Real-time Weather-Climate Discussion and Predictive Insights -- August 18, 2006

Edward Berry, NWS and Klaus Weickmann, ESRL/PSD

Since our last discussion (17 May 2006), tropical SSTs have remained generally above average.  Anomalies of +0.5-1.5C can be found from the South Pacific back into the equatorial Indian Ocean (IO), including the Tropical Northwest Pacific (TNWP), and west of equatorial Africa.  Actual SSTs of 29C and warmer are observed across most of these areas.  Recently, ocean temperature anomalies across the eastern Pacific cold tongue have also risen by ~1-2C to depths of ~100m.  Latest data from the TAO bouy array confirms another stronger, oceanic Kelvin wave has crossed the date line and moved into the central Pacific.    

The signal from the Madden-Julian Oscillation (MJO) has remained very weak, as indicated by monitoring tools such as the Wheeler plot (shown here) and the coherent modes Hovmollers (here). Instead, convection has been mostly persistent across the western Pacific Ocean (~150-160E) although a 50-60 day "oscillation" was prominent from late April to late June 2006.  Another recurrent mode of tropical forcing with a ~30 day time scale has been present, especially over the equatorial Indian Ocean.  The seasonal Eastern Hemisphere (EH) monsoon systems have become quite robust, particularly from Southeast Asia into the Phillippines.         

Part I presents an overview of the SST, tropical convective and circulation evolutions since April 2006.  One focus will be on the diagnosis/attribution of an excessive rainfall event along the USA East Coast during the period of June 24-29, 2006.  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 (updated September 9, 2006)

Figure 1 (below) shows two time-longitude sections of near equatorial five-day averaged SST (left) and SSTA (right) in deg. C. The life cycle of a recent central Pacific warm event (denoted by EN) is seen from boreal summer 2004 extending into spring 2005.  During this time, SSTs of 29C and greater, a threshold for supporting persistent tropical convection, extended east of the date line (vertical dash line).  The evolution toward La Nina (LN) is seen from fall 2005 up through February 2006, with 29C and warmer SSTs well to the west of the date line.  During March, surface westerly wind anomalies appeared across the equatorial central and east Pacific linked to an eastward shift of the tropical convective forcing to the west Pacific (discussed below).  These westerly wind anomalies initiated a weak oceanic Kelvin wave-like response, contributing to a warming of the basin.  Similar events occurred during May 2006 and just recently during July 2006.  The first two of these events are indicated by dashed orange lines on Fig. 1.  For the first time in 7 months, the 29C isotherm crossed the date line around May 1. 

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

Figure 2 (below) is a time-longitude plot of the near equatorial (7.5 N-S) anomalous outgoing longwave radiation (OLRA) field since late summer 2004.  Recall we use OLRA as a proxy for deep tropical convection.  The purpose here is to contrast the seasonal evolution of the tropical convective forcing during 2005 with that in 2006. 

In 2005, the stationary response to El-Nino peaked in Feb 2005 and broke down by early March so that by May 2005, the most intense tropical convection had shifted west toward Indonesia and the Indian Ocean (IO) as SSTs near the date line cooled.  This was associated with weakened westerly flow in the upper levels of the subtropical atmosphere during most of 2005.  During 2006, the centroid of tropical convective forcing and warm SST have moved back toward the east since about March 2006.  At the same time the weakened westerly flow of 2005 has shifted into mid-latitudes and been replaced by strengthened westerlies in the tropics during 2006. In the last few months, the climate has tilted toward conditions that favor El Nino warm event, possibly within the next 1-2 seasons, but its strength is still unknown. A possible impact already may be a reduction in the number of tropical cyclones over the Tropical North Atlantic (TNA) thus far this season.       

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

Figure 3 is a time-longitude plot of near equatorial OLRA focusing on roughly the last 6 months. Continuing with the numbering sequence from the discussion posted May 17th, multiple evolutionary behaviors are evident. For example, a strong convectively coupled Kelvin wave developed from the enhanced forcing across the TNA (tropical North Atlantic) during early May (dashed line segment A), and led to event #6 and arguably #7. These included a seasonal northward propagation of OLRA toward India as seen in a time-latitude diagram in Fig. 4.  The green rectangles along ~90E highlight the ~30-day recurring equatorial IO convective flare-ups, while the orange rectangle denotes an event that contributed to forcing the extreme precipitation episode over the eastern USA (discussed below).  The other feature of note is the ~50-60 day interval between strong convective forcing over the west Pacific (~1 May and then again ~ 1 July) and the suppressed conditions in between (10 May-10 June). Partially coinciding with the suppressed conditions over the west Pacific Ocean, Event #4 appeared to initiate a period of enhanced tropical convection across the warm SSTs of the TNA during mid April which remained persistent in this area until roughly mid-June (see dashed purple oval and line centered ~40W).  Such a cycle of persistent forcing "oscillating" between 80-180E and 80W-0W is expected to impact the large scale circulation, in particular the strength of the tropical to mid-latitude westerly flow.

Figure 3.  (Same as Fig. 2 except for the last 6 months)



Figure 4.  (Time-latitude section of the total OLR field for 60-90E).

The slowly evolving 50-60 day tropical convective forcing was associated with variations in the strength of the global westerly flow. Fig. 5 shows the global integral of relative atmospheric angular momentum, its global tendency, and the frictional and mountain torque, which force the total AAM. The times when tropical convection was active-suppressed-active over the western Pacific are shown as orange-green-orange shading in the first panel. The second panel shows that these instances correspond roughly with a positive-negative-positive relative AAM tendency. A similar relationship occurs during an MJO when convection oscillates over the west Pacific Ocean. Moreover, the global torques, which are seen in the next two panels, have the friction torque leading the mountain torque, just as with an MJO. This relation is highlighted with downward pointing arrows that connect torque anomalies of the same sign. While this phase relation is as expected, the magnitude of the torques tend to be more comparable during an MJO.  Here the mountain torque curve clearly includes shorter time scales as well as larger excursions compared to the frictional torque. These are generally induced by mobile mid-latitude wavetrains, which can be influenced by tropical convection but have faster inherent time scales. (Such interactions are generally averaged out when compositing over many MJOs.)  Nevertheless, the atmosphere seeks to maintain an AAM balance in the face of a large mountain torque and this is the essence of what we call "the mountain-frictional torque index cycle". The same wavetrains that induce the mountain torque transport momentum meridionally into adjacent latitude bands where it reaches the surface and is removed from the atmosphere by the frictional torque.  This process is represented by the upward pointing arrows that connect the mountain torque to an opposite sign of the frictional torque.  (During northern winter, the surface wind field associated with the PNA teleconnection pattern plays a prominent role in this removal or rebalancing process.) In summary, these curves represent a mixture of mid-latitude and tropical forcing processes that can become coupled via tropical-extratropical interactions.  For example, a transient interaction between the convection and mid-latitude wavetrains can produce a mountain torque while a mountain torque induced fluctuation in the tradewinds can feed back on tropical convection.  The two phenomena or processes (i.e., mountain and tropical convective forcing) are inexorably intertwined and difficult to disentangle in individual events without careful monitoring. Nevertheless, the curve in panel 1 shows a fairly simple variation in global AAM.

Figure 5. (The global integral of relative atmospheric angular momentum (AAM), the relative AAM tendency, the frictional torque and the mountain torque. The latter also shows contributions from various mountain ranges.)

A much more complicated picture emerges if one wants to regionalize these global anomalies. As a start, Fig. 6 shows the zonal and vertical mean relative AAM anomalies, and the global anomaly curve repeated underneath. Comparing the panels shows the primary contribution to the global signal comes from the tropics and subtropics with easterly anomalies present there during low global AAM and westerly flow anomalies during high global AAM.  The top panel also includes a well-defined poleward propagating signal, especially in easterly wind anomalies during late May through June 2006.  As the easterly anomalies move poleward they are replaced by westerly anomalies in the northern subtropics. Fig. 6 also has vertical lines that mark a period 16-30 June that brackets the USA east coast precipitation event, which occurred shortly after easterly flow anomalies shifted poleward to ~35N. Before examining the daily maps from this period, the zonal and vertically integrated AAM budget will be discussed.  Readers uninterested in these more technical insights can skip to Fig. 8.

Figure 6 (Plots of, top panel: vertically and zonally averaged tropospheric AAM anomalies; bottom: globally averaged AAM anomaly.  Latest reanalysis plots here and here; additional plots here.  See text for details.)

The relative AAM tendency and the terms that contribute to it (the flux convergence of AAM transport, the frictional torque and the mountain torque) are displayed in Fig. 7.  The annotations are based on prominent features in the relative AAM tendency plot. Each panel has the same arrow annotations with "pos" or "neg" starting on the equator and giving the sign of the poleward propagating signal.   At first (and second) glance the pictures are complex and challenging to interpret despite our having applied a 5-day running time mean and a zonal spatial  mean.  The best way to understand the features is to review the regional processes that give the AAM signal during an MJO or mountain-frictional torque index cycle. As indicated previously, both of these processes produce significant variance in the subseasonal band although the latter is characterized by a decay time scale and not an oscillation period.

To help focus the discussion of the budget we will consider only the time period during June 2006 when the convection goes from being inactive to being active over the western Pacific Ocean (see the shaded rectangles in panel 4) and bands of negative and positive tendency "jump-step" poleward into the northern hemisphere (panel 1). Visually the flux convergence of AAM transport makes the largest contribution to the tendency term in Fig. 7, and even more so since the color contour interval in panel 2 is twice that of the other panels. Much of the poleward propagation is due to systematic momentum transports with the transporting phenomena ranging from divergent circulations to Rossby wave dispersions to baroclinic wave processes.  The large momentum sink centered ~30N early in the period (along first dotted line) is followed by a strong surge in the tradewinds over the western hemisphere implying a large positive frictional torque (panel 3, orange shading).  As the trades weaken, the mountain torque starts increasing as convection becomes established over the west-central Pacific. Northerly meridional flows off of Asia support high pressure and a positive mountain torque from Asia, while easterly low level inflow into the west Pacific convection projects on an atmospheric Kelvin wave whose sea level pressure signal propagates rapidly east and gives a positive mountain torque from the Andes.   Thus the torques help "flesh out" the poleward propagation but it is mostly a dynamical  feature related to the momentum transports. We should note that the Coriolis and gravity wave drag torques are part of the vertically integrated AAM budget but are not shown. The former can be large and partially compensate the mountain torque. It is related to changes in the global mass distribution that accompany the mountain torque. Also, the zonal budget is not balanced using the NCEP/NCAR Reanalysis data, i.e., the sum of panels 2-4 plus the Coriolis and gravity wave drag torque is not equal to panel 1. On the other hand, the global budget (Fig. 5) is a better balanced field suggesting problems in the meridional structure of the torque and/or AAM flux divergence fields.

The specific feature of interest for the east coast precipitation event is the strong trade flow and positive frictional torque that develops around 10-25 June in  association with the poleward movement of negative AAM tendencies and a following band of positive tendencies. As already noted, this is associated with a 50-60 day oscillation in west Pacific tropical convection combined with a mountain-friction torque index cycle. However, there are other time scales involved in the event including: 2) ~30-day tropical convective variability, and 3) fast baroclinic wave /Rossby wave energy dispersions.

The 30-day flareup over the Indian Ocean (see orange box on Fig. 3) essentially indicated that convection had returned to the equatorial eastern hemisphere and started the process of changing the downstream atmospheric circulation anomalies. However, the connection between the tropical convection and the circulation changes over North America is fraught with "starts",  "pauses" and local energy sources along the way.  The wave energy disperses, gets trapped locally, disperses again and then gets involved in a major baroclinic development around 160W over the Pacific Ocean.  This all occurs during 16-30 June and provides the large scale dynamical forcing for the rain event.

Figure 7 (The zonal and vertical integral of the physical processes making up the atmospheric angular momentum budget.  A 5-day running mean has been applied to the data. The annotations emphasize episodes of  poleward propagation of zonal AAM anomalies. See text.)

To summarize the big picture during 16-30 June, zonal mean easterly flow anomalies move further north to 35N at the start of the period followed by a northward shift of westerly anomalous flow to 35N after 30 June. This change is primarily accomplished by a quadrapole structure of the AAM transport anomaly (see rectange in panel 2, Fig. 7) with some assist from the mountain torque (especially around 16 June).  The frictional torque shows a lagged "response", consistent with adjustment to an upper level momentum source.  A further "big picture" event during 16-30 June 2006 is the increase in westerly flow that develops in the northern subtropics. This is consistent with convection increasing again over the eastern hemisphere including the equatorial west Pacific Ocean. The increase is forced about equally by the frictional torque and the transports, a relationship quite different from the one at 35N and 50N (the northern portion of the transport quadrapole). This suggests a link between an equatorial-subtropical response to tropical convective forcing and a mid-high latitude response to organized eddy transports (the most likely contributor to the quadrapole).  Note also that during June-July the flux convergence of AAM in panel 2 is quite persistent near 35N undergoing a sink-source-sink cycle over about 30 days.  Its relationship with the 30 day oscillation in convection over the IO is unclear but on average these transport variations are "white noise" with 1-2 day decay time scales. Nevertheless, transports can be persistent during individual events and produce large signals that impact the circulation. We now examine the behavior of baroclinic waves and wave packets during the 16-30 June period. The big picture discussed in connection with Figs. 5-7 should be kept in mind.

Figure 8 is a sequence of daily mean maps of 250mb vector wind anomalies designed to illustrate the synoptic details leading to the excessive rainfall event along the USA East Coast.  Nevertheless,  The sequence begins with June 16th, about the time of the central IO convective flare-up and when AAM tendency was maximized on the equator (Fig.7).  In general, the dashed orange ovals denote the IO flare-up while the larger purple oval depicts the western Pacific tropical convective forcing.  The red H's and L's indicate anticyclonic and cyclonic circulation anomalies, respectively.  The orange lines for both June 17 and 24 indicate the approximate locations of a trough axis and the dashed purple arrows emphasize key wind field structures.

On June 16 and 17, a baroclinic wave packet moving rapidly east-southeast through southern Asia interacted with the divergent outflow from the IO tropical convective flare-up.  The extensive meridional flow on 16 June from 10N to 80N is the first indication of a strong tropical-extratropical interaction.  By June 18 the outflow from the now intense IO convection helps produce a jet streak west of Japan (curved purple arrow).  At the same time easterly flow anomalies are being established around 35N and westerly anomalies are appearing in the northern tropics of the eastern hemisphere. During June 19-21 energy propagates downstream over the North Pacific leading to L-H couplet along 145W-135W.  This represents an effective retrogression of the H that was over the southwest USA 17-18 June.  At the same time a large anticyclone just west of the dateline becomes intense and circular with nearly 40 m/s wind anomalies.  By June 21, the subtropical westerlies have expanded considerably from the Indian Ocean to the east Pacific where they feed into the 145W trough over the Gulf of Alaska.  By 23 June, the dateline anticyclone begins to disperse energy eastward leading to amplification of the ridge around 120W. The westerly flow in the subtropics continues to feed into the L-H couplet over the east Pacific-western North American region.  This pattern amplifies further on 25 June and also deepens the downstream trough over the central USA. By 28 June the subtropical westerlies have expanded across the western hemisphere and now appear to feed into the intense trough over eastern Canada and the anticyclone just to the east of New England (southerly wind anomalies in excess of 40m/s in between these systems).  This sequence from 23-28 June contributes the immediate dynamical forcing for the east coast precipitation event. Of course the weak flow aloft near 35N (e.g., June 26) is a reminder of the strong trades at the surface that were instrumental in transporting moisture toward the USA east coast.  These moisture laden winds from the deep Tropical North Atlantic were transported northward and dynamically lifted across the USA East Coast for about 6 days, allowing for tremendous and destructive rainfall.  At upper levels, there was also tropospheric moisture transport from the Pacific Ocean in association with the enhanced westerly flow. As the tropical convective forcing propagated into the central Pacific, the global circulation began a transition to GSDM Stage 3 (see this link for our accepted MWR paper that discusses the GSDM), allowing this East Coast excessive rain synoptic pattern to break down by the end of June.  

Figure 8. (Sequence of daily mean 250mb vector wind anomalies for the period from June 16-30, 2006.  See text for details, and this link for a useful RWD diagnostic from the Tokyo Climate Center .  Please see this link for animations of various fields including the operational 150mb and 250mb daily mean vector wind anomalies)

2.   Predictive Insights  

Since the time period of the sequence shown in Fig. 8, at least 2 other well defined RWDs have occurred. One was from about July 16-21 and the other from July 24-31. The former led to retrogression of the ridge position back toward the western USA and the latter allowed the ridge to shift farther west into the central Pacific, leading to a Pacific Northwest trough and an intense ridge across the Deep South. Figure 9 presents snapshots of these events with the annotations the same as used on Fig. 8.  Both of these situations were responsible for deadly heat waves across large portions of the country. The July 16-21 ridge retrogression was associated with a transport induced momentum sink near 35N (Fig. 7, panels 1 and 2) while 24-31 July was linked to a tropical convective flare-up around the Bay of Bengal, part of the ~30 day mode of tropical forcing.  We now update some of the information presented in Section 1. 

Figure 9.  (Same as Fig. 8 except for the dates shown.  Please see this link here for operational plots of various fields)

Tropical Ocean SSTs remain above average across most of the Western Hemisphere and the Indian Ocean, with cooler than normal values centered on Indonesia. Actual SSTs in excess of 30C have been observed along and just north of the equator ~160E, with 29C and warmer over the TNWP as well as from the East Pacific, Caribbean and into the Gulf of Mexico. Latest TAO data suggests oceanic Kelvin wave number 3 (for 2006) was located around 160W, and that surface westerly wind anomalies (~5m/s) have once again returned to the equatorial date line region.  Latest prediction from CPC expects a 50/50 chance of either a weak warm event or ENSO-neutral conditions to

prevail during the next 3-6 months (see latest TAO data here, ESRL/PSD data here, CPC data here).

The MJO remains very weak, and there is some evidence of weak air-sea coupling between the tropical convection and warm SSTs west of the date line. An eastward propagating dynamical signal has moved into the Western Hemisphere, moving along the East Pacific ITCZ (phase speed ~8-10m/s). Finally, consolidation of tropical forcing is taking place centered ~10-15N/110E that involves the Asian monsoon systems and an intense flare-up just north of the equator centered ~60E (see link to coherent modes Hovmollers).  Consistent with previous events during the last several months, we expect Indian Ocean flare-up to propagate northeast and also merge with the on-going convective activity from India to Southeast Asia. This whole region of enhancement could easily shift into the TNWP weeks 2-3 maintaining the tropical cyclone concern for that region (see link  for the latest Western Pacific tropical cyclone information). 

While tropical convection should persist west of the date line (SST boundary forcing), general suppression may occur around Indonesia during weeks 1 and 2. The TNWP forcing could reinvigorate the SST boundary forced convection west of the dateline thereafter. Finally, the Western Hemisphere dynamical signal may propagate through the Americas into the Atlantic by weeks 2-3, creating a tropical cyclone hazard for the East Pacific and perhaps the North Atlantic later on. The latter will need careful monitoring, especially for hybrid systems if the environment of the deep tropical Atlantic remains unfavorable (see link for latest Atlantic tropical cyclone information).  As would be expected, statistical and numerical models of the MJO (see ESRL/PSD MJO tools , BMRC MJO tools, CPC MJO tools) offer little useful predictive information.

Zonal mean westerly wind anomalies should continue to increase from 15S-25N, with magnitudes at 200mb ~5m/s at 15N. Animations of 150mb and 250mb daily mean vector wind anomalies (animations here; the reader can naviagate to the reanalysis data animations) shows the appearance of twin upper tropospheric anticyclones around the date line within the tropics, with distorted twin subtropical anticyclones ~120E. These anticyclones (with lower tropospheric cyclones) are the result of the divergent outflows from the tropical forcing discussed above. The pair at the date line is a response to the SST boundary forcing (and perhaps an ENSO signal due to an evolving possible warm event; a slower process than subseasonal), and are contributing to the recent increase of zonal mean anomalous westerly flow throughout the tropical and subtropical atmospheres. Some of the anomalous westerly flow is already impacting not only the Tropical North Atlantic, but also enhancing moisture transport into the Desert Southwest of the USA.

.

GSDM Stage 1-2 best describes the global circulation now (see links to AAM plots here). A transition to GSDM Stage 2 would be most probable by week 2, possibly persisting into week 3 although uncertainty remains extremely high.  This would suggest retrogression of a ridge back into the Great Basin and perhaps into Alaska, with a central and eastern North American trough. Most ensemble numerical models support this notion for week 2. However, this transition has been suggested for at least 2 weeks in a regular blog maintained by one of us (E. Berry, see link).  We will see what happens.

Figure 10.  (Week 2 calibtrated tercile probabilities for temperature from the CDC ensemble; forecast )

Figure 11.  (Same as 10 but for precipitation; forecast)

Week 1 (19-25 August 2006):  GSDM Stage 1 transitioning to Stage 2 seems probable.  The expected trend would be for well above normal temperatures to shift back into the Great Basin and Pacific Northwest while portions of the East cool down.  A relatively active MCS track is probable from the Northern/Central Rockies into perhaps the Ohio Valley (accompanying the synoptic systems) and portions of the East Coast.  Depending on the ridge position, the Southwest USA Monsoon may remain active.  Many parts of the lower 48 states are experiencing drought, particularly from the High Plains into the Rockies.  Little significant relief from this situation seems likely not only this period, but perhaps for several more weeks.  While there may be a "spin-up" particularly from a hybrid system across the Tropical North Atlantic, tropical cyclogenesis looks to be suppressed (see the Tropical Prediction Center for details) there.        

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 (26 August -1 September 2006): GSDM Stage 2 is probable. Please see Figs. 11-12.  A modification would be to suggest a shift toward below normal temperatures from perhaps the Upper Mississippi Valley/Great Lakes possibly into the Northeast with a stronger tilt toward the above normal tercile from the Great Basin to the Pacific Northwest.  Near normal may be most probable elsewhere.  Much of the country is likely to remain dryer than normal, especially from the High Plains into most of the Rockies and western states.  An area of near to above normal rainfall may be probable for the Desert Southwest and the Mid Atlantic-New England.  Tropical cyclone activity may remain suppressed across the North Atlantic.   

Week 3 (2-8 September 2006):  Per above, unclear.    

Latest CDC Ensemble Forecast

Latest NCEP Ensemble Forecast

Additional NCEP Ensemble Output

Latest Canadian Ensemble Output

Latest Deterministic ECMWF Forecasts