You’ve heard the com­plaints. You may even have uttered them your­self. “What happened to sum­mer?” “Call this a sum­mer!?” “Bloody Wel­ling­ton sum­mer.” And the half of offi­cial sum­mer that Wel­ling­ton’s had so far is def­in­itely sig­ni­fic­antly less warm than usu­al. But is it really that unusu­al, and are we per­haps expect­ing too much from a Wel­ling­ton sum­mer?

I decided to visu­ally com­pare the tem­per­at­ures for the 2016/17 sum­mer so far against the pre­vi­ous 5 years of data. The chart below shows the tem­per­at­ure every hour at Wel­ling­ton air­port between August 2011 and the end of Novem­ber 2016 as grey dots (they’re trans­par­ent, so they build up over the years to show the most com­mon tem­per­at­ures as dark­er regions). The orange curve shows smoothed aver­age tem­per­at­ures for 2011–2016, and the red dots show every hour of this sum­mer so far, anim­ated by day [tech­nic­al notes are at the end of the post]. Click the anim­a­tion for a lar­ger ver­sion.

summer_anim_jitter

What does this show? It’s just pos­sible to see that this sum­mer is on aver­age a bit cool­er than the last 5 sum­mers, par­tic­u­larly in early Janu­ary, but the dif­fer­ence isn’t strik­ing. We had a few days when tem­per­at­ures plunged into single digits, which is uncom­mon for this time of year, but not unheard of. We’ve only had a couple of days that reached the mid 20s, and we had noth­ing like that scorch­ing Christ­mas Day in 2012 when it nearly reached 30°C.

But I think what’s most notice­able is that the weath­er has been changeable…just as it is every sum­mer, and pretty much all of the time in Wel­ling­ton. We have a mid-lat­it­ude mari­time cli­mate, not a sub­trop­ic­al or a con­tin­ent­al one, so we don’t get extreme dif­fer­ences between sea­sons, or long, reli­able stretches of hot fine weath­er. The chart shows that the dif­fer­ence between one day and the next is often more than the dif­fer­ence between sea­sons. Tem­per­at­ure isn’t the only thing that indic­ates a fine day, and I’ll get around to look­ing at oth­er aspects, but for now here are some detailed charts of Wel­ling­ton cli­mate that show that the sea­son­al dif­fer­ence in windi­ness, cloud­i­ness and pre­cip­it­a­tion aren’t that massive either.

On that basis, some would argue that Wel­ling­ton does­n’t get a prop­er sum­mer at all. But what is “sum­mer” any­way? I have a the­ory that we’ve come to expect (or even demand) sev­er­al months of reli­able barbe­que weath­er, because our media and advert­ising are increas­ingly dom­in­ated by Auck­land, Aus­tralia and South­ern Cali­for­nia, and we’re sold an impossible vis­ion of “sum­mer” as a three-month-long Coke ad fea­tur­ing pretty people cavort­ing on beaches. For much of the world, sum­mers either don’t exist as such (as in the trop­ics) or are as change­able as Wel­ling­ton’s (in the mid-lat­it­ude west­erly belts, par­tic­u­larly on west­ern coasts) or are even less clem­ent (in the high lat­it­udes). If you say that Wel­ling­ton does­n’t get a prop­er sum­mer, you might as well argue that we don’t get a prop­er winter either since we don’t get snowed in for months at a time.

Sum­mer in Wel­ling­ton means longer days, more intense sun­shine, and on aver­age a bit more warmth and a bit less rain, cloud and wind. We still haven’t reached the nor­mal peak of sum­mer, which as the chart shows, is usu­ally in Feb­ru­ary and early March, so we might get some barbe­que weath­er yet. But you can nev­er guar­an­tee any­thing here, and as I sus­pect the next few days will show, we can get pretty much any sort of weath­er at any time of year. For some, that will be a con­stant source of dis­ap­point­ment, but some of us appre­ci­ate the met­eor­o­lo­gic­al dynam­ism. It keeps us alert, and watch­ing the ever-chan­ging skies. You can­’t live here by chance.


Tech­nic­al notes:

The hourly data for my chart comes from this archive of METARs, which only goes back as far as 2011 for Wel­ling­ton at the moment. A longer data­set would enable com­par­is­on to a cli­mat­ic norm, but with just five years we can get a sense of how this sum­mer com­pares to recent memory. Tem­per­at­ures are recor­ded to the nearest degree, but to spread them out on the chart I’ve added some jit­ter to the ver­tic­al dimen­sion. Dates are in UTC (GMT), which will only mat­ter if you’re paus­ing the GIF to look at indi­vidu­al days. The “aver­age” tem­per­at­ure curve isn’t strictly an aver­age, but is cre­ated by apply­ing LOESS smooth­ing to the whole data­set.

Wel­ling­ton air­port isn’t that rep­res­ent­at­ive of many parts of the city and region: it’s win­di­er, cool­er, and lacks the extreme high and low tem­per­at­ures you can get in sheltered val­leys (e.g. Karori or Wainu­io­mata). How­ever, there’s a lot of cal­ib­rated hourly data avail­able for it, and it’s use­ful for show­ing wheth­er there’s any dif­fer­ence between years.

The data was cleaned, trans­formed and plot­ted with the open-source stat­ist­ic­al lan­guage R, mostly using the pack­ages ggplot and ggan­im­ate.