Skip to content

CET Temperature Trends In March

April 7, 2016

By Paul Homewood




As is the case with most months, trends in maximum temperatures in England took a step up in the late 1980s. (Mean temperatures also show a similar picture).

But does this mean that all days are getting warmer? Or that the warmest days are getting hotter or more frequent? 



I have done some analysis on the CET daily series, which goes back to 1878, and its produces some surprising results.

The average March maximum temperature over the series is about 9.3C, and I have calculated the number of days each year that are over 13C and under 5C. These equate approximately to 10th percentiles.





We find that:

1) The number of hot days in recent years has not been high by historical standards, although they seem to be more regular since 1990.

2) There has been, though, a marked absence of really cold days since the 1980s. The only year that stands out in the last decade is 2013, the coldest March on record in England since 1910.


This all implies that average temperatures have not been increasing because all days have become slightly warmer. Instead it is the absence of cold days which is mainly responsible.

Another way to look at it is temperature extremes are narrowing, rather than the opposite which is often claimed.


We can make one further check. The distribution of the hottest March days shows that the warmest was in 1965, and there is no indication that such days are becoming more common – indeed, the opposite seems to be the case.






We find a totally different picture when we look at the coldest days. As before, there has been a marked absence of these for two decades or more. We also find that many much colder days were experienced prior to 1970, even when we compare the really cold month in 2013.





This is all very clear evidence that it is largely the absence of extremely cold days which has pushed up average temperatures in March.

This is only one month, so it will be interesting to see if we find the same pattern repeated in other months.

  1. Broadlands permalink
    April 7, 2016 5:29 pm

    For whatever it’s worth? Some trivia from the US… all data from NOAA.

    In the US 48 states, the coldest March was in an El-Nino year, 1965, 35.44°F. This past March (also an El-Nino year) it was 12°F warmer at 47.46°F. The March trend is clearly up, at 0.25°F per decade. And the annual US winters (DJF) also up 0.22°F per decade.

    BUT, since the 1998 El-Nino year through the 2015 El-Nino? The US winter (DJF) temperatures are down…at the rate of 0.78°F per decade. The full year, annual 1895-2015 trend is also down, minus 0.22°F per decade. These are all part of the US “pause” since 1998.

    All data “cherry-picked”, of course. 🙂

    • April 7, 2016 9:12 pm

      I refer to reanalysis data here:

      Looking at the latitude band 30-60°north temperature is most volatile in the months of January and February. Peak was experienced at the turn of the century and a decline set in from that point: 2016 data not included.

      January and February temperatures are tied to the evolution of the Arctic stratosphere. In fact all latitudes north of 30° south experience the greatest range in temperature in January and February.

      South of 30° south, volatility in temperature is greatest in June and July, tied to Antarctic stratospheric processes. Data here:

      Basically we are looking at the northern and southern annular modes of inter-annual and inter-decadal climate variation involving shifts in atmospheric mass from high to other latitudes that alters albedo and the intensity of the planetary winds. In the mid latitudes of the northern hemisphere it also involves a swing between the incidence of warm moist winds from the south and cold dry winds from the north. So the swing is much greater in the northern than in the southern hemisphere where, in the mid and high latitudes the westerlies that originate in the high pressure cells of low latitudes are always dominant. Massively low surface pressure prevails on the margins of Antarctica, albeit varying on centennial time scales, the central feature of the natural climate change phenomenon.

      Greatest volatility in temperature is experienced closest to the source of that volatility in very high latitudes. It is there that the interaction between stratospheric air rich in ozone and mesospheric air that is ozone poor determines the intensity of polar cyclones that are collectively responsible for the shifts in atmospheric mass. Mesospheric air descends in winter when surface pressure in high latitudes peaks. So, temperature volatility is very much a winter phenomenon.

  2. The Old Bloke permalink
    April 7, 2016 6:30 pm

    In view of what the CET is, and was historically, with the temperature data sets now being used, one needs to ask the questions: 1. Are the same number of reporting stations being used? 2: Are they in the same locations? 3: Are the stations used in the current CET calculation the same as they have always been?

    • David Richardson permalink
      April 7, 2016 10:07 pm

      The Old Bloke – from another old bloke.

      Assuming your questions are not rhetorical –

      The comments and short run-down by Philip Eden here are relevant –

      Note also the top two links at the left relating to CET.

      The provisional series from 1971 on seeks to recreate an extension to Prof Manley’s last published series in 1973 using the same stations as best they can. Prof Manley’s paper is available here –

      Click to access qj74manley.pdf

      I have compared the current Met Office CET with the Philip Eden series from 1971 and they do look much the same – the odd 0.1c here and there.

  3. April 8, 2016 8:22 am

    A question or two regarding these temperatures. In the mid 80’s I understand that the means of collecting ambient temperatures were changed. In fact about the same time that the instruments were changed. Out went Mercury thermometers and human collecting of the readings and in came digital and computers. Obviously the digital readings can be assumed to be very accurate as under any QA system they will need to be regularly checked.
    Q1. Were the two systems run together in the same weather enclosure to ascertain that they gave the same readings? Is this recorded anywhere?
    Q2. Were the two systems run together in separate enclosures to ascertain if different readings were given? Is this recorded anywhere?
    Q3. Were the temperatures normally taken at the same time in both systems?
    I have looked on the web but cannot find the answers, especially as regards possible different latency rates between the two different types of measuring devices.
    Some time ago on WUWT two different technicians from Germany announced that they had in fact collected the temperatures from the two different systems and there had been differences. Does anyone know the answer please?

  4. April 8, 2016 12:14 pm

    I purchased about 15 loggers to check temperature data in different vineyards back in 1994. Each was placed in a small Stephenson screen type apparatus and they could log every 20 minutes for three months at a stretch. Before I set out to use them I ran them on the bench together in the same environment for a few days. In my experience every logger gives a slightly different reading to every other logger. Furthermore, the difference between them can change across the range. They are a bit like thermometers that are sort of roughly calibrated. When I compared my data against that from another manufacturer in the same environment there was a difference, but larger.

    For my purposes these differences were not that material. I was interested in heat exposure above a threshold where flavour damage occurs.

    Historic data taken from thermometers is based on the daily minimum and the daily maximum temperature. The average of the two can be 2° or more either side of a mean of 24 readings taken on the hour. If you are gauging the effects of temperature on the growth of plants both time and temperature are important so you need hourly readings.

    Strictly speaking the mean temperature, half way between the minimum and the maximum tells us very little about the thermal regime of any location. In analysis it would be better not to use the mean but to look at whether the maxima or the minima are changing. Much more useful to know that.

    Climate change is by and large a winter month phenomenon at a time of the year that is undesirably cold. Its the coldest highest latitudes that have seen the strongest increase in temperature and it happens in winter. Do those polar bears actually enjoy the winter months when they starve and hibernate or summer months when they fish and fornicate? Do many humans live up there? This is the nonsense of the single average global temperature statistic. Its elevated by temperature increases in uninhabitable locations when they are least habitable…….in the process improving the environment for both plants and animals.

    That single global statistic is good for propaganda but useless for analysis. It conceals the things that we really want to know. Disaggregation is the way to go.

  5. Broadlands permalink
    April 8, 2016 1:20 pm

    Back in 1920 “a simple equation” was described…with some comments at the end on US variability.

    Click to access mwr-048-07-0394.pdf

  6. April 8, 2016 2:54 pm

    Would the temperature difference between max and min tell us about anything interesting?

    • Broadlands permalink
      April 8, 2016 3:01 pm

      Depends on when they are measured. NOAA repeatedly adjusts for TOB, “time of observation” bias, especially for some older measurements.

  7. Kelvin Vaughan permalink
    April 8, 2016 3:44 pm

    In the CET, If you take the highest maximum for each year since 1990 regardless of when it occurred in that year then the highest maximum is showing a falling trend.

    If you take the highest minimum temperature for each year since 1990 regardless of when it occurred in that year then the highest minimum is showing a virtually flat trend.

    If you take the lowest maximum for each year since 1990 regardless of when it occurred in that year then the lowest maximum is showing a rising trend.

    If you take the lowest minimum for each year since 1990 regardless of when it occurred in that year then the lowest minimum is showing a flat trend.

  8. April 8, 2016 5:42 pm

    Thank you Erl.


  9. A C Osborn permalink
    April 8, 2016 6:20 pm

    UHI anybody?

  10. justanotherpersonii permalink
    April 9, 2016 2:19 am

    The CET is an interesting counterbalance to the global datasets produced by GISS, NOAA, and the Met Office/Hadley Centre.

  11. JonA permalink
    April 11, 2016 7:27 am

    This looks like classic UHI effect – any analysis ever done on the temperature stations
    in the UK?

Comments are closed.

%d bloggers like this: