Home

A scientific discussion of the current atmospheric circulation patterns, monthly temperature and precipitation departure maps, and a general jet stream and storm track discussion. The tone of this outlook is centered on the mid-Atlantic.

Posted on 10/30/2007, updated on 11/14/2007

ENSO & PDO Discussion
GLAAM Discussion
QBO and Stratospheric Warming Discussion
AO/NAO Discussion
Seasonal Trends
Climate Trends
North Pacific Blocking & PNA
Solar Cycle Discussion
Winter Forecast Discussion - Updated
What Changed Since the Original Forecast

El-Nino & Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO)

We currently are in a moderate La Nina pattern. The ENSO 1+2 and 3 regions have been colder than -0.5 since around April-May, but ENSO 3.4 only since August and ENSO 4 stayed warm until dipping into the -0.5 territory during September 2007, according to this graph. You may be wondering why I gave ENSO only three stars. This is because I believe the influence of ENSO on the mid-latitudes can be a bit overrated and too emphasized upon in many winter outlooks.

One might argue that this La Nina had gotten off to a “late start”, but I disagree with that assessment. The surface temperature anomalies throughout this summer were a little warmer than it should have been. Remember when our El Nino last winter crashed suddenly and ENSO went cold in a blink of an eye? To me, that marked the beginning of a La Nina episode. After the El Nino crash, ENSO SST just flatlined for 2-3 months and then dropped to moderate La Nina levels at record pace. We can agree that this is strange behavior for equatorial SSTs, but that can be explained by an “interruption” in the development of the La Nina, in my opinion. In the following graphic, the thick pink line shows the actual temporal progression of the monthly ENSO SST trends this year. The dashed blue line I drew shows the progression that ENSO SST should have had, and would have had if it wasn’t for the interruption.

I talk about the above because we can easily misinterpret the “late development” by thinking that the La Nina will peak later than usual, when in actuality this La Nina has been closely following the temporal trends seen in 1988-89, 1970-71, and 1949-50. The point? This “late start” does not imply a “late peak”.

Considering the asymmetry of temperature anomalies that are biased cold towards the eastern coast of South America at the Equator, we are currently in a slightly east-based La Nina pattern, though it is becoming more basin-wide at the moment. The current La Nina is still strengthening, albeit very slightly, according to subsurface temperatures and re-strengthening equatorial trade winds. I see a peak in this La Nina during December and start weakening gradually around then, if not earlier. Subsurface temperature anomalies are still anomalously cold well east of 180° longitude dateline. However, once anomalously subsurface warm waters begin to penetrate well east of the dateline, we can anticipate a subsequent weakening of this La Nina. This may be happening already, so we will need to keep an eye on this subsurface animation. Even if it has begun to weaken, I don’t see this La Nina crashing like last year’s El Nino did.

Now we’ll talk about the analogs. I decided on the relevant analogs by eyeballing every individual La Nina and negative PDO SST map since 1949 and comparing each historical map to the current Pacific SST configuration (as of Oct 25).

Yes, I used singular-value numerical indices, but only as guides to help me locate the potentially relevant years. My top ten PDO-weighted analogs for this La Nina, using the MEI, are shown in the next graphic that shows the September to March progression of a 500mb geopotential height composite background signal:

The important signals that we can glean off this graphic is a late-fall/early-winter trough east of the GOA and neutral to slightly below normal heights in the eastern US. As winter progresses, a GOA ridge or blocking anticyclone builds, sending a downstream longwave trough into the heart of North America with a weak SE ridge signal. This is remarkably consistent and anti-correlated with an ENSO signal shown in a regressed 500mb geopotential height field with a La Nina pattern, and the resulting temperature and precipitation from these analogs:

These two graphics above are the single most important reason for a lot of warm forecasts for the east as of late. Yes, we probably will have to contend with the SE ridge and being on the wrong side of the rain/snow line at times this winter. But I warn about one thing - when we look at the above 500 mb composites using ENSO, we should not see them as if they are exactly what we will get this winter. Remember that ENSO varies slowly on seasonal time scales. For that reason, we cannot use ENSO to explain the daily-to-weekly variability of the PNA and other teleconnections. Now, let’s look at other factors and narrow down the analogs.

Back to Top

Globally-Averaged Atmospheric Angular Momentum (GLAAM)

GLAAM is one factor that I think has been under-utilized in seasonal forecasts, and I believe this factor is very important to include in seasonal forecasting. So I have given it a five star rating, despite my having problems with sample sizes in GLAAM data. GLAAM data only goes back to 1958, and when I break it up into 8 stages much like what I have done with the MJO and QBO data, the data gets a bit noisy. Due to the noise, I fall back onto the meteorological fundamentals and climate theory, as well as its relationship with ENSO:

The above animated DJF 500mb geopotential height regressions with GLAAM have remarkable similarities to those of ENSO. Positive GLAAM usually means a warm Canada and a cool and wet SE. Also, note what the North Atlantic Oscillation (NAO) does in the above graphic. The NAO starts out as positive early in an El Nino winter, and then goes negative during the later part. Sounds familiar? Warm with a positive NAO during the first half, then a pattern change; NAO goes negative and we go cold in the east. Now, we’re talking about a La Nina winter this year. Can the opposite NAO temporal pattern hold true this year? Start with a negative NAO during the first half giving us chances for cold and snow through December and parts of January, and then it goes positive for the second half and thus blowtorch the east during late January through February? Well, La Nina patterns are not always that straightforward, but I will get back to the NAO later in this discussion.

The GLAAM has gone negative down to record levels as of late summer, but has increased a bit since then. I think the GLAAM will continue to hover around current values (between -1 and -2) such that by December, the GLAAM will still be negative but not at such extreme values seen in the last two months. The GLAAM may increase slightly throughout the winter, especially after the La Nina peaks, the QBO starts to relax a little bit, and the northern hemisphere cryosphere grows in area and extent, perhaps reaching neutral at times this winter.

Therefore, the GLAAM will average negative this winter, much like the 1950s and 1960s, until around 1976 – according to this graphic. Like the PDO, GLAAM trends since 1958 show a 1976-77 shift from negative to positive. However, we have gone back to the negative side this year, and a negative PDO correlates with a strong Pacific Jet along a more northerly track. Hence, I think our winter will be much like those of the 1950s and 1960s except the arctic is warmer now. Here’s a graphic that shows my selected GLAAM analogs and a sequenced 500mb height signal for the composite analog winter:

We begin our late fall and early winter in stages 6-7, which corresponds to negative GLAAM with its first time derivative close to zero, according to my chart here:

Here’s an animation of what happens to the 500mb heights and atmospheric circulation as we move through the GLAAM stages.

I suspect we will stay around stage 6-7 through December, and then move to stage 8 and MAYBE stage 1 at some point later in the winter. We will not reach stages 2 and 3 because those stages correspond to an El Nino pattern. Stages 4 and 5 are the most IDEAL for a blockbuster KU storm, of which a historical eastern US snowstorm composite is shown here:

Notice the above snowstorm composite is an excellent match to GLAAM stages 4 and 5. Our chances of reaching stage 4 and 5 are quite low, but not impossible. In order for this to happen, giving us a real chance of a major snowstorm that can dump 2 feet along I-95, we will need the GOA ridge to retrograde and replace itself with a deep trough just off the North American coast. This would enable the Pacific Jet to take on a northerly aim and slam into the northern Canadian coastline, sending upper level shortwave disturbances around the mountain anticyclone and forcing lee cyclogenesis past the Rockies. The negative NAO combined with the 50/50 low would bring arctic air all the way south via cross-polar flow, forcing the lee cyclone to take southerly track, tap into Gulf and Atlantic moisture and detonate on top of us as the upper level trough takes on a negative tilt.

Chances of that happening are higher during a weak La Nina because GOA blocking events tend to be of short-duration and when they retrograde and fall apart, offering plenty of opportunities for a PNA ridge and a downstream eastern US trough – more on this in the Blocking/PNA section later. However, as long as we get a negative NAO during a moderate/strong La Nina, the Ohio Valley and Great Lakes are favored but we easterners can still get in on the game even in a much less than ideal setup – like what January 25, 2000 did for us.

Back to Top

Quasi-Biennial Oscillation (QBO) and Sudden Stratospheric Warming

This year, we have a strongly negative (or easterly) QBO, which is a measure of equatorial stratospheric winds in terms of strength and direction. The strong easterly QBO is among one of the reasons that this La Nina won’t go down without a fight. Fortunately, an easterly QBO favors a disturbed polar vortex and thus a warmer stratosphere, conflicting with the La Nina’s preference for a cold polar vortex.

Why are a warm polar stratospheres and disturbed polar vortices good for us? Because when the polar stratosphere warms suddenly, it usually sets off vertically propagating planetary waves of wave number 2 or less. These temperature anomalies can then propagate downward from a height of around 30 mb at a rate of 1 km/day and reach the tropopause in about a month, give or take.

When a sudden warming occurs and propagates downward, the Arctic Oscillation (AO) index decreases, and the NAO usually follows. Heights rise over the north pole, giving rise to a blocky, meridional, and as HM would say, a “convoluted” pattern, allowing cold air to spill southward wherever it wants, at will. Tight temperature gradients and baroclinic hot spots can then set off big storms that would be exciting to track. So, obviously, we want to root for a stratospheric warming, especially considering what we’re up against this winter. Unfortunately, stratospheric warming events are no guarantee. Some negative QBO winters had two stratospheric warming events within a couple of months, and some had none.

But wait, how is the QBO related to stratospheric warming, you might ask? When we have an easterly QBO, planetary waves freely propagate poleward from the tropics without getting ripped apart by meridional wind shear (which is what happens to planetary waves associated with a westerly QBO). When these waves propagate poleward, they eventually penetrate the pole and disturb it by exerting a zonal retardation force on the polar night jet.

When enough zonal force retards the polar night jet, the stratospheric polar vortex begins to break down. When the polar vortex breaks down, its swirling blob of high potential vorticity air is chopped up into many different pieces and the temperatures warm suddenly. You may see some very cool animations, demonstrating how this works. Then downward vertical propagation begins, per an EasternUSWX discussion on this topic a while back.

What does all this mean for the winter? I normalized the QBO by standardizing each month and putting this file together: Normalized QBO

Then I regressed that on 50mb heights, 2H, and 5H. The result is the following relationship the QBO has with geopotential heights at 50 mb, 200 mb, and 500 mb – corresponding to a positive QBO (for a negative QBO, just mentally reverse the +/- signs):

A positive (westerly) QBO signal obviously shows a positive AO and a slightly positive NAO and a decent Aleutian Low signal by Arctic-to-Pacific extension. For a negative (easterly) QBO, the reverse will be true, most notably with a trough near 50/50.  Next, I split up the QBO into different phases (QBO vs dQBO/dT).

I broke the QBO up into 8 stages, comparing it against its own tendency. To derive dQBO/DT, I used a 3 month moving average. Then I standardized both the QBO itself and its tendency in order to apply logic to place the index in certain stages, depending on whether the QBO itself or its tendency is stronger at a given month. This enabled me to pigeonhole each month into one of the 8 stages according to this chart:

Then I split up the raw data and created a date file for each stage, and ran the composites on CDC. Results are shown in the animated 50mb evolution given a clockwise trajectory through each QBO stage (1, 2, 3... to 8, then back to 1, and so on)

We are now in Stage 6 (both QBO and tendency negative, with the QBO strongly negative and tendency is weakening).  Note the SE ridge, consistent with October observations.

I suspect we'll be going into stage 7 soon (through November and early December), which consists of a strong negative QBO but with a weak positive tendency.  If so, this may be a good sign for eastern winter.  Note that deep negative anomaly near the 50/50 position along with a strong Greenland block. I'll address more of this in a couple of weeks.

Chances for stratospheric warming events are heightened because we now have an easterly QBO. If one does occur, that’ll give us a few weeks lead to a relatively cold and blocky pattern.  But there is no guarantee of a stratospheric warming event, even during an easterly QBO. And we don’t need a stratospheric warming event to crank out a snowstorm. Hence the two star rating for the importance of the QBO. The AO and NAO are also affected by other factors as well as forced stochastically – more on that in the next section.

Back to Top

Arctic Oscillation (AO) and North Atlantic Oscillation (NAO)

First off, I use these teleconnections concurrently because they are strongly correlated, as we all know. Second, I put a five star emphasis on this AO/NAO pair, which I will refer to as just the AO from here on, because it is a major wildcard as to how this winter will turn out – not only temperatures, but also the jet stream pattern and storm track evolution across the CONUS. Here’s a graphic that shows the difference between a positive AO (right) and a negative AO (left).

There are two main conflicting signals for the AO for this winter. La Nina usually favors a weakly positive AO with a relatively undisturbed polar vortex. An easterly QBO, however, usually favors a negative AO with a disturbed polar vortex. Currently, the AO/NAO cannot be forecast beyond one month, and predications made for two weeks and beyond are also tenuous at best. Chuck’s SST reemergence theory for predicting the NAO beyond one month has some merit. But, unlike the tropics, I see mid/high-latitude SST patterns only as a reflection of what’s going on in the atmosphere with some feedbacks from surface fluxes. However, it is those surface flux feedbacks may just prove to be the “tiebreaker” between ENSO and QBO influences.

Obviously the NAO is forced stochastically by several different factors including transient eddy and storm track variability, but I wanted to see whether Atlantic SSTs or stratospheric warming events have a larger impact on the NAO.  Here is a 50mb height correlation with NAO, with a 1 month NAO lag vs 1 month NAO lead.

North Atlantic SST correlations with NAO... 3 month NAO lag vs 3 month NAO lead

It's obvious from these two graphics that the stratosphere has an advantage in leading the NAO (or the NAO lagging the stratosphere), because the correlations are stronger when the heights lead the NAO by 1 month. These same correlations fall apart outside of 2+ months. As for SSTs, the main focus point/area of correlations is the gulf stream from the east coast to just off Newfoundland. When SSTs "lead" the NAO, only the gulf stream has marginal correlations while more significant correlations occur when the NAO leads the SSTs - not surprisingly due to surface fluxes, wind, and Ekman pumping.

This is not to say that we should completely disregard SSTs when it comes to the NAO... the SST reemergence theory applies only to the circled red area in the following graphic (NAO "lags" SSTs by 9 months):

What I mean by surface flux feedbacks is this: when a weak disturbance traverses over a patch of anomalously warm SSTs, the disturbance may grow due to additional moisture, instability, and vertical uplift. As the disturbance grows into a full-blown storm, strong winds will help deepen the ocean mixing layer and induce Ekman pumping to cool the SSTs north of the storm track and warm the SSTs south of the storm track. At the same time, that storm will transport heat northward, building a downstream ridge and an upstream trough that would adjust the upstream storm track, and so the pattern repeats cyclically guided one way or another by atmospheric and surface feedbacks. The same process occurs in the Pacific with the additional influence by the downstream mountain torque, as shown by daily-weekly fluctuations in the PNA (of which short-term variability cannot be explained by a much slower time-varying ENSO).

Here are the regressions of the AO and NAO. Notice a positive height anomaly between the Aleutians and Hawaii in the AO, which matches that of a La Nina background signal in the same region.

I think that these conflicting signals only means that one will dominate for a period of time, and the other will take over for a subsequent period of time, and go back and forth. As HM said, it isn’t whether one will win over the other, but rather represents a variety of patterns that we will experience throughout this winter. I think the NAO will go positive in late January through mid February, but my timing could be off especially considering a possible stratospheric warming event happening at any time throughout this winter since the atmosphere already favors it. Or, the La Nina could weaken faster than originally thought, allowing the NAO to drop further during the later half of winter. There is a good post on EasternUSWX comparing –NAO versus +NAO in La Nina winters.

Back to Top

Seasonal trends

We have had a record warm October in the east, with at least fifteen 80+ degree days at DCA and a maximum high of 94, which is almost unheard of in these parts. However, the east has received beneficial rains from a cut off low recently and models are showing agreement on a pattern change into November. November is an important month because it likely will set the tone for the rest of the winter. Although the models are showing a +PNA for a good part of early-mid November, the background –PNA signal of the La Nina will likely assert itself at some points during this winter.

Statistically, a cold November is a good sign for an upcoming winter, though there is an interesting take on the correlation between November and winter temps after a warm October – see EasternUSWX thread. I still think a warm October and warm November during a moderate La Nina is bad for an east coast winter. However, one snowstorm is all it takes to make or break an entire winter for many locations. And making snowfall predictions is like throwing a last second hail mary into the dark.

In the beginning, with little arctic air genesis due to record low sea ice and low snow cover compared to climatology, the air will actually be merely cool to marginally cold in an otherwise very cold pattern in the first stages of winter. As snow cover expands and sea ice fills in, genesis of true arctic air may then commence, which I think will happen by mid- to late-December.

Climate Trends

In the last ten years, we have seen a probable climate shift to a negative PDO regime, which indicates a change in atmospheric circulation that involves a strengthening Pacific Jet and a positive EPO:

Several of these recent winters, we have seen a strong Pacific Jet flood the North American continent with mild Pacific air and only so often allowing arctic air to get involved further southward. As long as the Pacific Jet plays a role, we need a negative NAO or a North Pacific blocking event to disturb the flow and allow the cold air to filter southward. Being a La Nina winter, the Pacific Jet will continue to play a role and the EPO will probably average out on the positive side. However, blocking events will be probable – more on that the next section.

It may be too early to call, but I think we are now entering a new climate era of more La Ninas than El Ninos, just like the 1940s through the early 1970s. The Pacific “looked bad” in those years. But we seemed to get a lot of good snows despite the Pacific, perhaps due to a favorable Atlantic. However, we still have global warming, a warm arctic, and record low sea ice. We still have to see how that plays into arctic air genesis this winter. Personally, I think arctic air genesis starts a little later relative to climo, but the timing of the onset of spring should remain normal.

Back to Top

North Pacific Blocking and the Pacific North American (PNA)

I have already written a great deal about the PNA and blocking, and you scan through my research here and look at the graphics - ENSO-PNA relationship and North Pacific blocking

The following graphic is a basic state un-decomposed PNA regressed on 200 mb, 500 mb, SLP, and temperatures:

Not surprisingly, the PNA leads the PDO as shown in the North Pacific variability section of my research paper. As you can see, a positive PNA gives the Pacific an imprint of a positive PDO. Conversely, the negative PDO that we have now is a negative PNA signature. The following graphic demonstrates the strength of PNA variability with or without ENSO influence, as well as showing how ENSO would influence the PNA with its background signature.

The above animation starts with a basic state un-decomposed PNA regression, then shows a PNA signature WITHOUT the ENSO signal, and the last (and weaker) one shows the ENSO background imprint onto the PNA. Credit for the latter two indices due to Hart; a while ago, we worked together on decomposing the PNA to investigate what factors influence it (thanks, Hart). The ENSO signal biases the PNA wave train slightly to the east while pushing the Pacific Jet further south during warm episodes. A cold ENSO would induce more North Pacific blocking and push that jet further north instead, however. However, even without ENSO, the PNA regression still looks quite strong, indicating that there are several other factors that influence it, such as the MJO, storm track and eddy variability, heat and momentum fluxes, mountain torque, polar teleconnection variability, and changes in relative angular momentum.

The strength of the La Nina can affect the strength and duration of blocking episodes in the North Pacific. A weak La Nina would have mostly short duration events such as these:

While a stronger La Nina would have more persistent blocking anticyclones over the North Pacific:

What a North Pacific blocking event means to the eastern US depends greatly on the NAO. A positive NAO coinciding with a blocking event would likely warm up the eastern US, while a negative NAO can get us in on the cold - click here for animated demonstration.

I think we will see both types of patterns, making this a quite variable and interesting winter. I think we’ll start off with a neutral to negative NAO this November through December with only marginal cold air and a few rain storms that “should” be snow, and a couple of significant snowstorms late December through January. At that point, the NAO may go positive, thawing us for a while, and go back negative later in the month through February and March. It might even be more variable than that, giving us a roller coaster type of winter. One day, 60 to 70 degrees and sunny; next day, 30 and snowing… I think it will be that kind of winter.

Back to Top

Solar Cycle

We are currently in a minimum of the solar cycle. Solar minimum analoging shows an east-based negative NAO signal and a slightly enhanced jet across the Gulf of Alaska:

When solar minimum analoging is focused on La Nina conditions with an easterly QBO and negative GLAAM, the jet is turned more to the northeast and the negative NAO is more basin-wide.

Last, but not least… a special new index that I created:

The Squirrel Behavioral Index

When I was walking home from a convenience store yesterday, I saw the fattest squirrel I have ever seen in my entire life. I mean, he was fat. Sitting on the sidewalk with cheeks chock full of acorns. And he was quite slow. As I got closer, he took one look at me, frowned reproachfully at my invading his territory, and slowly moved out of the way. He didn’t even dart off the sidewalk and 30 feet up a tree as most squirrels do. He just crawled tranquilly off the sidewalk under a shrub, and after I passed him, he came back out on the sidewalk and went about minding his own business. Talk about bizarre behavior for a squirrel.

I don’t know whether he’s hoarding all these nuts thinking a harsh winter is coming, or he’s got no sense of urgency thinking that this winter will be quite mild.

And finally...

Back to Top

The 2007-08 Winter Temperature and Precipitation Forecast Discussion (Updated Version)

This winter will be an interesting one, despite a bad prognosis based on past La Nina’s influence on the eastern temperature departures. We will see quite a bit of variability regarding temperatures, precipitation events, and storm tracks. It won’t be one of those winters that can be defined in two halves like last winter, but rather a pattern change every two to three weeks or so. The base longwave pattern will probably consist mostly of a Midwestern / Plains trough, with the SE ridge poking in once in a while. I decided not to do monthly jet stream maps because I think the jet stream will vary wildly, especially over the Rockies and downstream. I have created the following jet stream map - averaged for the entire winter - that shows the main feature (the Pacific Jet), along with different scenarios per the NAO and the dashed line is the analog jet stream:

Lots of storms will slam into the Pacific NW and western Canada, but once these storm reach the west coast, they will immediately encounter lots of cold air, especially after late December. The clash between these storms and the cold air will produce lots of snow for the northern Rockies and Cascades as well as some snow showers with lots of wind across the Upper Plains. As for east of the Mississippi River, things look a little more uncertain and would depend more on the NAO. I foresee 2 or 3 blocking events, at least, and while they may bring warm temperatures to the east, it is the decay and retrogression of a blocking event that can give us some cold air outbreaks that originate from northern and western Canada. At times, the storms from the west will be able to dive far enough south to give the eastern seaboard a decent chance at snow. Many times, however, the Midwest and Ohio Valley through Ontario will be favored for significant snows as storms will track along or west of the Appalachian Mountains. In these instances, precipitation type along the eastern seaboard will probably be mostly rain, or wintry mix if we can get some cold air damming. It appears that we have a good chance at least one 6”+ snowstorm along the east coast despite long odds, and I expect that to come within late Dec - mid Jan and/or late Feb – mid March time frames, when I foresee cold temperatures for the east. This is not to say we won’t get any snow during the warmest months of my forecast (late Jan through Feb), but I don’t want to get more specific than that. My confidence in the seasonal averaged temperatures is at medium, but quite low on the monthlies. First, the seasonal departures:

Orange being > +3
Yellow being +1 to +3
White being -1 to +1
Light blue being -1 to -3
Dark blue < -3

Lastly, the monthly temperature forecasts – however, my confidence in these is quite a bit lower than in the seasonal-averaged forecast above.

Back to Top

What changed since my original October assessment?

It appears that the cryosphere has done a great job in catching up to climatology, and it appears that there'll be no lack of cold air throughout the winter. However, the Pacific Jet is still quite strong, and it may keep the cold air from invading the CONUS too much, especially early this winter.

The La Nina has continued to strengthen and may reach "strong" criteria soon, so it will exert some influence on the mid/high-latitudes, favoring a cold north pole and a strong zonal jet. However, notice that I didn't blowtorch the entire east for the month of January, because I wanted to keep open the possibility of a early Jan stratospheric warming event which may propagate downwards, tank the NAO, and send some cold air southward. The QBO, I think, has reached its peak, so I would look for a stratospheric warming event a little sooner rather than later.  I predicted that we will go into stage 7 of the QBO and the observations seen in early November confirms this.

My confidence in the temp monthlies east of the Mississippi River is also quite low, as I still think this will be a volatile roller coaster winter... whereas a few days of utter blowtorch or extreme cold in the month can skew the entire month one way or another. I have better confidence in the seasonal temperature map, however. I believe it will be a cold winter across the Pac NW, the northern Plains and the Midwest. The OV and NE will average around normal (with above normal precipitation), while the mid-Atlantic averages slightly above normal, with the southern states averaging above normal and continued dry.

Back to the Seasonal and Storm Forecasts Page

Home | About WLH | Contact WLH | Site Map
Website Designed by Nabeel Keblawi | Copyright @ 2006 All Rights Reserved