Abstract
Measuring the Earth’s Temperature
We read a fair amount in the press about global warming. Some of the journalists are reasonably well informed. Others are less so. A few have an agenda driven by their editor or their publisher and try either to disprove or confirm that human beings are affecting the Earth’s climate – mostly the former!
But how do we decide what the temperature of the Earth is today let alone what it has been in the past? Today, there is a reasonably well-established network of worldwide meteorological observatories. However, observatories are only the beginning. We need to think about their geographical distribution and how to combine their results into a single global figure. Roughly two-thirds of the Earth’s land area is in the northern hemisphere and that is where most observations have been made. Historically, this uneven distribution made it difficult to arrive at a representative world average. Today, things are better because ships, remote automatic recording stations and satellite information can help fill in the gaps.
The general approach is to grid surface of the Earth and then to attribute a temperature to each grid square or grid node based on the information that is available in that area. Before that is done, however, allowance has to be made for the local circumstances affecting each observatory. For example, observatories in mountains will tend to show cooler values than those nearer sea-level, and there may be more annual variability. There is also a well-established thermal ‘urban-warming effect’ that reflects energy use within cities, and it means that temperatures measured in the middle of cities are likely to be 1 or 2 °C warmer than sites on the outskirts. Given that the observations are made by different bodies in different countries, there may well be differences in both precision and accuracy. These are likely to be particularly relevant for some of the most remote and thus important stations!
There is therefore scope for argument about the temperature of the Earth today. For questions of climate change, however, the absolute value of temperature is much less important than temperature changes. However, local observatories and met-organisations that provide the data for the global model are a law unto themselves. Without announcing the fact they may simply move to a new location keeping the same name and thus to a different local microclimate. They may change their way of deriving a station temperature, for example, changing from averaging a maximum and minimum to simply quoting midday temperature. And a city may grow up around a station that was previously in the country. As we go back in time, the challenges increase. There are fewer stations, less evenly distributed, and with poorer records of how their observations were made. Finally, we reach the 17th century, when the earliest measurements were made with mercury thermometers. We also know that a variety of physical processes, including intensity of solar radiation and minor perturbations in the Earth’s orbit can generate periodic temperature changes with cyclicities of a decade to tens of decades.
The bottom line is that it is pretty difficult to judge changes in global temperature from a relatively short time series of variably controlled measurements. However, over the last decade or so, the global temperature does not appear to have varied very much. A better indication of global change may be given by changes in sea-level that have been recorded for centuries at ports all over the world and more recently by satellite. Sea level is rising partly through thermal expansion and partly because of the melting of continental ice in Greenland and Antarctica. The oceans appear to damp out much of the high-frequency temperature variation observed in the atmosphere but that is separate story.