Real-time Weather-Climate Discussion and Predictive Insights -- December 1, 2006  

Edward Berry, NWS and Klaus Weickmann, ESRL/PSD

Since our last discussion (18 August 2006 and updated September 9), warm ENSO conditions have continued to mature.  The overall spatial distribution of global SSTs has warm anomalies across the Indian Ocean (IO) and the central/eastern tropical Pacific (anomalies ~plus 1-3C) with cool values across southern Indonesia and the eastern Indian Ocean. Warm anomalies also persist across the Atlantic particularly from the subtropical Caribbean to the west coast of Africa, possibly attributable to the positive phase of the Atlantic Multidecadal Oscillation.  Actual SSTs of 30C and higher are present around the equatorial date line with 29C and warmer across the Indian Ocean.   Latest TAO buoy data suggest that positive temperature anomalies along the equatorial cold tongue extend to depths of about 250m with magnitudes up to plus 5C at 150m around 120W.  This represents additional anomalous deepening of the oceanic thermocline due a Kelvin wave currently moving eastward along it.  

After several months of inactivity, a Madden-Julian Oscillation (MJO) developed over the Indian Ocean in early September and became fairly strong and coherent, as indicated by monitoring tools such as the Wheeler plot (shown here) and the coherent outgoing longwave radiation (OLR) modes (here).  When the MJO emerged into the west central Pacific during late September, it initiated a strong low-level westerly wind event that forced the oceanic Kelvin wave discussed above, accelerating the development of the current warm event.  Convective flare-ups continued west of the dateline during most of October, aided by the warm ocean temperatures there. Significant impacts on the global circulation (with feedbacks) occurred during the eastward trek of the MJO from the Indian Ocean to the central Pacific Ocean. The MJO weakened abruptly as its signal came back into the Indian Ocean during late October.

Part I begins with an overview of the SSTs, tropical convective forcing and large scale circulation anomalies. We briefly review key aspects of the development of the current warm ENSO.  We then turn to case studies of the complex interactions between the global circulation and tropical flare-ups including those with the MJO.  These interactions led to several high impact weather/extreme events.  A synopsis of the current weather-climate situation with predictive insights follows in Part 2. For information on the status of El Nino and the MJO (including a week 1-2 global hazards outlook) please see the following links. 

Latest CPC ENSO Advisory  

Latest CPC MJO Discussion and tools

ESRL/PSD MJO web site

Latest MEI discussion

Part 1.   Weather-Climate Overview 

Figure 1 (below) shows a time-longitude section of near equatorial SSTA for November 2004 to the present. The life cycles of the 2004-05 central Pacific warm event (denoted by "EN"), the weak 2005-06 La-Nina ("LN") and the recent evolution toward basin wide El-Nino conditions are seen.  See past weather-climate discussions for details on the earlier warm and cool events.  There has been a biennial component to the recent ENSO cycle with warm events during 2002-03, 2004-05 and now 2006-07. Consistent with this quasi-periodicity, the recent warm event developed quite rapidly, surprising climate scientists who typically expect a slower 3-4 year ENSO cycle. The persistent warmth in the Indo-Pacific warm pool in recent years (+1.0-1.5 sigma anomalies), and especially east of 140E since ~July 2001, has maintained the climate system in a precursor El Nino state for much of the time. The period since 2001 is comparable to the January 1990-July 1995 period, which also saw weak-moderate basin-wide warmings every year or two. The increased temperature of the warm pool presumably reflects "global warming" and seems to be having an impact on the behavior of ENSO.

SST anoms
Figure 1  (Hovmoller plots of SSTs and SSTAs; latest images of various TAO/TRITON SST monitoring tools are here )

Figure 2 (below) is a Hovmoller plot of near equatorial (7.5N-S) OLR anomalies since December 2004.  Their large scale, slow evolution is consistent with the warm-cool-warm SST anomalies seen in Fig. 1. During warm conditions, convection is enhanced near the dateline and the western Indian Ocean but suppressed over Indonesia while during cool conditions it is enhanced and concentrated over Indonesia.  Subseasonal convective activity is superimposed on the ENSO cycle. The MJO contribution to the activity was relatively weak with only three moderate to strong events (~April 2005, January 2006 and September 2006) during this period. The remainder of the variability has eastward propagating characteristics but its recurrence time is faster, typically around 25-30 days. In fact, since the 2001-02 cool season MJO activity has been somewhat unusual with fewer multiple MJOs and more summer activity. The warmth and expansion of the Indo-Pacific warm pool may not only be affecting ENSO but also the MJO.

Of particular note in Fig. 2 is the close sequence of three subseasonal events during March-early May 2006 and an event during June 2006. These were closely linked with important changes in the pattern of Pacific Ocean SST anomalies, as detailed below.

OLR anoms
Figure 2  (Hovmoller plot of near equatorial OLRA for roughly the last 2 years; latest image here; additional plots here )

Figure 3 (below) shows the anomalous zonal wind and SST along the equator for 2006. First we note that the negative anomalies with the weak La Nina did not penetrate west of 160E. As a result, the 29C isotherm extended to 160E providing the potential for deep convection and oceanic forcing of the west Pacific. In fact, substantial convection anomalies reached 160E from the sequence of three subseasonal convective events noted in Fig. 2. The Fig. 3 zonal wind plot shows a weak westerly signal over the west Pacific from the March event, a stronger anomaly over the east Pacific during early April and a subtantial eastward propagating wind anomaly from the May event. A separate analysis indicates the east Pacific wind anomaly was forced from the extratropics as part of the response to convective forcing from the March event. These anomalous winds were associated with a substantial warming of east Pacific SSTs (see first black dotted lines on Fig. 3). The May event was also associated with positive SST tendencies that extended well past the dateline (second set of dotted lines on Fig. 3) and returned the west Pacific to a precursor state more typical of the northern fall before an El Nino. With the 29C isotherm now east of the dateline, the June event led to the first substantial westerly wind burst, which cooled the west Pacific and shifted the maximum positive SST anomalies to the dateline. The stage was now set for further warming of the central-eastern Pacific although its eventual strength could not yet be known.

Finally we note that the rapid sequence of these subseasonal convective events, with little suppressed convection in between, can also be viewed as a single, slow eastward propagating feature, especially if monthly averaging were performed. Nevertheless, two distinct oceanic Kelvin waves were excited by the sequence as shown by the March and May wind events (not shown).

zonal wind and sst anoms
Figure 3.  (Zonal wind (left) and SST (right) anomalies along the equator during 2006.)

Now we review the last year's variations in the global circulation. Figure 4 shows a contour plot of vertical and zonal mean anomalies of relative atmospheric angular momentum (AAM, top panel) and a time series plot of the global mean AAM (bottom panel). Consistent with the ENSO cycle (see bottom panel), AAM is low during the La Nina of 2005-06 and has been increasing since that time. However, a substantial portion of the global increase comes from the stratosphere and is due to westerly phase of the QBO, which peaked in ~July-August 2006 . The top panel shows a prevalence for easterly (or negative) AAM anomalies (green/blue colors) in the tropics and subtropics with some evidence for poleward movement. At the same time, westerly AAM anomalies (yellow/orange colors) developed abruptly in equatorial regions in late January 2006 and have dominated the deep tropics, especially over the northern hemisphere, since late June 2006. Recently the westerly anomalies moved to the northern subtropics in association with the September-October MJO. There are many other cases of poleward movement of anomalies on subseasonal time scales that have been detailed in previous discussions. Eventually, subtropical westerlies are expected to intensify and persist when convection becomes established over the central equatorial Pacific in association with the current El Nino.


Figure 4. (Zonal and vertical mean AAM (top) and global AAM (bottom) anomalies during last year. Contours on top are total field while colors are anomalies.)

We now turn our attention to recent subseasonal variations. Figure 5 is a Hovmoller plot of OLRA for the equatorial (top) and northern tropics (bottom) with coherent OLR modes superimposed. The subseasonal event during June 2006 (mentioned above) is evident at the top of each panel. More recently a MJO developed over the Indian Ocean in early September and moved into the west Pacific region by early October. The MJO initiated surface westerly wind anomalies in excess of 3 m/s west of the dateline during October as it interacted with the warm SSTs in that region. The westerlies forced a strong oceanic Kelvin wave which has since accelerated the development of the El Nino. Additionally, several discrete flare-ups of convection occurred within the MJO envelope as it propagated east, and some of these are numbered on Fig. 5. In the following, tropical-extratropical interactions associated with these flareups are discussed.                


Figure 5 (Outgoing longwave radiation anomalies in two tropical bands. Shading shows 3-day running means while contours are a time-space filtered version of the color shaded field.)

Fig. 6 is a sequence of daily mean 150mb vector wind anomalies during flareup #1 when convection first developed over the Indian Ocean. The number on each plot depicts the location of the flareup, which began on the equator ~ 4 September.  The divergent outflow from the convection interacted with a baroclinic wave train moving through Asia, so by September 10 Rossby wave energy dispersed eastward through an intense anticyclone northeast of Japan to a cyclonic anomaly south of Alaska.  By September 14 additional wave energy tracked across Asia to the now amplifying anticyclone across Alaska.  A low over the Gulf of Alaska four days earlier was forced inland into the USA Pacific Northwest.  At the same time, twin anticyclones and easterly flow on the equator developed over the Indian Ocean with downstream twin cyclones across the west Pacific as the circulation responded to the persistent convection. By the 15th the trough over the Pacific northwest was digging southward along the USA west coast. During this period there were also weaker flare-ups occurring across the warm date line SSTs.  The divergent outflows from the central Pacific forced a subtropical jet that interacted with the western USA trough.  This combination contributed to an outbreak of severe local storms across the central part of the country.  Finally, as the MJO moved east-northeast, wave trains combined across the central Pacific and led to an extreme weather event involving baroclinic cyclogenesis across western Kansas on September 21 (not shown).    


Figure 6  (Daily-average 150 mb vector wind anomalies for the dates shown.)

Fig. 7 is a sequence of maps depicting the total surface wind during October 1-5, approximately the time of flareup #2. MJO convection had now shifted eastward to the west Pacific and twin surface cyclones developed there on 1 October. These low level twin cyclones were the first sign of the westerly wind burst shown on Fig. 5 that subsequently initiated the oceanic Kelvin wave still moving along the east Pacific thermocline at the time of this writing. By 5 October, the northern low had moved poleward toward Japan and interacted with wave energy dispersing from eastern Asia.


Figure 7 (Total surface vector wind for the dates shown.)

Fig. 8 shows the consequences of this interaction using a sequence of 150 mb vector wind anomalies. By 7 October, a strong anticyclone had developed east of Japan and over the subsequent four days  a well defined Rossby wave energy dispersion contributed to a strong positive PNA on 11 October.  This included colder than normal temperatures over much of the central USA and intense early season lake-effect snowstorm for Buffalo, New York. 


Figure 8 (Daily 150 mb vector wind anomalies for the dates shown.)

Finally, Fig. 9 is a sequence of 250mb daily mean vector wind anomalies for October 23-25. By this time the MJO dynamical signal was very weak, and the dominate tropical forcing was from the central Pacific (purple oval).  As flare-up #3 occurred, twin anticyclones responded around 160E (not shown).  A baroclinic wave packet from Asia interacted with them, and resulted in an anticyclonic wave break west of Hawaii by 23 October (shown by dark arrow and purple colored low). As  the "purple low" shifted west another flare-up  occurred by the 24th leading to cross equatorial flow and an intense subtropical jet into the Desert Southwest by 25 October. Meanwhile, wave energy dispersing from the central Pacific contributed to a "digging" cyclonic circulation anomaly along the USA west coast (the orange low).  As the "orange low" moved along the subtropical jet into the central Plains, an early season blizzard occurred across Colorado while roughly 2 dozen tornadoes were reported over southwest Kansas.  Given the rapid evolution of these events, most prediction models did not capture this high impact weather event until about 2 days before it occurred.


  

Figure 9 (Daily mean 250 mb vector wind anomalies for the dates shown.)

Part 2.   Predictive Insights

In these discussions a Global Synoptic-Dynamic Model of subseasonal atmospheric variability (GSDM) is utilized as a framework for real-time monitoring, and to help evaluate model predictions and assist with the preparation of  week 1-3 forecasts.  The GSDM organizes the interactions of four subseasonal time scales into a sequence of repeatable events.  The MJO provides the primary oscillatory component while teleconnection patterns, monthly oscillations and synoptic scale wave energy dispersion contribute additional time scales. The global and zonal earth-atmosphere angular momentum budget (AAM) is used as a dynamical framework for the GSDM.  The GSDM consists of four stages with Stage 1 having low AAM, Stage 2 having a positive AAM tendency, Stage 3 have high AAM and Stage 4 have a negative AAM tendency. GSDM Stages 1 and 3 tend to be more persistent while Stages 2 and 4 are generally transitional.  

Focusing on North America during Northern winter, the synoptics of GSDM Stage 1 include split mid-latitude flows across the North Pacific and Atlantic Oceans and combined flows over Asia and North America, as is also typically observed during La-Nina events.  This situation favors a negative phase of the PNA with western USA troughs and an active southwest flow storm track including possible high impact weather events across the Plains during winter and spring.  During GSDM Stage 2, a west Pacific wave train linked to a large amplitude ridge just off the North American west coast into Alaska is probable.  High impact weather includes outbreaks of bitterly cold Arctic airmasses targeting the central USA.  GSDM Stage 3 is characterized by an anomalously extended and southward displaced jet from East Asia to the west coast of North America, with split flows over the continents.  As during El Nino, a positive phase of the PNA and/or negative TNH is favored.  Often the USA west coast and Deep South can be the target of extreme precipitation (including severe thunderstorms) and high wind events. During GSDM Stage 4 zonal mean westerly flow collapses throughout the subtropics, leaving individual subtropical jets, including one which impacts the USA Desert Southwest.  Locations such as Arizona into Colorado may receive excessive precipitation during Stage 4.     

See this link for our accepted MWR paper that discusses the GSDM  (in press)

Now back to the real time situation.

Full disk satellite imagery show two prominent areas of tropical convective forcing at the present time. One is located over the warm SSTs around the equatorial date line (including weak tropical cyclone activity) with the other larger region centered over the IO at ~5S/60E but extending from southern Africa into northwestern Indonesia. The latest 3-day averaged OLRA are ~minus 70-90 W/m**2 for both areas. Based on animations of 150mb and 250mb daily mean vector wind anomalies (ERSL/PSD), the IO forcing appears to be organizing into an MJO. The IO convection is flanked by quadrapole anomalies of upstream twin subtropical anticyclones (and low level westerlies) and downstream subtropical cyclones, and these features are interacting with the extratropics through wave energy dispersion processes. In fact, the latest Wheeler diagram shows an MJO projection well above 1 sigma and there is also a projection onto the coherent OLR modes.  A phase speed computation has a movement of roughly 5m/s or about 20 deg of longitude during the 5-day period from November 15-20.   

Daily wind animations show that a recent Rossby wave energy dispersion linked to the MJO arcing into the Southern Hemisphere extratropics contributed to the flare-up of the South Pacific tropical convection roughly a week ago. This region of forcing has been expanding back to the west-northwest toward the Philippines during the last few days. Moreover, convectively coupled Kelvin waves emanating from the MJO along with the expanding date line/South Pacific convection have allowed some filling in to occur between the 2 regions (working against suppression).   Twin upper tropospheric anticyclones supporting cross equatorial flow from the Southern Hemisphere are also present around the date line, allowing the linkage of a subtropical jet with the trough currently moving into the western USA. 

GSDM Stage 4 best describes the weather-climate situation for roughly the past couple of weeks. However, a transition to GSDM Stage 1 may be in progress. For instance, animations show the recent Kamchatka block has become  "dislodged" tied to Rossby wave energy dispersion from the IO convection. As shown by most models (but with large differences in the details) the trough which has been present across the Gulf of Alaska is about to deepen into the western USA leading to a greater projection onto the negative phase of the PNA. Arctic air has been building up across Alaska and northwest Canada for the last several weeks, and this trough will have that airmass as a cold air source.  This possibility has been discussed in a Blog (link) maintained by Ed Berry starting November 4th.  

Where we go from here remains tremendously uncertain. There are going to be impacts from tropical convective flare-ups and these are mostly unpredictable by current models. The positive AAM tendency we are now seeing is partly attributable to the current MJO. MJO forcing is expected to consolidate into a large region of intense tropical rainfall extending from near the Philippines to the South Pacific centered ~150E by around week 3. This region may then stall before shifting east-southeast toward the central Pacific/SPCZ by early January 2007. Thus GSDM Stage 1 (with subtropical jets) may be the preferred weather-climate situation through week 2, followed by GSDM Stage 2 for much of  mid-late December. Afterwards, GSDM Stage 3 may appear, which is "typical" during the warm phase of ENSO. In the USA outlooks that follow, confidence is below average for week 1 and then as low as it gets for weeks 2-3.  

Week 1 (26 November -2 December 2006):  GSDM Stage 1 with subtropical jets is most probable. This situation generally favors an active weather regime for the Rockies and Plains, with a southwest-northeast storm track across the central USA. An important issue is how soon deep tropical moisture transport through the Gulf of Mexico can resume after the recent surges of cool dry air. Latest observations suggest low level moistening is occurring across the Gulf of Mexico. Most models predict baroclinic development on the Plains by around the middle of next week, but with still serious phase and amplitude issues. They are "catching-up" to the changing tropical convective forcing and other processes. A thought to offer would be a slower and more amplified solution for trough development, and the models are now trending there. In any case, bitterly cold Arctic air is likely to penetrate into the Rockies and Plains by the end of this period while the Deep South warms up in advance. Depending on the details, portions of the Rockies and Plains may have severe winter weather conditions while severe local storms become a concern for the locations such as the south central states to the Ohio Valley.

There is still an on-going tropical cyclone risk across the central Pacific and possibly other regions (see the Tropical Prediction Center for links). Satellite pictures show evidence of development trying to occur west-northwest of the equatorial date line. Locations such as the Philippines may be impacted later this period. Tropical cyclone Yani is weakening in the area of the South Pacific islands; however, more development may follow. Finally, locations across the Indian Ocean may become at risk for tropical cyclone development as the MJO slowly moves east.       

Please see the CPC Drought Monitor for areas of dryness and the latest official outlooks and statements from Storm Prediction Center not only for severe storms, but also fire weather concerns. Finally, the CPC  CPC USA Hazards Assessment for offers additional insights not only for possible week 1 high impact weather, but week 2 as well.  

Week 2 (3-9 December 2006):  Same as week 1, but with the usual synoptic variations in amplitude. Perhaps another episode of baroclinic development across the Rockies and Plains may occur toward the end of this period. The week 1 storm system will not be the "last western/central USA" trough in this series. In fact, the stronger and slower moving troughs may not occur until the last half of December, particularly if a transition from GSDM Stage 1-2 occurs.

Week 3 (10-16 December 2006):  We may have a period during weeks 4-6 of extremely cold air covering particularly the central USA should a mature GSDM Stage 2 evolve. That may allow a snow pack to build across locations such as the Upper Mississippi Valley and Great Lakes. Given the magnitude of our warm ENSO (and other factors) I would be surprised not to see an anomalously strong combined jet ~30-35N extend from East Asia into the western USA (with split flow across North America) by ~ the middle of January 2007, which would be GSDM Stage 3. This would significantly increase the probability of high impact weather (heavy precipitation, high winds, etc.) for the USA west coast perhaps affecting California the most (other regions for hazards and weather understood).    

Latest CDC Ensemble Forecast

Latest NCEP Ensemble Forecast 

Additional NCEP Ensemble Output 

Latest Canadian Ensemble Output

Latest Deterministic ECMWF Forecasts