Why climate data sometimes need to be adjusted
How do climate scientists identify a real temperature trend? For example, what happens if some temperature measurements were made on top of a hill and other measurements were at sea level?
Measuring sites have been moved over time
NIWA’s “seven-station” temperature series is a composite record put together from more than 30 individual measuring sites in the seven locations: Auckland, Masterton, Wellington, Nelson, Hokitika, Christchurch, and Dunedin.
Measurement sites at each location have been moved around over time, and individual measuring sites have operated for varying lengths of time, but, collectively, these measurements extend back at least to the early 1900s, and form a valuable historical record of New Zealand temperatures.
To calculate a trend over such a long period, it is necessary to merge the raw temperature data from different sites.
Some measuring sites are colder or warmer than others
Sometimes there are climatic differences between the sites which mean that if temperature measurements were taken at both sites over the same period of time, the reading would be consistently different. Reasons why one site might be intrinsically warmer or colder than another include:
- altitude: a higher site is generally colder than a lower site (eg in Wellington, the measuring site at Kelburn is about 120 metres higher than the one at Thorndon)
- exposure to the elements: a more wind-swept site is generally colder than a more protected site
If you join raw data from different sites together, and plot a graph, you might see a “step jump” (fig 1), or discontinuity, when a site change occurs.
That step jump has to be removed before calculating the trend or you would mix a change in temperature over time with a change between sites. You would not be able to tell whether the location had genuinely warmed or cooled because your result would be confused by the effect of shifting the measuring site.
Original raw data are kept secure
When we create a time series using adjusted data, we retain all the original raw data. It remains available on-line in the NIWA climate database so others can conduct their own analysis.
Here are a couple of examples to illustrate why adjustments are necessary: