(1) Some Land Surface Temperature Records (BEST project, CET 1659-2011, Frost Fairs)
(2) Comparing Tempeature Records (HadCRUT3 & 4)
(3) Some Rainfall Anomalies 1900-2011
(Global Land, Sahel)
(4) Graphing Ice Volume & Snow Cover
(5) IPCC Assessment Report Palaeo-Climate Graphs 1990-2007
(plus Groveman & Landsberg)

(1) Some Land Surface Temperature Records

Berkeley Earth Surface Temperature project (BEST) first developed an average land surface temperature data set based on thermometer records reaching back to 1800 and has since (July 2012) developed the data set back to 1743 with a continuous record from 1752 (although the uncertainty increases greatly with these early years
). Other data sets only venture back to 1880 (NASA GISS & US NCDC) or 1850 (Hadley-CRU) as the worldwide availability of thermometer temperature records become much more difficult even back into the 19th century.
As well as the BEST land surface temperature record, the differences between land temperature anomalies & global (i.e. land & ocean) temperature anomalies is also demonstrated in the first graph using NCDC & HadCRUT3 data.
How well the BEST land surface temperature record with its large 18th Century fluctuations reflects the actual global temperatures is yet to be discussed fully. Rohde et al (submitted) suggest a volcanic cause for most of these fluctuations but caution that they have "rather uncertain ampitude" as this early part of the BEST data set is derived mostly from only European data with some N American data after 1785, this data being all that is available for such early years. 
Data & further info at BEST website.   NCDC data.   HadCRUT3.

Central England Temperature record (CET) is the longest existing thermometer record providing monthly average temperatures from 1659, and from 1772 daily average temperatures. For more description see Wikipedia.
Accurate phenological records (timing of flowers & 'budburst' of leaves) which date back to 1736 in Stratton Strawless, Norfolk and from 1760 nationally are described as showing a shortening of UK winters in the last few decades that is unprecidented over 250 years & reflected in the C.E.T. Springtime temperatures over that period.
Data source - HadCET, Met Office Hadley Centre.

FROST FAIRS The CET also demonstrates that it is wrong to consider the iconic Frost Fairs on the River Thames as evidence of a "Little Ice Age."
The historical records suggests there may have been as few as seven Frost Fairs with historical record of the Thames freezing over in London about two-dozen times.
Walford lists 24 years when records talk of the Thames or more generally rivers freezing over between AD134 and 1814 but lists only 4 Frost Fairs. MacKay lists 19 freezings since AD250 with 7 evident as Frost Fairs (one Fair implausibly described happening in AD695). Jones talks of 23 freezes since 1250 and Lamb lists 25 freezes (2 partial ones) since 1400, concluding his 1977 account:-
"The river has never been frozen over in London since the replacement of
London Bridge
, the building of the embankments and the channelling of the
tributary streams in pipes in the early ninteenth century, but considerable
amounts of broken ice were reported in 1839, 1895 and 1929.

The CET coldest-month temperatures graphed below show that recent winters have been as cold as some years that saw Frost Fairs but the London Thames never froze during these cold winters. In 1963 for instance the river froze only far up-stream above the tidal limit.
Of course the Thames is not the only major European river. The River Rhine last frozen over in 1963 and records show such events were more frequent in the early twentieth century than in the nineteenth century. (The 14 frozen Rhine years from 1780-1963 only match three of the years labelled in the graph below eg 1830, 1947 & 1963.)

(2) Comparing Temperature Records

HadCRUT3 is one of three Global Surface Temperature records (with NASA's GISSTEMP & NOAA NCDC) and is soon to be replaced by the updated HadCRUT4. HadCRUT3 has less coverage of land temperatures than the other surface two records ( with reducing coverage in recent years - see ) particularly in high Northern latitudes. As both land temperatures and high northern latitude temperatures are rising faster than the global average, this results in HadCRUT3 having a low bias.

The increased land temperature coverage of HadCRUT4 ( see maps on pages 5 & 6 here ) adjusts current global average temperature by +0.04 deg C (compared with HadCRUT3) and greatly reduces but does not eliminate the low bias.
The largest differences evident between HadCRUT3 & HadCRUT4 occur in the mid-1900s and are due to a re-analaysis of the ocean temperature data where sea water samples collected in canvas buckets over the ship's side would read low due to evaporative cooling.

(3) Some Rainfall Anomalies 1900-2011

The trend in global land rainfall (precipitation) anomalies through the 20th century showed only a very slight rise compared with the large interannual variation
(see New et al 2001) and the discernible regional trends.  Recent years with large positive anomalies may yet be part of this interannual variation. Global average precipitation is 990mm and the global land average is 715mm.
Data scaled from image here.

The graph of Sahel Rainfall Anomalies and map below were created as part of an item published here debunking a skeptical thesis about "The Greening of the Sahel."

(4) Graphing Ice Volume & Snow Cover

was begun because useful graphs of certain climate data were often not readily available. Sometimes this was less a matter bad graphs and more a matter of awkward data that requires a bit of thoughtfulness during the graphing process.


PIOMAS anomaly graph of Arctic Sea Ice Volume, for instance, has featured dramatic annual negative anomaly 'spikes' since 2010. Yet graphs of the superb PIOMAS data do not (at time of writing) show enough detail for the profile of these 'spikes' to be understood.

When as here they are graphed in detail, year-on-year, these 'spikes,' the mid-summer anomaly minimum can be seen to be followed by an increased bunching of the annual anomalies during the latter half of melt season. The rate of bunching seemingly appears to cease with the end of the melt. This bunching perhaps suggests there is some mechanism (perhaps the location of ice remaining after late June/July) that makes further melt increasingly difficult and so allows years with more ice remaining (which thus includes ice that is more easily melted) to close the gap by the end of the melt-season.

Another and more extreme example of awkward data concerns snow cover. Rutgers University Snow Lab maintains an exellent record of weekly Northern Hemisphere Land Snow Cover that, with three short breaks, stretches back to 1966. Yet illustrating the decline in Snow Cover graphically is particularly awkward and this situation may be why the decline remains poorly reported.
In June 2012, for instance, it was reported that Northern Hemisphere Land Snow Cover had broken a record. Golly! The June snow cover anomaly was the lowest figure for June in the whole 45 year record, besting the previous record set in 2010 by 1 million square kilometres. This statement complete with a graph of June anomalies, was reported as the last item in the US NSIDC's first July Arctic Sea  Ice News & Analysis bulletin. It understandably never gained widespread coverage. Indeed, it had perhaps too much of the flavour of fresh-picked cherry.

Yet there is something dramatic lurking within the Snow Cover data.

Graphing out the entire 45 year Snow Cover record shows a large annual cycle with no really discernible trends. And adding monthly points onto the graph mainly shows a lot of variability. It takes hard study to spot that anything is happening to Snow Cover with this data and this type of graph.

A different graphical approach is to plot a year-on-year graph. In this case, the result is a colourful cat's cradle although there is a 'rainbow' trend beginning to appear with the more recent years definitely having less Snow Cover during weeks 15-35. The width of this 'rainbow' trend, the advance in melt over this part of the year is 4 to 5 weeks.

If the annual minimums on this graph are examined in ridiculous detail, it can even be shown that 2012 Snow Cover has broken the 24 year record for Minimum Snow Cover set in 1988. It remains, however, somewhat underwhelming.
(There is also an issue here with the Greenland data which during the pre-satellite years has some very low values. These graphs avoid this issue by not unreasonably excluding Greenland from the analysis.)

As a means of seeing past the large annual cycle, a graph of anomalies can be plotted. (This is of the same form as the daily NCEP/NCAR Snow Cover graphs except NCEP/NCAR count both Sea Ice and Land Snow as Snow Cover. And also NCEP/NCAR only plot the last 10 years with a 1995-2009 anomaly base.)
The graph here is of monthly anomalies and yet another record is shown. June 2012 was not just the lowest June on record, it was the lowest month on the entire 45 year record. The June 2012 anomaly stood at 5.74 million square kilometres, pipping the December 1980 record by a whopping 130,000 square kilometres, although this again is hardily something that is going to make front page news.
Beyond this lacklustre June 2012 record, trends are still not greatly convincing in this graph. Increasingly large negative anomalies since 1996 are preceded by very large negative anomalies in 1980-81 and almost continual negative anomalies between 1988 and 1995. Also, positive anomalies have been getting stronger since 1988, although not to the levels seen prior to 1988.

It is only when the anomalies are plotted year-on-year, that the last few years can be seen to droop alarmingly from the 'rainbow' section of the graph. The early and later parts of the year remain a colourful cat's cradle.

Simplifying the graph by plotting only the averages for past decades (rather than individual years) at last allow the drama of the declining snow cover to be seen clearly.

At the height of this summer's melt, 2012 was 8.5 million sq kilometres ahead of the 1972-1979 average. In terms of time, this is a full month advance in the melt over this mid-1970s-2012 period (ie a 27 year period.) That is twice the advance seen over recent decades in the the UK's 250 year-long phenological record which is itself dramatic enough. Another way of visualising this early melt is that the average snow limit at the height of the melt season is now a whole 500 miles further North than it was 27 years ago.

But it is made far more dramatic because this process is visibly accelerating. The average retreat North over the 27 years,  since the mid-1970s, is 18 miles per year. Since the mid-2000s it is 38 miles. In this last dramatic year, at the height of the melt season the snow has been melting away 71 miles further North than in any other previous year.

If the purpose of this analysis is climatology, this next step at illustrating the decline is perhaps a step too far, (although not so if the snow itself is the point of study). By plotting % change in anomaly, the relative variation in snow cover through the year is shown with the dominating feature now being the changes in summer snow levels.

In the very early part of the calendar year, the lowest snow levels occurred during the 1990s and on average have been increasing since. For final part of the calendar year, average snow levels have been increasing over the entire 1972-2012 period.
It is these increases in winter snow levels that have masked the summer decline. And the summer decline is itself less influential within simple annual figures because for two months of the year the snow is effectively down to zero. Thus even a measure of full summer snow levels show a less-than-dramatic decline.

Yet, however well hidden the decline in summer Northern Hemisphere Land Snow Cover, it is just as dramatic and for the climate just as consequential as the Arctic Sea Ice decline.

The 2007 IPCC AR4 project an average annual loss of Northern Hemisphere Land Snow Cover by 2070-2100 at over six times the equivilent figure in recent years (13% by the end of the century, 2% today). Yet the over summer months the 2070-2100 average monthly projection (as per the
ACIA 2004 used by the IPCC) is already being exceeded.
Indeed, at -2.9 million sq km, the NH snow anomaly for June-August 2012 approaches the largest of all summer loss projections found by ACIA 2004. This represents a very large increase in Arctic warming through albedo reduction arriving over half a century before it was expected. We ignore the dramatic loss in the summer snow cover at our peril.

(5) IPCC Assessment Report Palaeo-Climate Graphs 1990-2007
The UN IPCC Assessment Reports presentation of palaeo-climate have not been without controversy. Specifically their graphical representations of global temperatures over the previous millenium has generated much popular controversy, particularly the so-called "Hockey Stick" curve.
Here the original 1990 graph is plotted onto the graphs from 1995, 2001 and 2007 to allow a full comparison that is usually not presented within coverage of the subject. A final graph plots the temperature data that was used to create the more modern part of the 1990 graph (CET) onto that 1990 graph along with the "Hockey Stick," with tthe CET data plotted up to 2010.
Fuller explanation & links to references presented within a 'debunking' article here. 

It is pointed out by Frank et al 2010 that the original reconstruction of Northern Hemisphere average annual temperatures using proxy data was made by Groveman & Landsberg 1979 who produced a temperature series covering 1579-1975. The resulting series has been reproduced on the second graph below by way of a comparison. (Data scaled from Figure 7 Laut 2003.)
Groveman & Landsberg published a decade prior to the 1990 IPCC assessment and over a decade after the work of Lamb on which the 1990 IPCC assesment based its schematic figure 7.1c.

Original IPCC 1990 graph (figure 7.1c).