Ophir or Ranfurly?

Which is really the coldest place in New Zealand (900 words)

A couple of weeks back I mentioned the controversy that was kindled when Ranfurly, with a low of -25.6 on 17 July 1903, eclipsed Ophir’s claim to fame as the coldest place in New Zealand, with its -21.6 on 3 July 1995. Here I fan those flames a bit ….

Here’s the temperature record for Ranfurly during the time in question. The Met site is actually at Eweburn, about 1.5 km NW of Ranfurly, which operated there from 1896 to 1922. There were no data yet from Ophir. That site opened in 1924.

Date    Tmax(C)    Tmin(C)
1903-07-13     -3.3           -8.9
1903-07-14     -6.7           -8.9       
1903-07-15     -15.6         -20.0
1903-07-16     -20.0         -23.3     
1903-07-17    -21.1        -25.6     
1903-07-18     -12.2         -20.5     
1903-07-19    -3.3           -8.9       
1903-07-20      2.2           -10.0     
1903-07-21     -7.8          -13.3     
1903-07-22     -5.6           -8.9       

The minimum temperature recorded easily surpasses that at Ophir, where the lowest recorded was a comparatively balmy -21.6C. In fact the maximum temperature at Ranfurly that day was close to to Ophir’s record. But is it correct?

One of my colleagues at NIWA, Gregor Macara, looked into this a few years ago. His report - from 2014 - is well worth reading, and I’m glad I took the time to do so before barging in on the debate.

It was definitely cold at the time! Apparently snow cover was about 1 metre deep. Here’s a photo from his report showing snow that month at St Bathans, about 35 km to the north-west (photo credit: F.M. Pyle).

His report concluded that the 1903 record at Ranfurly should stand, but he did mention a few caveats about the accuracy of reporting during that time period, so perhaps it’s not a done deal after all.

The graphs below show the long term temperature records for several sites in the area. The top panel shows the means of the minimum temperatures each July: the coldest month of the year. For that year with the record low in Ranfurly, the mean minimum for July was -9 C, and by far the coldest on record. So, taken at face-value it looks like Ranfurly wins.

But ever since records began in Ophir, their minimum temperatures are generally lower than in Ranfurly, especially in more recent years. In the last 2 decades Ophir’s mean minima have been lower than Ranfurly’s every year except 1 (2005). Ranfurly’s altitude is higher, so I wonder if they - but not Ophir - had snow cover that year. For a period from 1982 to 1998, there were no data from Ophir, and the region was represented by Lauder (about 10 km away). The coldest winter there was in 1995, another year with heavy snow cover throughout the region.

When seen in the context of mean annual temperatures (middle panel), and the January maxima (bottom panel), those early data from Ranfurly do indeed look a bit out of kilter with the rest - unless there was a rather dramatic increase in temperature in the last decade of the 19th century. Any changes due to increasing greenhouse gases are much smaller than that, and are more in line with the gradual increases seen at all sites over the last one hundred years: in the period from 1920 to 2020.

Macara’s report drew attention to some anomalies in the maxima reported for Ranfurly during the period of interest (that was his main caveat). I wanted to compare these early data with data from a nearby location, but the nearest site I could find with reliable data going back to the 1890s is Dunedin - about 100 km to the south-east. The plot below is just like the one above, except that I’ve replaced Lauder’s data with Dunedin’s and expanded the y-axes to fit in the wider ranges of temperatures. I’ve also added two dotted lines which bound the period when the observations were carried out by the observer at Ranfurly who retired in 1917. We should be suspicious about data between the those lines. Gregor was suspicious of just the maxima, but I’m suspicious of both. It seems surprising that 3 of the 4 coldest temperatures at that site over the last 120 years all occurred during his 20-year period of observations. That imbalance is not present in the Dunedin data.

During period from 1896 to 1917, the Ranfurly mean and maxima are clearly too low. It’s not so clear whether the minima are also too low (the mean temperature is defined simply as the average of the max and min values). Unlike Ranfurly, Dunedin’s temperatures in those first 20 years show changes that are entirely consistent with those expected from New Zealand’s seven-station record. Indeed, Dunedin is one of those seven. I should add that no adjustments have been made to the data to account for changes in observing site in my analysis, but all sites selected were from within less than 7 km and I’ve been careful to use only data from similar altitudes at each site.

Dunedin’s mean annual temperatures (middle panel below) also show a remarkable congruence with Alexandra’s. Its seems that the colder winters at Alexandra compared with Dunedin are almost exactly balanced by its warmer summers.

Over the last one hundred years, maximum temperatures have increased by between 1 and 2 C at Ranfurly, Ophir and Alexandra, but any increases in minima are less than 1 C.

My assessment is that while Ranfurly holds the record, the real minimum may be a couple of degrees warmer than currently stated (which would still probably beat the 1995 record of -21.6 C at Ophir). But perhaps Ranfurly wouldn’t hold the record if measurements at Ophir had been available at the time. Minima at the two site generally track closely with each other, with Ophir usually being the lower of the two.

As a loyal Alexandra resident, I’m happy to note that our summer temperatures do tend to be highest of all these sites. In January 2018, the mean maximum was an astonishing 30 C at one site (Marslin St) and 29 C at another (Pioneer Park). More on those differences in a later posting. So it looks like our record is safe. But I note that this analysis omits temperatures from Alexandra Airport, where maximum temperatures are lower than for the other Alexandra sites.