be best to ride the new trend of lower AO values than forecasted and predict a neutral or negative Arctic Oscillation in the medium and long range time period.

This image shows the Pacific North American index (PNA) forecast over the same timeframe as the AO forecast above. We also see those superimposed panels like we saw in the Arctic Oscillation image. The PNA has two phases: positive, which means high pressure dominates the West Coast and leads to wintry weather over the East US, while a negative PNA depicts low pressures over the West Coast and warmer weather in the Central and East US. The forecast members here appear in sync with the idea of the Pacific North American index trending more negative in the future, and I agree. Some model systems haven't really caught on yet, and are still projecting frigid forecasts for the long range. I anticipate we see a gradual warming of those forecasts, though the Pacific should still be conducive for at least some decently cold weather. If we use that trend idea in the bottom panel here like we did with the Arctic Oscillation image, we don't see a particular trend right now- the PNA has stuck relatively close to its forecasted values in the long range. Thus, I see no reason to back off the trends now. I expect an eventual -PNA to evolve.

Why are models so confused? Well, we can look to the shorter term for that answer. This image shows an ensemble projection from the Climate Prediction Center for 5 days out, unlike the week-plus forecasts we discussed above. On the left is the ensemble projection for one height level (5520m), while the right panel shows the forecast for another height level (5760m). These images are both valid at the same time, they're just forecasting for different parts of the atmosphere, if you will. The pattern is clearly defined in the left panel, with ensemble members agreeing on the northwest flow pattern continuing, but the picture is more muddled with the panel on the right. On the right, we see high model disagreement with the pattern in the waters west of Mexico, with some members wanting to develop a closed low, and other members wanting to retain high pressure over the West US. Still other ensemble members want to keep the pattern more zonal and not contaminated with high or low pressure. So, who do we believe? Well, considering the Pacific-North American (PNA) index looks to still be positive over the next few days, and other ensemble guidance agreeing on intense cold weather staying in the US 5 days away, I would buy more into the high pressure on the West Coast solution than the closed low idea.

Long range ensembles still aren't faring too well.

This four-panel image shows the 500 millibar height anomaly forecast over the Northern Hemisphere from the GFS model (top left), the GFS Ensemble Control (top right), and the GFS Ensemble mean (bottom left). The bottom-right image shows the ensemble spread across the Northern Hemisphere. All four forecasts are more or less similar to the GFS image we discussed at the beginning of this post, but take a look at the ensemble mean. We can observe the continued model confusion with deeper purples emerging over both the Bering Sea region and over Greenland, signaling higher model uncertainty. The model uncertainty in the Bering Sea is likely due to how strong developing high pressure in that area will be, with some forecasts suppressing the ridge of high pressure to the south of the Bering Sea, and other forecasts flooding the body of water with high pressure. Similarly, models don't know how to handle pressure anomalies over Greenland. If there's low pressure present there, as some models indicate, we would see the positive North American Oscillation (NAO), which would support that warmer and zonal weather, while a negative NAO would support a continued northwest flow and cold weather in the East US.

So what do I think will happen in the long range, since models aren't giving too much to work with at face value?

Well, we can probably expect a break from the intense cold we will see to round out February. In its place ought to be a warm Southeast as some ridging takes hold thanks to that negative PNA, but also some cold weather in the Northern US as the negative Arctic Oscillation persists. It is possible the ridge in the Southeast is suppressed if we can get high pressure over Greenland to get a negative North Atlantic Oscillation going, though model guidance everywhere is having trouble with that aspect of the forecast. The West can anticipate a well-deserved break from the persistent high pressure, with stormy weather making a comeback... to some degree. We'll need to see how model uncertainty in the Bering Sea resolves itself before we can be sure the West will get at least a bit of relief.