Record hottest month

Further updated: All series have now reported.

October 2015 was the hottest month on record globally.  That’s since 1850 — a total of 166 years, or nearly 2000 individual months.

The six popular instrumental global temperature compilations have now reported, and all except HadCRUT4 have October 2015 hottest (HadCRUT misses by 0.02 °C):


Series, vs 1961-1990 October 2015 Previous record in
NASA GISTEMP +0.94 °C +0.87 °C Jan 2007
HadCRUT4 +0.81 °C +0.83 °C Jan 2007
Cowtan & Way Krige+hybrid UAH +0.93 °C +0.83 °C Jan 2007
NOAA NCDC +0.86 °C +0.79 °C Sep 2015
Berkeley Earth +0.95 °C +0.87 °C Jan 2007
Japan Meteorological Agency +0.77 °C +0.75 °C Sep 2015
Average +0.88 °C +0.78 °C Jan 2007

That is of course hottest in relative terms — compared to the long term average for that month of the year (adjusted here to a common 1961-1990 reference interval). The hottest month in absolute terms is unlikely to be any October, because global mean temperatures show a fairly strong annual cycle caused by the large excess of land in the northern hemisphere (land warms more than ocean in summer and cools more in winter). The annual cycle is implicitly removed from all of the instrumental series¹ (they report anomalies), but reasonable estimates are available from reanalysis series (this is NCEP/NCAR version 1):


The warmest month of the year — July — averages about 1.7 °C above October. Unsurprisingly perhaps, the record hottest month in absolute terms turns out to be July 2015, at about 16.1 °C global average temperature².


Coldest month

The coldest month on record in relative terms was January 1861 at 1.1 °C below its 1961-1990 average (-1.1 °C monthly anomaly³). January also happens to be the coldest month in the annual cycle, so January 1861 is also the absolute coldest month in the record, at about 11.1 °C global average temperature.

The coldest and hottest months are here:




  1. The global instrumental series all work with temperature anomalies right from the start. That’s because the spatial coherence of temperature differences from normal is much stronger than that of straight temperatures, so spatial averages computed with anomalies have much higher confidence than averages computed directly from station temperatures. Berkeley Earth does publish some absolute monthly averages (for 1951-1980), which are a little above the NCEP/NCAR estimates, but track them closely.
  2. Applying the average (1961-1990 based) anomaly from the six instrumental series to the NCEP/NCAR V1 1961-1990 global average temperature for July. (In the straight NCEP/NCAR V1 reanalysis results, August 2015 just pips July at 16.15 °C … but the all-time winner there is actually July 2011 at 16.2 °C. The instrumental series probably provide better estimates, at least for now.)
  3. Average anomaly from three of the six series. The others don’t extend back to 1861; they stop near 1900.


2 comments to Record hottest month

  • Richardoz

    Back in the late 1950s when I was learning to fly we learnt about the Standard Atmospheric Lapse Rate (SALR) which is 2°C drop in temperature for each 1000ft gain in altitude. There were warnings of global warming in the 1970s and 80s and I figured that the average snow line in the Snowy Mountains would go up 500ft with every 1­°C increase in the temperature based on the SALR. Unhappily I think that we are seeing the relentless realisation of this this in your snow graphs. I have seen photographs from early in the 20th century of people skating on frozen ponds at Mt Buffalo and Hotel Kosciusko, Diggers Creek. Not possible in recent decades.

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>