Global Surface Temperature

Estimates of global surface temperatures [GST] from gridded dataset of historical near-surface air temperature anomalies over land and sea. At the moment I have applications to download and display data from the NOAA GISTEMP dataset, and the Met Office HadCRUT4 which is a collaborative product of the Met Office Hadley Centre and the Climatic Research Unit at the University of East Anglia.

The gridded data is a blend of CRUTEM4 land-surface air temperature dataset and the HadSST3 sea-surface temperature [SST] dataset. The dataset is presented as an ensemble of 100 dataset realisations that sample the distribution of uncertainty in the global temperature record given current understanding of non-climatic factors affecting near-surface temperature observations. This ensemble approach allows characterisation of spatially and temporally correlated uncertainty structure in the gridded data, for example arising from uncertainties in methods used to account for changes in SST measurement practices, homogenisation of land station records and the potential impacts of urbanisation. The HadCRUT4 data are neither interpolated nor variance adjusted.

Info courtesy of the Met Office.


CRUTEM4 sites and the placement of Stevenson screens

I am surprised that some of the UK climate records used by the Met Office to calculate their global land temperatures for CRUTEM4 with, are from sites where the instrument enclosure, primarily the Stevenson screen, has been compromised over the years by the encroachment of buildings, car parks, and runways and the various ‘climate’ sites around the country, to such an extent that it must in some way be affecting the temperature sensors. Creeping urbanisation has been happening for years, and is not a new problem, it’s a bit like how politicians suddenly realised that life expectancy has been on the rise for the last 100 years.

Before I go any further these concerns have been voiced before, and a review of the observing sites of the UK has been done before, and much more thoroughly than I can do in this short article, most notably in the Surface Stations Survey by Tim Channon on the TallBloke blog.

The Surface Stations Survey work was done a few years ago now, and as far as I see wasn’t directly linked to the ‘raw’ monthly CRUTEM4 temperature data that you can freely download from the Met Office, and which is used to calculate a monthly estimate of global land temperature with. In recent years the Met Office, for some reason known only to themselves, have reduced the number of the UK sites from well over 100 twenty years ago (fig 1), to just 18 sites in 2017 (fig 2).

Figure 1 – CRUTEM4 Sites 2000

Figure 2 – CRUTEM4 Sites 2017

Here’s a graph (fig 3) of how the total number of UK sites that are currently used in the CRUTEM4 calculations has declined in recent years.

Figure 3 – WMO Block #03 sites

The irony of this 80% or more reduction in UK sites used, is that two of the three sites used to calculate the composite CET series, the longest instrumental record of temperature in the world, are now no longer used – Rothamsted (1872-2012) and Preston Moor Park (aka Stonyhurst 1960-2012).

Poor siting of instrument enclosures

But I digress, what I really wanted to moan about bring to people’s attention was the precarious siting of the Stevenson Screen at some of the 18 sites that we still use to calculate a global temperature with. Generally the siting of the screen didn’t look too bad, but there are a number that are poor, and here are three of the worst sited Stevenson screens that I found using Google Maps. Of course guessing where the screen is an art that has become a bit of an obsession with me. The biggest offenders are all at airports, namely Aberdeen, Valley and the infamous Heathrow (figs 4, 5 & 6).

Figure 4 – Courtesy of Google Maps

Figure 5 – Courtesy of Google Maps

Figure 6 – Courtesy of Google Maps

At this point I would like to say I wouldn’t be able to do this without Google maps, but I have noticed that the generally the quality of the highest zoomed images is inferior to those in the Google map images of the Surface Stations Survey. This might be just a Google maps issue, or it maybe a deliberate restriction on quality and zoom level requested by the MOD for RAF stations. The yellow circle is at a radius of 10 metres and the blue circle at a radius of 30 metres. I won’t go into detail of what I estimate the WMO classification for each site would be as regards temperature, I’ll just leave it your imagination.

What can the Met Office do about it?

When I was an observer every so often at an outstation, someone would come round and inspect the ‘met’ enclosure to see if it was being maintained correctly, I wish now that I had taken a keener interest in what the inspector was looking at other than if the bare patch had been weeded recently! I wonder if there was tick box to confirm that no jet engines were being run up within 30 metres of the screen? I can remember quite clearly being wafted by warm gusts of air from an F3 Lightning at Binbrook en route to the Stevenson screen across the pan to do the 09 UTC observation even in the middle of winter.

They could if they wanted to without much effort do the following with the climate records used from the UK in CRUTEM4:

  • Reinstate the best of the climate stations that have been lost in recent years, but not the records from RAF Waddington or RAF Brize Norton please!
  • Immediately reinstate the temperature climate records for Rothamsted and Stonyhurst, at the same time adding the one from Pershore, so that the three stations used for the renown CET series are included in the calculations, which to my mind would be only fitting!
  • Remove Heathrow until the enclosure has been relocated possibly in the middle of Bushy Park!

This would be very easy for the Met Office to do, they wouldn’t have to go cap in hand to any other meteorological service to ask them to supply the data, as they already have those temperature records.

I know just how sensitive temperature sensors are in AWS these days, I have a Vantage Pro, and over the years I’ve relocated it a number of times in our garden, each location had its different weaknesses, too close to trees or the hedge, or too close to an area of paving, now it’s far too close to the garage. It certainly is a very difficult, if not impossible task to find a location on a modern airfield that’s totally unaffected by external influences on temperature. But in this day and age of advanced wireless communication, I just can’t believe it’s not possible to install AWS as far away as possible from any runway, car park, building or road, at any site, which invariably is at an airport, be it military or civilian. I’ve been doing it with my AWS without a problem for the last 13 years, albeit at a range of less than 10 metres! Inevitably this will have to be done as the demand for green space on airfield sites increases till the whole damn place is paved for a parking lot.

Cool start to August

12 UTC Mean Temperature Anomalies 1-14 August 2017 – courtesy of NCEP reanalysis

Fascinating anomaly chart for the first two weeks of August 2017. A band of cold anomalies more or less stretches from central north America, across the north Atlantic, to northwest Europe, with anomalies ranging from zero to -2°C. The recent heatwave across southeast Europe has left its mark there, with a belt of positive anomalies, as warm as +6°C over the Ukraine and eastern Black Sea, and curving down through Iraq, Saudi Arabia and into Sudan.

May 1-15 northern hemisphere temperatures

Figure 1 – Data courtesy of NCEP reanalysis

The extremely high temperature anomalies seem for the moment at least seem to have disappeared from the Arctic. This temperature anomaly chart (fig 1) is for the first 15 days of May 2017, and yes anomalies are still mainly positive in the Arctic, but in a range -2°C to +6°C, rather than in excess of +16°C that they were back in January and February. That might explain why Arctic sea ice has staged a bit of a recovery during late spring. There are a number of anomalous cold areas, one north of the Great lakes in Canada (-4°C) and another one over Scandinavia and northeast Russia (-6°C), the latter causing problems with heavy late spring snowfalls in places. Central North America, Greenland and large parts of central Asia have seen a very warm spring with anomalies typically +4°C above the 1948-2014 long-term average.

Latest GISS global temperature data

There are two key graphs central to the debate about climate change, one of them is arguably the primary reason why global temperatures are rising, and that’s a graph of monthly CO2 (fig 1) as recorded at Mauna Loa (I hope they’ve made allowances for any CO2 being vented by the volcano itself). The other is a graph of monthly global temperature anomalies (fig 2), either from the GISS  data series produced by NASA, or maybe from the CRUTEM4 series produced by the Met Office and CRU. The latest GISS values are now in, and the anomaly for March is +1.12°C, which although slightly up on the February figure (+1.10°C), is down on the +1.28°C of March 2016. That drop might be expected, because last year in March the last El Niño event was just starting to wind down.

Figure 2

Depending on the range of the graph, and this one’s for the last 30 years (fig 2), you produce a different slope in the linear trend, but I think a period of 30 years is a fair length to sample, and I make the decadal rate from the GISS series is +0.181°C per decade. If you look at a linear trend for the last 10 years of global temperatures that decadal rise increases markedly to +0.383°C. I will admit now that in some screenshots in my earlier blogs I may have screwed this value up and I apologise, not that any of my readers seemed to notice it, or if they did they never commented, hopefully I’ve now got it right. Here’s a table of the recent monthly values of CRUTEM and GISS that I used to create the graph with (fig 3).

Figure 3

The latest data from Mauna Loa is ever upward it seems, and the first graph (fig 1) clearly shows the annual cyclical nature of CO2, and the obvious fact that CO2 has been rising inexorably by around 5% per decade since 1958, that’s a decadal increase of around 15.3 ppm if I’ve got that linear trend right, but I much prefer the following graph (fig 4) which shows the 12 month rate of change in CO2 levels. As you would expect the 12 month rate of change in CO2 has been increasing over the last almost 60 years and oscillates a fair bit – and yes I know the linear trend is probably not a good idea.

Figure 4

March 2017 temperature anomalies

Figure 1 – Courtesy of NOAA/NCEP Reanalysis

You can see why the southeast of England had the highest regional temperatures in the UK during March, it was thanks to that warm anomalous area (+6°C) centred over Eastern Europe, but once again the central Atlantic was a little cooler than average. The central United States had another warm month (+6°C) but most parts of Russia were even warmer, with a massive +12°C anomaly over Novaya Zemlya.

Northern hemisphere temperature anomalies – February 2017

Figure 1


Early March temperature anomalies

Figure 1

Apart from a belt of negative anomalies (-2 or -3°C) stretching from the mid-western Atlantic to Sweden, the first eight days of March have been very warm in most areas across the rest of Europe. There is a large positive anomaly (+9°C) over Ukraine, and another one equally large over the Greenland sea.

Global temperatures drop back

Figure 1

The 12 month moving average of Global temperatures has fallen back in the last few months, after climbing quite sharply for the last four years or so. I’ve included a plot of both the CRUTEM4 and GISS monthly series as you can see (fig 1). I like to use a 12 month moving average because it’s a simple way of removing any seasonality, I’ve added a linear trend for both even though it’s probably not correct or in any way scientific. The linear trends are not aligned because the Americans use the 1961-1990 long-term averages to calculate their anomalies, whilst the British in their wisdom still use the 1951-1980 averages. The rate of increase in global warming is higher in the GISS series during the last 50 years at +0.292°C per decade, as opposed to the slightly lower +0.273°C using the CRUTEM4 series. For a bit of fun (yes I am weird), I thought that I would overlay the El Niño events of recent years (reddish vertical bands). I did read somewhere, that there was a strong correlation between global temperatures and ENSO, so I thought that I’d see if there was. Well that does seem to be true in some events, but not all. The 1991-1992 El Niño for instance, although the cooling that did occur then may have been due to the dust released in the Mount Pinatubo eruption at around that time. Talking of volcanic events the next chart (fig 2) is for the last 30 years and shows all volcanic events during that time, although I’ve only include the ones that scored 4 or higher on the VEI scale. You’ll also notice that the linear trend for both series is much lower during the last 30 years than it was in the last 50 years in the first chart. I have written about this before with regard to linear trends and the CET series, it can be misleading and I’m not going to bother to go into it all over again.

Cold start to January across Russia and Eastern Europe

Figure 1 – Data courtesy of NCEP

Don’t look for much in the way of cold air from the north or Scandinavia this Winter, a far better bet is finding a thin conduit of air that stretches westward from the deep cold in the heart of central and northern Russia. Of course the cold anomaly of -10°C is in stark contrast to the +20°C warm anomaly in the Barents Sea between Spitsbergen and Franz Josef Land. It’s quite apparent to me that ‘bullseye’ anomalies like this one in the Arctic, are due to the fact that air temperatures over open sea are much higher than temperatures over pack ice, which of course is absent.

Image 2 – Courtesy of NSIDC

Latest global temperatures and volcanoes…

Mount Pinatubo June 1991 Courtesy of Wikipedia

I was just going to post the latest GISS global temperature anomalies for October 2016, and thought it might be a great idea to overlay the graphs with the volcanic activity of the last 136 years that would have impacted on global temperatures. Finding the source of the raw data on the internet was not easy, but there were some interesting Wikipedia articles about a term called the Volcanic Explosivity Index [VEI]. The VEI is a relative measure of the explosiveness of volcanic eruptions devised by Chris Newhall of the United States Geological Survey [USGS] and Stephen Self at the University of Hawaii in 1982. The scale runs from zero to eight, and there have only been 42 VEI 8 mega-colossal explosive eruption events in the last 36 million years. To give you an idea of the VEI scale Mount St Helen’s in 1980 was a type 5 event, Mount Pinatubo was a type 6 event in 1991, and Mount Tambora was a type 7 event in 1815. Anyway I finally tracked down the data to NOAA, and what’s now called the National Centers for Environmental Information [NCEI], they maintain a  database they call ‘The Significant Volcanic Eruption Database‘ and I’m very grateful to them because I can now use the data in my global temperature application, and will add it to the graphs in my daily and monthly CET applications as time permits.

The caldera of Mount Tambora – courtesy of Jialiang Gao (

Here’s the full monthly GISS anomalies and a 12 month running average since 1880. The first event volcanic in 1883 was the eruption of Krakatoa, which may have resulted in some cold winters in the 1880’s, and undoubtedly some very colourful sunrises and sunsets.

And below is a zoom into the GISS global temperatures since 1980. As you probably know 2016 will undoubtedly be the warmest year in the series, although the 12 month running average has already peaked. The three volcanic events marked on the graph from left to right are Mount St Helens (1980), El Chichón (1982) and Mount Pinatubo (1991). I’ve placed a six month span on the volcanic event but this of course may have been much longer for Pinatubo. Looking at the graph there is no doubt about the obvious effect that Pinatubo had on global temperature, but it’s not obvious with the other two. I’ve tried drawing a left, right and a centre aligned 12 month moving average, and none of them synchronize well with any fall in global temperature, in fact the 1982 event almost appears to have caused a spike in global anomalies. Interestingly, although Wikipedia claims the Puyehue-Cordón Caulle event was a VEI 5 event in June 2011, the NOAA database lists it as a VEI 4. I suppose in reality even events with a VEI of less than 5 still emit a lot of dust, ash and gas into the atmosphere so they all must have some effect, looking at the graph there was a decrease in anomalies in 2011 which possibly had something to do with the 2011 eruption. I’ll have another think about the eruption data and maybe come up with some kind of annual index to overlay rather than specific events. It’s now been almost twenty-five years since we’ve had a type 5 VEI event on the planet, I wonder if 2017 will continue in the same vein?