What exactly is the global temperature?  How is it calculated?  And most importantly, from the climate change perspective, how can we know the global temperature is changing?  First, let’s look at how we measure heat.

Thermometers measure the heat energy of objects.  A thermometer uses a sensor, for example, the venerable glass tube filled with mercury, that responds to the thermal energy of its surrounding.  As the surroundings of the glass tube increase/decrease in heat, the mercury expands/contracts within the tube.  Place a scale alongside the tube, and the reading on the scale tells you how much heat is in the surroundings.  Calibrate the scale against standard temperatures, such as water boiling at 100 degrees Celsius, or with a known accurate thermometer, and the scale readings can be converted into temperature.  The accuracy of the reading tells how closely the measurement corresponds to the real temperature.  If your thermometer reads 51 degrees when a very accurate lab thermometer reads 50.0, the thermometer is accurate to at best one degree.  The precision of the reading tells how much information the scale gives; for example, if the marks on the scale are every two degrees, it is usually possible to read the scale to the nearest degree (on a mark or halfway between two marks).  Finally, the reproducibility of the thermometer tells if the same temperature always results in the same reading.  If the reproducibility is poor, comparisons between different readings become problematic.

If we are measuring air temperature, we must make sure the thermometer is in a neutral setting.  The thermometer can’t be in direct sun, or near buildings which are heated and cooled.  It can’t be shielded by a canopy of trees, and so on.  Every site has some environmental influences, such as that tree canopy, or the nearby city and airport, that have measurable effects on the temperature readings.  Ideally, the thermometer stays in the same place and nothing changes, but in reality, things change.  Trees grow, buildings go up and down, the population of the nearby area grows, the weather station is moved to make way for progress.

If a weather station has been measuring temperatures for a long time, without question a careful scientist would want to adjust the record a bit to account for these local environmental changes.  If one knows the station moved, one can look at the measurements in the time surrounding the move and see if, on average, the readings are just a bit higher or just a bit lower than before.  Similarly, when a new instrument is installed, one can compare before and after readings to see if the old instrument was reading a little lower or a little higher than the new instrument.  There are lots of reasons the longterm temperature record from a station might need some tweaking to give a consistent set of measurements.  The important point from a scientific perspective is that the adjustments are objectively calculated, and everyone knows why and by how much the raw readings were changed.

So now we have a reasonably consistent set of readings from a single station.  To compute a global temperature, ideally we’d place weather stations all over the global, take measurements, average the daily minimum and maximum temperature for each station, then take the daily averages of all stations and compute a global average.  In turn, each day’s global average could be compared against the same time period in prior years, and after collecting decades of readings, scientists would be able to predict the expected global average temperature for a time period.  The difference between that predicted ensemble average and the actual measured ensemble average is the anomaly. The difference between the measured temperature and a long-term average, or reference value, is the anomaly.  A warming climate gives a positive anomaly, and a cooling climate gives a negative anomaly.  Over time, the rate of change in the anomaly tells us if the worldwide climate is tending toward increasing or decreasing temperatures.  Note that the ensemble anomaly can be positive, but decreasing, and vice versa.

Satellites add an interesting variation to this.  Remote reading thermometers calculate the thermal energy in objects by measuring photon emission from the object in one or more wavelengths.  In the case of satellites, microwave emissions from atmospheric oxygen is measured.  Lots of work is necessary to accurately measure atmospheric temperature from a satellite, as Roy Spencer details.  Satellite measurements have lots of advantages over ground station measurements.  A single satellite can see large portions of the earth, the same instrument is used to measure temperatures in many places, and there are fewer tricky environmental changes to content with.  But satellite data only goes back to the 1970s.  However, it does provide a good check of the consistency and reproducibility of ground measurements.

Now we are ready to look at the various global temperature records in the next post.