Whether or not to shoot the messenger
SUNDAY JANUARY 24, 2010
The weather-cock on the church spire, though made of iron would, said 19th century German poet Heinrich Heine, soon be broken by the storm-wind if it did not understand the noble art of turning to every wind. Nine-tenths of people couldn’t start a conversation if the weather didn’t change once in a while, and the same can be said for a forecast that goes bottom-up. With that in mind I will address what seemed like a wrong prediction made for Ashburton last December. Some have said an opposite occurred, with double the monthly rainfall average falling instead of half. Did the method fail? Prior months had attracted no criticism presumably because the prognostications were reasonably okay. These include p41 in the 2006 Predict Weather Almanac that warned of the cold wintry blast about 12 June, and unsettled North Island weather this January, part of a summer that was going to be long in arriving. So perhaps an overhaul of the method may be postponed, temporarily at least. But if wintry weather of 2006 was predictable 18 months in advance, again, why was December way out?
Weather forecasting is not really a 'science' in the way bridge-building and chemistry are, because it is opinion based, blending available information into accumulated experience, just as commentators do in economics or mental health or political science. As political commentators do occasionally get elections wrong, and doctors can make wrong diagnoses, so it is at times with weather forecasts. Skeptics will always be skeptics, and will selectively focus to prove their case, and this is especially true of rivals in the same profession who wish to sustain a competitive edge. For them no explanation will suffice because they have a mindset to find fault in something they do not understand. But readers with no axe to grind other than an annoyance of unexpected weather arriving when they have chosen wrong clothes or have made wrong farm decisions and who may have come to rely on a system are owed an explanation. Besides, this provides a good opportunity to discuss some of the pitfalls in this business of prediction.
Whilst not wanting to look like weaseling out of a sticky situation it must be said that if any product is to be reviewed it is only fair to note the actual claims made by the manufacturer. Review should then be based around how much of what was promised was delivered. With the moon method there is never any claim to get every forecast right 100% of the time, for no one has all the answers, but about 80-85% always seems reasonable, which is what most metservices claim. That means out of a 12-month year there is an allowance for two months of unavoidable error. Was December one of those months?
Mainly the moon method determines the timing of rain, rather than amounts. The daily position of the moon dictates the changing height of the air and this results in the airtide. The air varies in altitude and by its changing height sets up colder and warmer fronts by letting in more or less of the sun’s daily heat and more or less of the nightly cold from space, and these control condensation times. But no matter how much I say it, there will be some who will hold me to account on rain amounts, failing to recognise that the size of the rainfall depends on the quantity of water vapour drawn upwards by prior evaporation, a function of intensity of the heat from the sun.
Looking back at timing, the rain days as written in 2006 Predict Weather Almanac on p363 are summed up for S Canterbury during last December as being the 3rd, 6th, 9th-12th, 17th, 25th and 29th. As a location that is representative of the district let's look at when rain fell in Rangitata. Metservice records from Environment Canterbury indicate actual rain came on 3rd, 6th, 8th, 11th-13th, 16th, 25th, 28th and 29th. This can be checked by anyone choosing to visit NIWA’s historical records.
Rain for Ashburton came on 3rd, 4th, 7th-9th, 11th-12th, 19th, 28th and 29th. So the potential for rain was around these times, and some certainly came. There can be no rain without potential, and never rain when there is no potential, but whether or not rain eventuates from potential and how much, is variable. I maintain that the moon-method remains intact because rain fell when it was supposed to.
The rate of weather changeability is dependent on moon-speed relative to Earth. Moon-speed changes come in three ways, faster in new moon phase, faster in closer moon-earth distance and faster around lunar equinox (moon rising due east). Consequently there are opportunities for the skewing of predictions during these periods and for this skewing to last up to 2-3 days. Further, when peaks of these three independent lunar cycles combine, a skew may last the best part of a week and the sequence of isobaric maps may also be delayed accordingly. A few days mid-month won’t affect the overall month rain quota but a month's predicted estimates can be way out if a bundle of expected rain falls slightly before the change from one (solar) calendar month to another, or late enough to be registered as part of a new month's rain amount. The 60mm that fell on Dec 29 is a case in point. Taking that out would have made a considerable difference to December's rain amount figure because Ashburton would then have registered only 80mm for December (less than 10mm outside the monthly average) instead of 160mm.
Another reality is that a weather event can be generated anywhere between 100m and about 35 kms up, and can be dropped from 10-15kms, which means that between dropping time and ground arrival, depending on wind strengths, there can be a potential overshoot of rain by up to 80-90 kms. This is the radial distance from Ashburton describing a circle containing Studholme, Mokikihi, Fairlie, Mt Arrowsmith, Arthurs Pass, Rangiora and points within. So if a forecast of heavy rain for, say, Fairlie, instead lands in Ashburton, given the tools available it is as close as one can reasonably get, and could be considered a fairly successful forecast. Locals know that a winter thunderstorm in Christchurch usually means snow in Fairlie. For coastal locations this 80km error can mean a rain dump might land out to sea leaving a town dry. Ashburton received heavy rain for December but not so for Timaru. Well within this error range, Timaru only received average December amounts.
A glance at actual figures shows clearly that over the 3 months prior to this writing (8 January) in a geographic circle centred over Ashburton, there was a variety of rain amounts. To the south Opihi received 240mm and Hadlow 246mm, but almost double fell in both Geraldine (400mm) and Mt Somers (413mm). More to the north over the same period Okuku School received 206mm, Halswell 214mm, Ahuriri 246mm but Arthurs Pass 1946mm. Even Rangitata on 331mm differed from the 306mm recorded over the same timeframe at the Winchmore Station. In an area that goes from craggy mountain to coastal plain, the variation within this error circle was fairly considerable.
If the accuracy of moon-weather is to be compared with the track record of other metservices, then we should compare parallel predictions from about a year and a half away. My forecast for December 2006 was on my website database for four years before October 2004 which was when I compiled the 2006 almanac. The question is whether or not this long range method provides something useful that therefore makes it worthwhile. I repeat that I think the timing of rain is the likeliest factor. Those who do work with it find how to use the method properly and will gain from it. I recommend the use of farmers’ own daily records, stretching back one, two, three, four or however many moon cycles you have the figures for. In the fourth week of December 1988 and again in the same week in 1989 (but not 1987 or 1990) sudden downpours came to Methven and surrounding districts, including Ashburton, and that was one cycle ago. Two cycles or 36-38 years ago - in the first week of 1970 heavy downpours hit the region. Three cycles ago - on 11 January 1950, Fairlie received 57mm in 3 hours. Four cycles i.e. 72-74 years ago - on 6 January 1934 much heavy rain fell throughout the province. (Methven had 128mm in one day, and Ashburton 52mm in 3 hours. Several roads in the district were impassable due to flood waters). Going back before this, I only have Methven observations. Five cycles - in 1916 on 26 December Methven recorded 37mm. Six cycles? The data begins to get scarce, but on every beat of the 18-19yr cycle heavy rain will probably be there, waiting in farmers’ records to be discovered, such as the rain that came at the end of December of 1895(6 moon cycles of 18.613yrs ), and even half a cycle later at the end of December 1899 - when 85mm fell in Methven in a day.
Historical figures are nowadays stored not with Metservice but held entirely by NIWA who sell them. As anyone who tries will find out, data for only a handful of towns costs over a thousand dollars, putting historical research mostly beyond the means of an outsider. Hence I am always on the lookout for snippets, cuttings and old diary entries. If anyone does have figures for the area and records of floods etc in antiquity, I would love to hear about them. Getting it wrong now and then may not necessarily mean the moon method is flawed, it may just mean I am only as good as my tools.
Coming up for Canterbury I expect a February that may be warm and wet in the first week, wettest and windiest in the third week, but drier in the second and fourth weeks. March should see very changeable weather around 19th/20th, and widespread rain. April goes wet and cool in the second week.
© Ken Ring January 2007