The book is a tour de force by nearly 40 contributing authors, who aim to provide a manual of techniques for interpreting satellite and radar images of weather systems in mid-latitudes. It is nothing if not comprehensive, with about 900 images and diagrams.

To produce a thorough training aid for students and forecasters, the editors have had to condense a huge amount of information. The result is not an easy read and, to make the most of many of the examples, you will need a sound knowledge of both the events described and basic meteorology. In particular, an understanding of atmospheric physics is needed to interpret the many conceptual models used in the book to translate two-dimensional images into three-dimensional reality.

Armed with this knowledge, it is possible to learn a great deal about the beauty and complexity of weather systems.

Anyone who gets on top of everything in this beautifully produced book is likely to improve their forecasting success rate. The rest of us can simply enjoy the intoxicating mixture of order and chaos, as revealed in the images presented here.

The underlying assumption in my efforts was that the climate had a memory.
This proposition is based on the notion that the slowly varying components
of the climate – notably sea-surface temperatures (SSTs) and the extent
of pack ice – could produce effects which lasted for months, if not years.
There is some evidence of these processes at work in the statistics, which
show greater variance at low frequencies.

A more compelling reason for believing in seasonal forecasts is the
growing evidence that in the tropical Pacific, changes in atmospheric pressure
patterns and SST – the so-called El Nino – are responsible for altering
global weather patterns. These quasi-periodic warming and cooling events
over vast areas of the Pacific are not only closely linked with extreme
weather throughout the tropics, but are also associated with winter temperatures
over North America and winter rainfall over the British Isles. Furthermore,
computer models of El Nino events have demonstrated the capacity to predict
their development many months ahead. So seasonal forecasts, of cold winters,
hot summers and so on, looked an increasingly realistic proposition.

What has changed my mind is recent research at the Met Office. This
has simulated climate variability during the 20th century to see whether
observed variations in SSTs and sea ice make a difference to seasonal weather
patterns. The preliminary conclusions of the Met Office’s meteorologists
are that, beyond the tropics, these supposedly major players in the climate
game are less important than the inherent chaotic variability of the atmosphere.

Using a sophisticated model of the global atmosphere, the group at the
Hadley Centre for Climate Prediction and Research, Bracknell, has incorporated
the historic data for SSTs and sea ice to conduct a series of simulations
of global climate. The historic record is the product of many years work
at the Met Office and consists of global SSTs going back to 1871, together
with data on Arctic ice from 1953 and Antarctic data from 1973 onwards.
Where the sea ice records are incomplete, the Bracknell scientists have
used climatological data.

The results show that for Europe, the use of historic annual figures,
as opposed to longer-term climatological averages, to drive the models
make little difference to the seasonal forecasts. The most important factor
is the initial atmospheric conditions. The El Nino has an impact in the
North Pacific in winter, and may play a minor part in defining the summer
weather over Europe. This suggests that in mid-latitudes the variations
between seasonal patterns from year to year are largely the product of chaotic
variations of the atmosphere. In striking contrast, in the tropics SSTs
dominate the patterns of excessive rainfall and drought. An example of how
the variance in rainfall is defined by SSTs is in northeast Brazil during
the wet season. In the average of four model runs for the period from 1949
to 1985, the correlation coefficient between the simulated and actual rainfall
was extraordinarily high (0.93).

The one additional crumb of comfort for forecasters of seasonal weather
in mid-latitudes is the hint of a signal in the longer term patterns. The
simulation for the winters during the period 1961 to 1970 showed an increased
tendency for stationary high-pressure systems to form in the vicinity of
the Greenwich meridian. These ‘blocking anticyclones’ produce extreme weather
patterns, and this simulation tallies with experience as this decade did
feature some notable winters, of which the extra-ordinary cold one of 1962/63
in Europe, North America and Japan is best known.

This straw, together with the faint signs of El Nino, is all I have
left to clutch on to. So there may be runs of years when the global SSTs
make it slightly more likely that we will get extreme winters. But when
I look at the record of the ups and downs of winters of the 1960s, it fills
me with little confidence of hitting a winning streak.

The good news is that in the tropics the links between SSTs, especially
in the equatorial Pacific, and rainfall are so clear that reliable seasonal
forecasts look like a realistic prospect. This had major implications for
more than two billion people whose lives in the tropics are dominated by
the success or failure of the seasonal rains.

As for those of us stuck in the turbulent westerly winds from the North
Atlantic, the message is quite clear – do not put your faith in forecasts
of cold winters or hot summers. The latest evidence is that, despite their
apparent appeal, they are built on sand and liable to be swept away by
the ever changing chaotic eddies which make up so much of our weather.

So when it comes to future predictions of seasonal weather, as Richard
Nixon said, you will not have my efforts to kick around. Though, drawing
on this parallel, if meteorologists can get a better handle on any signals
hidden deep in the atmospheric noise then I may be lured back into the
fray. As with so many aspects of life, the temptation to seek to impose
some order on what are largely random events is irresistible.

William Burroughs is author of Weather Cycle, which Cambridge University
Press is publishing in paperback this summer.