My research .. and Trans-atlantic connections -- an unexplored potential?

M.T. Cranium

Now that I have finished (better term, brought up to date) my research files for Toronto and NYC, I have a gold mine of information available to do correlation studies of temperature and precip trends on both sides of the Atlantic. NYC is probably the better choice since it would correlate better with all coastal locations for precip, Toronto tends to run a bit out of phase having a stronger correlation with places like Chicago and even St Louis or Minneapolis.

Having been around this weather forum and Net-weather for quite a long time now, I have noticed that in discussions of long-range outlooks there is never much thought given to any trans-Atlantic connection. The important teleconnections are seen to be the arctic and Atlantic oscillations, and even the Pacific oscillation, the MJO (a tropical index) and the usual solar and QBO topics.

All of those are important of course. But I have to wonder, since the jet stream is normally coming across eastern North America and the Atlantic to reach Ireland and Britain, why no real interest in eastern North American weather patterns? People in North American weather forums seem to think that non-random prediction capability exists for that region, so if that's right (and I think it is, although not an impressively high score) then it should add some measure of predictive capacity to European forecasting, if any reliable connections can be made.

Here's what I've found already, and I'm in the early stages of this. About two-thirds of the time, the anomaly sign will be similar on both sides of the ocean. Also about two-thirds of the time, a warm pulse moving through that region of the northeast U.S. will show up in western Europe 3-4 days later. The other one-third of the time, the correlations are negative and these pulses do not cross the ocean, they die out somewhere around the Azores.

That one-third of the time is probably related to or governed by blocking and retrograde weather patterns. So I've been looking for a secondary peak of correlation in the opposite direction, signals that show up first in Europe and later in North America. This time sequence is considerably longer, I have assembled a list of potential cases and the average lag time is 2 months, suggesting that the mechanism is totally related to upper level patterns and has a roughly hemispheric-annual sort of motion (the difference in longitude from Dublin to NYC is about 70 deg which is just under 20% of the earth's circumference, an annual-rotating feature in either direction would require about 2 months).

My weather research led me to construct a theoretical grid for all of these motions and interactions which is a distortion of the terrestrial grid for both latitude and longitude. Its poles are skewed towards magnetic poles, and the lines of "meteo-latitude" are therefore lower than terrestrial from about 30 deg west of the IDL to the middle of the Atlantic and higher than terrestrial in the other half. The peak differentials are over central North America (-10 deg) and Siberia (+10 deg). I would imagine that a working version of this would need to have second-order latitude parameters based on the interference in the free flow of the atmosphere created by the larger mountain ranges.

Then instead of longitude I have a concept of (nine) timing lines. These run generally south-east from the north "meteorological" pole to the south "meteorological" pole, and so they cross the equator about halfway around the world from where they leave these poles. Because I did the research in eastern North America at first, timing line one runs through the Great Lakes region towards South Carolina and then curves gradually around to enter west Africa near Senegal, crosses the equator near Kenya, and begins to curve back south through the Indian Ocean. The Kerguelen Island research station is pretty much right on timing line one as visualized (the south meteorological pole is placed over the Antarctic continent south of Australia even though the south magnetic pole is closer to the coast and is not exactly opposite the NMP, especially in recent years where the NMP has drifted further west-northwest to lie at about 86N 170W. I have not changed my original grid from the 1980 start of research since I think that the atmosphere has not totally responded to this change yet anyway, but if I did change it, the results would be negligible away from a region close to the North Meteorological Pole.

That was a roundabout way of getting to the interesting part, that Ireland is close to timing line three in the system (that one then runs through western France into the central Med, northeast Africa, Saudi Arabia, crosses the equator south of Sri Lanka, and ends up south of Indonesia and eventually east of Australia).

Why nine timing lines? And what is a "timing line?" (what is it timing?).

I will answer those in reverse order. The timing lines attempt to time events in the atmosphere that I believe are responses to external drivers such as the earth's passage through different sectors of the rotating solar-system magnetic field. The original theory (which later research has modified somewhat) was that when the earth was in a given sector, if it had net outward solar flux, timing line one would warm up differentially (because the magnetic field is stronger in that vicinity). Then this warming would attempt to rotate progressively around the hemispheres in the mid-latitudes, a concept that I was able to verify partially by comparing trends in the Toronto and CET temperature records. The lag time was about as expected, 2-3 months. I also identified retrograde features that seemed to move at higher latitudes generally as well as rippling through the tropics (I think this is essentially the same process as the Madden-Julian Oscillation or MJO is tracking).

Also, there was some correlation for the timing line one weather data with lunar signals, suggesting that the Moon does have some influence over the atmosphere. I visualize that as an interference pattern. Oceans have tides because they have boundaries (shorelines) where the tidal energy has to dissipate. The atmosphere has only the partial boundaries of high elevation intrusions and perhaps climate boundaries which tend to be indistinct anyway. So any lunar influence on weather would likely take the form of an interference pattern, the timing lines may be those locations where timing is in sync with the triggering gravitational peak, and places between timing lines are out of phase, although will get the same tidal effect at a different time.

Why nine then? It just happened that map inspection revealed that there were often nine locations for low pressure at predicted times being studied mainly for timing line one (and by about 2005, timing line three since I got heavily into European weather and for that reason mainly). I found that each timing line would produce a similar patterned low pressure complex at these predictable times (such as full moon, new moon, northern and southern declination maxima) but these would be distorted according to what upper level pattern each timing sector was in. A warmer than average ridge pattern near or just east of a timing line would send that impulse further north; a fast flow near-average temperature pattern would concentrate the low at moderate meteo-latitudes, and a split flow or blocking pattern would likely create a split in terms of two different lows on the timing line (but separated NW-SE since the timing lines run that way).

Having nine timing lines means that any lunar-generated waves would require roughly three days to move from timing line to next timing line downstream, some lunar events average 29.53 days and some 27.32 days, and a few others intermediate times, so that the observed motion of low pressure systems seemed to fit this general concept, in that for an energy wave to move once around the hemisphere would require it to advance by about 12 to 13 deg longitude per day (28 x 13 = 364).

The trans-Atlantic connection for prograde warming or cooling sectors should have two components; the faster moving lunar energy peaks and the slower moving solar system magnetic field peaks. Those are on a slower timetable since they persist for 30-60 days at source and will require a "synodic year" to rotate. A synodic year is simply the terrestrial year plus whatever time is required to catch up to a rotating feature. For example, I identified sectors in the solar system magnetic field that were co-rotational with Jupiter. You could picture them as flexed so that the earth would encounter them just before and then 2 months after we pass Jupiter, something that happens every 399 days on average. These field sectors were then found to have corresponding trans-solar components which makes sense because the Sun would be attempting to restore a balance to its energy flux by sending out equal and opposite additional energy in the opposite direction.

The net result of that was a four-peak temperature profile on the time scale of the "J-year" which is the synodic year for earth-Jupiter oppositions. These would then show up in somewhat less prolific form downstream near timing line three. I have also verified that the warmings begin to show up ahead of the time predicted in western North America (timing line nine curves down through the Rockies and western plains states into the Gulf of Mexico, timing line eight is some distance off the west coast of North America then curves down close to the Mexican Pacific coast and into northern South America, etc).

Another finding was that the timing lines seemed to have some degree of east-west oscillation and were not fixed in location although they would rarely exhibit enough motion to get beyond the halfway points between them, more like a quarter of a timing "sector" which is roughly ten degrees of longitude. So timing line three can sometimes ripple east far enough to be into Ireland and sometimes far enough west that Ireland is close to a midpoint between that timing line and the next one (four). This would have some impact on long-range forecasting since places near timing lines are more likely to get the low pressure systems passing in a higher energy phase. This commonly observed tendency for recent winter storm seasons to be less intense than in the 1980s-1990s could be a result of timing line three having drifted some distance further west, there is nothing in the research to say definitively that timing lines are constant in their locations. Also a factor is that meteo-latitude may have shifted so that Ireland is now at a higher meteo-latitude than it was in 1990 (I would think that factor might be 2-3 deg at most).

Does any of this contradict climate change or global warming concepts? Not really, this system could be operating independent of air mass temperatures which would be changing upward due to human activity. Or the net input of heat energy into the atmosphere could be a factor in shifting meteo-latitudes.

The key concept for trans-Atlantic connection would be this -- one would not look due west from Ireland for the best index location, but along the lines of meteo-latitude. As I've indicated, these are lower in North America at given terrestrial latitudes, than in Europe. The terrestrial latitude of NYC is about 41 to 42 deg north, but the differential in meteo-latitude is probably less significant. Toronto is near 44 deg north and may in fact be at a very similar meteo-latitude to the southern regions of Ireland.

What I'm doing next with the research with these expanded data bases is to examine year by year how the two regions respond to postulated energy peaks and what unexplained differentials remain after taking away modelled event temperature profiles. The first set of differentials will be "at source" near timing line one, the model does not perfectly predict temperatures there. Then taking these same predictions with suitable lag times for western Europe, I will be assessing whether or not the results are similar (same unexplained differentials for example), and what lag times are appropriate. If there was a 75% overlap between one signal and the next one downstream, it would advance long-range forecasting since we would already know two months in advance what signal to expect. I don't think it will be that easy. But I am hopeful this is a pathway towards unlocking some of the mystery, and I have to say that probably we have been overlooking the potential of the trans-Atlantic connection because that is after all where a lot of the atmospheric flow originates for this part of the world. Maybe concepts can be developed about how this interacts with already known factors like arctic oscillation. And no doubt a certain amount of Pacific (or SOI) data will be automatically incorporated since North American climates respond to the SOI quite readily.

I opened this up partly to get any feedback, like thoughts you might have on the trans-Atlantic connection, and partly to let you know that work is underway to get this somewhat more operational, my research over a long period of time has suffered from a general lack of time and resources to do it at institutional level speed and complexity, and also from the barrier imposed by complexity, it's a lot to manage for one person with one empty cranium and a home computer.

I have probably sparked some interest in the lunar energy component of this and would want to caution that the results of the research, while interesting, are not some giant eureka sort of moment, the signals are weak compared to the range of variability, although at certain times of year they rise into the low-moderate range. I have already discussed this one or two times, how there are identifiable peaks in temperature and precipitation associated with at least the better phased winter versions of the lunar energy peaks, and an identified pressure trough component at certain times also. But I am almost 100% certain that the Moon's activity is too weak to influence ridge-trough positioning, that seems more likely to have a solar system magnetic field source of natural variability, perhaps driven second-order by other resultant processes like AO and SOI. It's all very complicated of course. But the Moon's contribution seems to be a series of energy peaks moving along at relatively high speed through that background. Other researchers as far back as Bryson and Lamb (mid-20th century climatologists) had looked into the 18.6 year lunar declination cycle and found some climate correlation; I looked at this too and found roughly the same as they had found, namely, a tendency to blocking near the high declination range and a tendency to enhanced fast flow near the low declination years. What causes this cycle? Simply this, the moon's orbital plane is inclined by 5.1 deg to the ecliptic, and the nodes progress rather quickly taking only 18.6 years to rotate around the full circle, so that at intervals that 5.1 deg range is added on to the ecliptic component of declination (what causes the Sun to look high in June and the full moon high in December), so that the declinations range from +29 to -29 deg, and then half a cycle later the reverse is true and the Moon is ranging only from +18 to -18 deg declination. We are currently four years from a declination peak, so that the last "fast flow" peak was around that very mild 2015-16 winter, the next blocking peak will come around 2024. The correlations were far from being 1:1 however and while the 18.6 year cycle left a mark on temperature trends it did not account for more than about 20% of the variability in them.

M.T. Cranium

The exciting part of this research potential is that I keep finding slight improvements on the timing aspects, some of the signals are moderate in size rather than small, and if they all blend into a complex model they have potential to add up to variations well into the two-thirds to three-quarters of actual variability. Of course the rub is this -- are these index values robust enough to continue on into the future, or will slightly better combination techniques just yield another percent or two of accuracy (at the moment, I'm finding about a 30-50 per cent improvement over random chance, the problems are twofold -- one, the climate continues to warm slowly meaning that if you factor that in you are likely to beat random by a considerable percentage so you need a more flexible verification on that aspect, and secondly, there is always some chance that the model, while valid, is not fixed to the terrestrial grid, but is wandering in the same way the magnetic field wanders, so your model has to anticipate that. Variations that took place over central France fifty years ago might be migrating northwest over time and might actually be more helpful to forecasting the here and now than variations pulled out of a study of British or Irish weather trends (which might give a dandy forecast for the waters south of Iceland, etc).

Now that I have NYC and Toronto data bases, I have some useful data sets to study this possibility since NYC is just about directly southeast of Toronto and therefore at the same timing "number" although at a lower meteo-latitude. But what happened there around the 1950s to 1970s might prove to be better for Toronto predictions than what happened in Toronto then (although the two things, like France and Britain-Ireland, have high correlations while not perfect overlap).

I did hope that some people would offer thoughts about the trans-Atlantic connections, we don't seem to think about them very much except that they are somewhat swept up into the other things we do monitor (such as SOI, AO, NAO PMDO, MJO etc).

After a long time looking at the data in both fields, I've come to the conclusion that solar variability is basically only useful for long-term annual or decade scale forecasting and not much use in "finer" time variations.