Utility companies would like to know now what to expect through the next few months in respect of temperature, precipitation and sunshine. For instance, what is the projection for Manley's Central England precipitation (most often rain). To compute such quantities, assess in a balanced manner all of the relevant cyclically varying or otherwise changing influences, and put them together to yield a projection.

Let us look at James J. McCarthy's Presidential Address "Reflections on: Our Planet and Its Life, Origins, and Futures", Science, 18 Dec 2009, 326, 1646-55. Assessing monthly global average temperature, he compared a time series from "CRU observations" with a computer-model time series "Model: r=0.87": Covering the years ~1980-2008, his Fig. 7A is impressive. In his Fig. 7B are shown the individual contributions from "Solar cycle", "Anthropogenic effects", "ENSO" and "Volcanic aerosols". Together the four influences explained 76% of the variance in the global temperature observations. The squeamish reader is advised to avoid looking at "How will Earth's surface temperature change in future decades?", the original paper from which McCarthy took the graph (Judith L. Lean and David H. Rind, Geophys. Res. Lett., 36, L15708, 15 August 2009). Earth's surface temperature rises inexorably.

McCarthy reported on the El Nino Southern Oscillation which refers to coupled oceanic and atmosphere oscillations along the equatorial Pacific - globally influential because its span in longitude, ~125°E (New Guinea) across to ~80°W (the coast of Ecuador), 155°, is almost a hemisphere. His graph of ENSO shows irregular fluctuations with extremes at intervals of ~5 years. For a regional quantity (difficult) rather than a global quantity (easy) - a slightly different selection might be made: -

(1) Solar activity leads. This was the subject of many interesting exchanges and comments in the pages of Physics Today of March and October 2008, January and November 2009. Nicola Scafetta and Bruce West criticized the conclusions of the Intergovernmental Panel on Climate Change of 2007. The establishment figures Philip Duffy, Benjamin Santer and Tom Wigley had the last words in the Physics Today arguments. The calculation of an index of solar influence remains interesting. An overlooked review by Robert G. Currie, August 1995, "Cyclic signals in Lake Saki varves", Int. J. Climatol., 15, 8, 893-917, is discursive and contains interesting philosophy forcefully expressed. With an engineering background, Currie was passionate about the application of properly designed digital filters to cyclic variations; he deplored the sort of statistical modelling practiced by Scafetta and West, which is indirectly related to the fractal geometry of nature (compare Benoit B. Mandelbrot who died in October 2010 and is missed). After a prodigious publication record, Currie faded from the scene in ~2000. To extrapolate a computation aimed at producing a time series for a regional climate quantity, would we follow Robert Currie, or Scafetta and West? Within roughly an 11-year sunspot cycle, we appear now to be getting reducing amounts of sunshine.

(2) The steady rise in atmospheric carbon dioxide dominates McCarthy's "Anthropogenic effects".

(3) Volcanic aerosols like the assignation by McCarthy, is a third independent forcing, in the nature of a sudden event or cluster of events, followed by gradual decay.

Atmosphere and ocean interact, so we must monitor oceanic activity. Note that a column of atmosphere weighs 1 kg per square centimetre of surface, the same as a depth of just one metre of ocean. For every kilogram, ocean can store roughly four times as much heat as atmosphere. We should know about oceanic modes of oscillation, with time-scales ranging up to hundreds of years.

(4) A ribbon of anomalies generally stretches diagonally across the north Atlantic from the Bahamas to the British Isles. A line from Portugal to Iceland crosses this ribbon, and variations in atmospheric pressure between these two places shows a North Atlantic Oscillation. This is a prime candidate as a variable to monitor.

(5) An index of ice coverage in the Arctic would probably be good for inclusion in projections for the UK (a simple linear decline ??).

     Stephen Zebiak and Mark Cane wrote a classic paper on "A Model El Niño-Southern Oscillation", Mon. Wea. Rev., 115, 10, October 1987, 2262-78. From one end of the tropical Pacific to the other, they dealt with winds pushing upon the ocean surface, winds converging to yield heating from clouds, and shallow water waves in the ocean because of an upper ocean layer overlying a lower layer. With strokes of genius they sketched complicated processes with simple relationships - delightful. And now we have laptops to do in seconds what took them months - so just program your laptop first and ask questions afterwards; play about with their system and get a feel for why it takes more than a year for a pulse to propagate all the way across the Pacific.

(6) For a first idea for ENSO, one might try the index NINO3.4, the Sea Surface Temperature (SST) anomaly over 10° of latitude and 50° of longitude in the equatorial central Pacific.

(7) Bearing in mind "Convectively Coupled Equatorial Waves" (George Kiladis et al., Rev. Geophys., 47, 2, June 2009, 42pp), indices to describe the amplitude and phase of any current tropical Madden-Julian 30-60 day oscillation might be considered.

(8) Oscillations at the troposphere-stratosphere interface at altitudes around 15 km are associated with the Quasi Biennial Oscillation. The state of this phenomenon, with behaviour partially linked with weather in the troposphere, might be included in a treatment.

Items (4)-(8) present difficulties. A balanced assessment is sought, which captures major forcings at selected scales of time and space. Too much complication must be avoided. Don't just get a better result than anyone else before; provide seasonal outlooks which carry conviction with users of projections.