Temperatures in urban areas are higher than in nearby lowly populated locations because of the heat generated by man’s activities.

Temperature is measured in a Stephenson Screen, a slatted box of standard design at a standard height in surroundings that must fulfil certain requirements. Daily temperature is taken as the mean of maximum and minimum. Max and min can be measured by thermometers with sliders; the sliders are pushed by the mercury as it expands or contracts and left stranded at the extremes.

Estimates of global mean annual temperature are made by dividing the world into areas and estimating the temperature for each from the records of the operating stations in each. I am not clear how this is done. Height affects temperature. Are all stations reduced to sea level by applying the isothermal lapse rate? The temperature is that of the world as it is not some hypothetical flat world. How does one allow for nearness to the sea? Cornwall is different from say Nottinghamshire. If we just take the readings as they are how do we accommodate new stations? If the coast is warmer than the inland mountains ,it is not fair to incorporate new stations if they are set up only along the coast.

The “heat island” shows these problems most clearly. The obvious way to estimate the heat island effect is to compare the city return with that – or better those – from nearby rural stations. But then we cannot use the city return as a reading in its own right since it depends solely on the rural reading.

If Tc = temperature in the city and Tr = rural temperature and Tmc the modified city temperature we have:-

Tc- Tr = diff Tmc= Tc – diff Tmc= Tc-(Tc-Tr) Tmc= Tr

I suspect that is what has been done but in so roundabout a way that the researchers have managed to disguise from themselves what they are up to.

The only other way is to estimate the heat generated and the effect that would have on temperature. That I would have thought effectively impossible. Estimating the heat produced say by greater London would be out of the question with any accuracy. The amount would vary with time of day and with ambient temperature; less on a hot summer’s day than in the depth of winter, more at 07.00-09.00h than at 03.00 to 05.00h.Then from how far does heat affect the reading? The effect clearly decreases with distance but to what degree? Is it the first 100m that is critical? Does the effect persist to some extent for 10 km? Then, when the wind is strong, the effect will be dispersed. But on a still cold night the inversion that will build up in a rural area may be prevented by heat generated close to the ground in the city. It is all fiendishly complicated; a multitude of factors all difficult to quantify.

But excluding sites where the heat island effect might be significant would cut down the number for analysis –possibly decimate. On the one, had new sites in areas where records have been few- third world- are likely to be set up in or close to towns. Then towns have been spreading at a great rate so sites that were once rural no longer are. In the third world towns have spread so rapidly that sites that a few years ago might have been in the country now no longer are.

Greenwich in the mid 19^th Century, which is the start of the analysis period – the base period is from memory 1868 -1892- was then in the country. Greenwich is I suspect the first station to record consistently in a standard way. It isn’t in the country now.

What is actually done to allow for the heat island I do not know. Some years ago I got hold of a paper by Jones et al (1999); Philip Jones is the Director of the Climate Research Unit at the centre of Climategate- Jones has been forced to stand down by the way. The paper said simply that the heat island effect had been allowed for and gave 2 references. One was to an Australian Ph D thesis by someone then on the staff at CRU and the other to a paper in an obscure Swedish journal. I got the latter from the Met Office library and read it 5 times without detecting even a mention of the heat island.

(I suspect this practice is more common than one might suppose. Who checks references unless they are about what you are working on? Most references are I suspect just put in to make the paper look impressive. Others are to shift responsibility for a dodgy figure to someone else; I did that once myself[1]. Some are even circular. One by dear old Vernon Joyce now no longer with us, referred to a paper that I was surprised, knowing the authors, had dealt with the matter. When I looked up that paper it gave the authority for the figure as V Joyce pers com. – personal communication!)

I have an uncomfortable feeling that we ought to restrict estimates of change in temperature to stations that have operated throughout the analysis period and whose surroundings have not changed much- have not got swallowed up in towns. That would not make for an impressive number or a comprehensive cover of the world.

Dec 2009