Edward Berry, NWS and Klaus Weickmann, CDC
Section 1 - Climate Evolution Since November 2003
Since our last report of 10 March 2004, additional Madden-Julian Oscillation (MJO) activity has occurred. Periodic convective flareups just west of the equatorial dateline region have also continued due to relatively warm SSTs there. The combination has contributed to episodes of coherent tropical forcing of the atmospheric circulation. Also, starting in June, observational evidence of air-sea coupling across the equatorial central Pacific Ocean basin emerged. In fact, The Climate Prediction Center's latest El Nino/Southern Oscillation (ENSO) diagnostic discussion concludes that a warm ENSO event is now more probable for the boreal fall/winter.
MJOs and Ocean-Atmosphere Interactions
Figure 1 presents a Hovmoller plot of near equatorial (2 deg S-2 deg N) surface zonal wind, SST and 20 deg C isotherm depth for the past 2 years.
Figure 1: (latest
image)
A complementary time-longitude plot showing 3-day average OLRA anomalies and an estimate of the MJO signal along the equator for the past 2 years is shown in Fig. 2
Figure 2: (latest image).
For this discussion, the focus is on the period October 2003 to the present; however, it is informative to compare the current SSTA and OLRA with those during the 2002-03 ENSO warm event. For instance, Fig. 1 shows the current SST and 20C depth anomalies are comparable to those present during August 2002. Likewise the westerly wind anomalies of 4 m/s observed west of the dateline during June 2004 are similar in magnitude to those seen during July 2002 (not shown in Fig. 1).
The line segments on Fig. 2 represent the most recent six MJOs. As typically occurs, they are a diverse group. MJO # 1 had the best-defined envelope of convection anomalies while MJO # 4 had a large atmospheric Kelvin wave component and therefore a faster eastward movement. The eastward penetration of coherent convection anomalies past ~140E also varies from event to event.
The MJOs are reflected in the zonal wind and 20 deg C depth shown in Fig. 1. MJO # 1 crossed the dateline during early January 2004 and the accompanying equatorial westerly wind event (WWE) set off an oceanic Kelvin wave. A brief eastward shift of warm water to ~160W accompanied the MJO. MJO # 2 had a weak wind signal beyond 140E. MJO # 3 produced another strong WWE that also excited an oceanic Kelvin wave. Wind and convection anomalies persisted near 160E after the passage of this event as displayed in Fig. 1 and 2. MJO # 4 induced only weak wind anomalies. Despite MJOs 1-4, SSTA in the central equatorial Pacific generally cooled during the late boreal winter period. For example, the 0.5C isotherm shifted westward from ~160W during early January to ~150E by the end of April 2004.
Fig. 1 shows quite a different pattern of zonal wind anomalies during June 2004 compared to previous WWEs. As MJO # 5 evolved in the Indian Ocean during late May 2004, a WWE developed near 150E and shifted very slowly east to 170E by early July 2004. The 4 m/s contour in Fig. 1 (left panel) shows this evolution clearly. SSTs cooled behind the WWE and warmed ahead of it. The slow eastward movement (<1 m/s) is suggestive of a coupled air-sea interaction. By the end of June there were westerly wind anomalies of 4 m/s just west of the dateline, and another stronger oceanic Kelvin wave was excited. As the westerly winds weakened and shifted east of the dateline in association with MJO # 5, 1 deg C SSTA spread east of the dateline and persist to the present.
Global and zonal mean circulation anomalies
We now turn to the global circulation anomalies that accompanied the MJO activity. Figure 3 presents the time evolution of global relative atmospheric angular momentum (AAM) during the last year
Figure 3: ( latest
image).
MJOs produce a robust signal in global AAM, which is a measure of the westerly flow in the atmosphere. The top panel depicts zonal mean anomalies as a function of latitude and the bottom panel shows the time series of global AAM (see Discussion #3 for some additional information on AAM). In the bottom panel, vertical lines mark the time when MJO convection anomalies were over the western equatorial Pacific Ocean.
Since December 2003, quasi-oscillations are evident in the time series of global AAM. On average, AAM is increasing or near a maximum when MJO convection is located at ~150E. Figure 3 (top panel) shows westerly (easterly) anomalies in equatorial and subtropical regions as the convectively active phase of the MJO shifts into the western Pacific (the Indian Ocean). This is best defined for the first four MJOs.
On the top panel, poleward propagation of zonal mean zonal wind anomalies can be seen, starting with equatorial easterlies around 1 December 2003. The easterlies propagate poleward to ~ 50 deg N by the end of December and are followed by a strong westerly pulse that moves poleward more slowly. There is considerable variability in the strength, coherence and speed of poleward movement of zonal wind anomalies for the different MJOs. Nevertheless, they reflect a broad-scale interaction that can produce zonal index-like variations as they move poleward. For example the poleward propagation of easterly and westerly wind anomalies into the southern hemisphere starting around 1 May contributed to enhanced midlatitude blocking across that winter hemisphere during June and July 2004.
Figure 4 is a time-latitude plot of zonal mean zonal wind anomalies at 150 mb. It is similar to the vertical integral shown in the top panel of Fig. 3. The dashed lines connect westerly wind anomalies as they propagate poleward from equatorial regions and are labeled with the MJO numbers in Fig. 2.
Figure 4
Again, poleward propagation can be seen. The "H"s in early March near 30N and in late June near 30S signify periods of strong anticyclonic flow (including blocks) that are linked to specific MJOs. Note that some of the poleward propagating events persist well beyond the lifetime of one MJO. For example, large westerly anomalies linked with MJO # 1 persisted for at least 90 days. MJO # 2 produces a perturbation in the poleward movement as westerly zonal mean flow shifts south to 35N in the latter half of February 2004. This was a stormy/wet period for much of the USA, including the drought-stricken southwest.
The above figures give the reader a sense of the global and zonal mean circulation variability that occurred during the past 9 months, and attributes at least some of it to the MJO. Other physical/dynamical processes such as transient mountain torques, index cycle like variations, baroclinic wave packets, etc., also contributed. For example, the strong zonal mean flow anomalies in the subtropics around mid-January and March 2004 were partially forced by mid-latitude mountain torques. For such reasons the details of the response of the atmospheric circulation associated with each MJO will be different. In the following subsection some of the differences (noise) and similarities (signal) among the last six MJOs will be illustrated. Those readers interested primarily in the real time situation can skip to the following subsection.
Signal and Noise in recent MJOs (Note: still under construction)
Figures 5 and 6 are plots of three-day averaged 150-mb vector streamline and OLR anomalies (blue negative/red positive). The "H"s and "L"s denote anticyclonic and cyclonic circulations or high and low height anomalies. Individual rows denote different MJO events (numbered on the left, see Fig. 2), while individual columns denote different locations of MJO convection. Not all six MJOs are shown and not all MJOs have an entry for all four convective locations.
Figure 5
The left column on Fig. 5 is when the convectively active phase of the MJO is at ~90E. In general, there are twin subtropical anticyclones with equatorial easterlies across the Indian Ocean and twin cyclones with equatorial westerlies near the dateline. During MJOs # 3 and #4 there are also equatorial convective flareups west of the dateline, reflecting subseasonal activity over the positive SSTA there. This is just one illustration of why the circulation anomalies associated with an individual MJO may deviate from the composite picture, which is an estimate of the MJO signal. The details of the extratropical circulation are also different. For instance, during MJO # 2 and #3 there are easterly anomalies in high northern latitudes reflecting the sudden stratospheric warming that developed in mid-December 2003. In a broad sense, there is a tendency for a trough to be along the North America west coast.
The right column of Fig. 5 is when the MJO convection is at ~130E. Compared to the left column the OLR anomalies are now more spatially coherent and dominate the tropics. The twin subtropical anticyclones cyclones have moved east and generally straddle the convection while the twin cyclones with equatorial westerly anomalies are also further east. MJO #1 has a well-defined Rossby wavetrain emanating from the convection toward the USA while MJO #3 has a large scale zonal pattern across the eastern Pacific featuring a large anticyclone over the Gulf of Alaska and subtropical cyclones to the south. The latter still reflects a large negative mountain torque in early March that acted to decelerate the subtropical flow. In general, for the USA, there is a tendency for a trough to be over western North America.
The left column of Fig. 6 is when MJO convection shifts into the western Pacific. Again, there is the familiar pattern of the divergent outflow response with subtropical anticyclones generally straddling the convection with cyclones along and just east of the dateline. While MJO #1 still has the wavetrain response into North America, there is still zonal symmetry across the eastern Pacific for MJO #3. Also observe the large anticyclones across northeast Canada for MJOs #1 and 2 with one near Greenland for MJO #3.
By the time the convectively active envelope of MJO # 1 reached 170W, an "El-Nino" like base state evolved with twin subtropical cyclones across the Indian Ocean and the twin anticyclones east of the dateline. Observe the nice Rossby wavetrain response into the lower 48 states, with blocking anticyclones over Alaska and Canada. For the USA there was a ridge across the west and a trough in the east (in contrast to when the moist convection was at 140E). MJO # 5, with the convection at 160W, lead to the formation of a wavetrain from the west Pacific favoring a blocking ridge across Alaska and a trough over much of the Rockies and Plains.
Figure 6
One purpose for this section is to illustrate that the signal to noise ratio in the extratropics associated with the MJO is relatively small. This means knowing the phase of the MJO does not by itself translate very simply into information about the future location of a mid-latitude trough or ridge. However, monitoring individual events can provide early warning about the development of tropical-extratropical interactions that lead to significant synoptic events and possibly large scale pattern transitions.
Ridges and troughs in the Pacific-North American region
During the 2003-04 DJF winter season the ridge position in the Pacific/North American region was closely linked to the location of convection anomalies of MJOs #1 and #2. Discussion # 3 , Fig. 2, shows evidence for this relationship. For instance, during the last half of December and early January, the convection with MJO # 1 moved east into the western and central Pacific and the ridge moved east to ~ 70W. The ridge retrograded to ~150W by ~25 January as MJO # 2 was organizing across the South Indian Ocean. It then shifted back east as convection from MJO # 2 moved into the west Pacific. In general, on the MJO time-scale, ridges in the Pacific North American sector tend to retrograde to ~150W as the convectively active envelope shifts from the Indian Ocean to ~110E and tend to progress to ~central North America as convection shifts from ~120E to the dateline. This relationship is best seen during the boreal winter season, and is linked with the extension and retraction of the east Asian jet stream.
After February 2004, the evolution of the ridge in the PNA sector is less
clearly linked with MJOs # 3 and # 4. However, starting around 1 June
2004 tropical convection becomes fairly persistent centered at ~160E.
Convection at this location can excite a west Pacific wavetrain during the
northern winter season as illustrated in Fig. 7. It shows Stage 2 of
a synoptic-dynamic model (SDM) of subseasonal variability that will be described
in more detail in a revamped Experimental MJO Prediction Website.
At this stage tropical forcing over the western Pacific favors a positive
phase of the west Pacific wavetrain (wPw) pattern, which gives an anomalous
ridge just off the North American west coast into Alaska and a trough across
the central USA. During northern winter, the wavetrain is produced
by interactions between the convection and a jet streak coming off of Asia.
Figure 7
Fig. 8 shows the 150 mb vector streamline and OLR anomalies for the month of June 2004. With only a few exceptions, a ridge has persisted at ~130W and a trough at ~90W. This circulation pattern has contributed to the below normal temperatures and above normal rainfall over much of the eastern two-thirds of the USA during the last two months (see Interactive Monthly/Seasonal Composites). The wavetrain across North America resembles that seen in Fig. 7 although the circulation anomalies over the western North Pacific Ocean are considerably more complicated. Nevertheless, there are large negative OLR anomalies over the tropical west Pacific suggesting tropical forcing there is contributing to the circulation anomalies, i.e., they represent a summer version of the wavetrain shown in Fig. 7. This is further supported in the next section where the reappearance of this pattern in the last week (10-16 August 2004) is linked to a convective flareup over the western tropical Pacific associated with MJO # 6.
Section 2 - Present Conditions
A better understanding of the processes that contribute to present conditions can help with interpretation and evaluation of the forecast. The current pattern of circulation anomalies appeared to develop through a transient interaction between extratropical wave energy dispersion and a tropical convective flareup related to MJO # 6. Such interactions help determine the timing of circulation transitions and the possible onset of persistent anomalies.
Fig. 9 shows time-longitude plots of 3-day average (color shading) and space-time filtered (contours) OLR anomalies for a northern tropical and equatorial latitude band. An estimate of the OLR envelope or signal of MJOs #5 and 6 is given by the blue contours. At this time of year the MJOs are better defined in the northern tropics (top panel) than along the equator. While the contours are an estimate of the MJO signal, there are distinctive convective flare-ups (encased in squares) within the envelope that can initiate the circulation response during individual events. From a composite or MJO signal perspective these are considered noise but from a prediction viewpoint are critical for timing transitions in the circulation. The most recent such flareup occurred in the northern tropics from ~1-6 August 2004. It contributed to an eastward shift of MJO #6 convection from ~100E to ~140E. The shift is associated with further tropical-extratropical interaction that leads to the return of the persistent summer pattern.
Figure 9 ( latest equatorial
/ latest northern
tropical )
Fig. 10 shows a six-day sequence of 150 mb vector wind and isotach anomalies. The interaction is best seen at this level. On 2 August, there is a strong cyclonic anomaly east of the Caspian Sea (40N, 70E) while east of the Philippines the convective flareup is intensifying (see Fig. 9). Divergent outflow is evident toward Japan (red arrow) and especially into the southern hemisphere across Australia (not shown). On 3 August the cyclonic anomaly disperses downstream (northern red arrow) while outflow continues from the region of the convective flareup (southern red arrow). Then on 4 August the southerly divergent outflow "phases" with the southerly flow between a low-high couplet around Japan (long red arrow). The couplet amplified partly due the energy dispersing eastward from the Caspian Sea low. The phasing of outflows may be a key indication of important tropical-extratropical interactions. Also at this time, a wavetrain has been established across the North Pacific terminating with a trough just off the U.S. west coast. On 5 August the couplet around Japan amplifies and and takes on a northeast-southwest tilt. The southerly flow extends further north toward the Arctic. The extratropical wavetrain decouples from the convective outflow on 6 August with the anomalous anticyclone propagating to 55N, 180E. A low-high-low anomaly pattern is now established over the USA partially via continued energy dispersion from the Pacific. By 7 August, a low has developed south of Alaska and downstream of the propagating northern anticyclone. Somewhat distorted twin anticyclones (southern hemisphere not shown) are present in the tropics along ~150E depicting a more conventional response to the convective forcing that is still present there.
Figure 10
Finally, on 10 August a major pattern change has occurred (Figure 11).
The east Pacific low is now quasi-stationary, the anticyclone has propagated
to northwest Canada and a ridge has amplified along the U.S. coast. A broad
trough has been established over the U.S. The pattern is reminiscent
of that favored during the northern summer thus far. Mulitple factors
help determine the initiation, phase and amplitude of the wavetrain pattern
including the convective flareup over the western tropical Pacific associated
with MJO #6.
Figure 11
Section 3 - Predictive Insights
Figure 12 presents the 144-hr ensemble mean prediction of 500-mb geopotential heights and anomalies from the CDC and NCEP ensemble systems. The initial conditions are from 0000 UTC Wednesday 11 August 2004 and the forecast is valid 0000 UTC 17 August 2004. Given the appearance of equatorial and subtropical westerly wind anomalies across the western hemisphere, and increasing westerly flow across the North Pacific Ocean, the 144-hr hour predictions were judged to be most useful for the week 1-3 forecast (discussed below).
Figure 12: ( latest
CDC forecast / latest
NCEP forecast )
Both models show a general pattern of a trough at ~ 140W, a ridge from the central Rockies into Alaska, and a downstream trough from Hudson's Bay to the southeast USA. Positive height anomalies are also predicted across much of the polar latitudes. Subsequent integrations of both the CDC and NCEP ensemble from more recent initial conditions (last updated 08/16/2004) continue to support this prediction. However, the models maintain the positive phase of the west Pacific wavetrain pattern (western North American ridge, eastern trough) longer into week 2, before the westerlies penetrate into the USA. This additional information is reflected in the discussion below.
Important issues for the prediction include the evolution of the tropical convection currently centered around 10N, 140E and the continued eastward movement of MJO # 6 along the equator. Unfortunately, statistical and GCM predictions do not offer much guidance about the MJO at this time. The anomalously warm SSTs in region of the equatorial Pacific suggests convective activity with MJO # 6 should persist. This consideration combined with the SDM and the current ensemble predictions suggest a week 2 circulation pattern consisting of split flow across the continents with enhanced polar jet streams across the oceans. Also, blocking at the polar latitudes may enhance the split flow across North America.
The following gives some predictive insight for weeks 1-3. Uncertainty is higher than average, yielding low confidence predictions.
Week 1 (12-18 August): Much of the country east of the Rockies is likely to remain cooler than average (at least negative 1-2 sigma), with above normal rainfall. Contributing to the above normal rainfall for the east will be tropical cyclones Bonnie and Charlie. Much of the far west should continue with above normal temperatures and below normal rainfall. Toward the end of week 1 or early week 2 the trough may come onshore into the west and bring cooler temperatures and some precipitation for the northwest states. High impact weather would include the possibility of severe local storms and flooding rainfall for the south and east (see CPC Hazards Assessment page), and continued fire danger across the western states. Severe local storms may be a concern for portions of the Front Range of the Rockies into the Plains and Great Lakes states.
Week 2 (19-25 August): Particularly after day 10 (21 August), much of the eastern half of the country should warm while parts of the west cool down. The westerly polar jet stream should penetrate well inland into at least the Plains states, with a weak ridge across the southeast. There may be several anomalously strong mobile troughs from the western states into the central USA, which then weaken as they approach the northeastern part of the country. That would suggest a greater than climatological risk for severe local storms/MCS activity across especially the central and northern Plains and Upper Mississippi Valley states, including possible tornadoes. Other details, including tropical cyclone activity, are unclear.
Week 3 (26 August - 1 September): Any predictions at this lead have to
rely on the MJO. Assuming a new MJO develops soon, there should be
another significant convective flare-up west of the dateline by this time.
The result for the PNA region would be a circulation pattern similar to that
observed currently. However, the ridge may be at ~130W and trough ~90W
with another ridge just off the southeast coast of the USA. If this
occurs, anomalously cooler and wetter conditions would again return to the
central and eastern part of the country, with anomalous warm for the far
western states. In addition, there would be the possibility of severe
local storms for much of the Front Range of the Rockies, Plains, Mississippi
and Ohio Valley states.