It is not uncommon to find obvious non-climatic trends in  temperature records that are not related to the growth of the Urban Heat Island (UHI).  These features can be termed inhomogeneities and unless complete station diaries have been kept it is often impossible to accurately assign causes.

[1]   The administration and hardware needed to collect temperature records were put in place for purposes of weather reporting and forecasting.    NOT to define subtle climate changes.  So this whole process of measuring "climate change" from instrumental records is to some extent trying to make a silk purse out of a sows ear.   Computing climate change from  thousands of temperature records from around the globe is akin to attempting to make a set of silk travel baggage from a few sacks full of sows ears.

[2]    It so happens that the best administered,  most complete and easiest to research records are those from larger towns and cities where permanent staff may be better trained and motivated to make more accurate and complete observations and keep records in good shape.  Naturally, any records in growing urban areas will be contaminated to some extent by artificial warming due to the UHI although inhomogeneities such as an uncorrected move to the urban fringe can cancel out the UHI component, see the example of Sydney.
Figure 1   shows  the gradation in  UHI  trends affecting temperature records from a range of population centres in California.  Graphic thanks to Dr James Goodridge, retired State Climatologist  and Dr. Gary S. Sharp,  Center for Climate / Ocean Resources Study, Monterey, CA.
These divergent trends from California that prove the UHI in large data sets is similar to the result from South Africa obtained by Hughes & Balling 1996, who examined records from 1960-1990.
Figure 1

[3]    Conversely,  records from rural sites are more prone to have data gaps,   site / equipment changes are more likely to be poorly documented and generally the data are not as easy to work with as those from cities,  Yet, it is in these imperfect records that a truer definition of the last centuries climate fluctuations will be discovered.   It is not uncommon to find obvious non-climatic trends in  temperature records that are not related to the growth of the Urban Heat Island (UHI).  These features can be termed inhomogeneities and unless complete station diaries have been kept it is often impossible to accurately assign causes.    In many cases the causes can be due to changes in  the immediate environment of the Stevenson screen ( or other enclosure) that houses the thermometer or sensor.  This is quite separate from the steady rise in  temperatures due to growth in the UHI over entire urban areas.

Additions to buildings might increase the radiation of absorbed solar radiation to the thermometer enclosure or for example a hedge could be allowed to grow up, these cases could result in non-climatic warming until site conditions were restored.  Conversely, the clearing of shelter vegetation leading to increased ventilation  or damage to the screen louvres could result in non-climatic cooling over a period of years.

Two examples are given below, Figure 2  from Orange Post Office in NSW Australia and Figure 2  from Greenville in North Carolina, USA.

In the case of Orange, a town with a population in the 20-30,000 range now, the temperature graph shows a  large steady warming from 1940 to the 1960's when compared to the neighbouring  country towns of Bathurst and Forbes.   This warming looks too  rapid to be only due to UHI growth.  Relevant factors are, the Bathurst record was measured peripheral to the city which is similar in size to Orange.  Forbes is about one quarter the size of Bathurst or Orange and the station is urban.   It is possible that much of this rapid  warming in Orange relates to progessive building changes in the vicinity of the Post Office yard over a period of twenty or so  years from 1940 to the 1960's.  In the late 1960's the station moved to Orange Airport and it looks from the record that  improvements were made to the Post Office site  from the mid 1960's.  The writer is not aware that Orange Post Office has been used in large area datasets.

Figure  2

In the case of Greenville, South Carolina,  there is a pronounced non-climatic cooling in the few years around 1904.   Note that Jones 1991 (CRU) did not use that cool period in Greenville.  Both sests of  Greenville data have been subject to quality control  or statistical tests and this graph must call into question the adequacy of the GHCN  procedures.  Note that non climatic cooling in the early years of  these time series inserts artificial warming into the trend, just as artificial warming in recent years.

Figure  3

 

[4]    The UHI phenomenon has been  well documented  in many scientific papers.  The most successful definitions come from studies of one or a few heat islands using perhaps real time temperature recording on a particular night or maybe comparisons over years with nearby rural stations with good quality data.   Definition is clearer too in areas of flat terrain, as opposed to comparisons in hilly terrain where all manner of microclimatic variations could  mask the UHI.   In recent years we have the advantage of seeing the NASA / NOAA images of UHI's in various sectors  of the USA from  GOES satellite  land surface infra red  data.    Remember these images are snapshots on a particular date,  at a time of calm weather.  Obviously during  windy conditions the UHI plume will be dispersed.   There is a lot of explanatory information on the Net on the UHI, just use the search engines, type in heat island or urban heat island and many sites should be listed.   In my view the extensive evidence on the existence of the UHI in urban temperature records  should lead naturally  to an assumption of UHI presence, with only the magnitude to be defined. The failure of climatology in general to take this position has lead to the current situation where the concept of century long global warming,  a cornerstone of global economic policy on carbon emissions,   is based on very poor climate data often contaminated by the UHI.
Consider for a moment if  positions were somehow reversed and the greenhouse industry & IPCC wanted to emphasize the UHI; with all the evidence available they would simply say that the "precautionary principle" dictated that they  accept the existence of the UHI in all urban areas, thus urban records would not be used in studies of long term climate.

[5]    The UHI can be elusive when studies are made on large datasets and there are many reasons for this, some of which will be explored below.

• The growth in the HIE with population is not a straight line relationship but unit increases in population cause smaller increments in the HIE as population rises.
• Many researchers have assumed very small or negligible UHI in populations of less than 100,00, yet this is not so.  The above bullet point on a non-straight line relationship to population plus the underestimation of the UHI in small to medium towns has lead to many studies where city UHI's are compared to so called rural stations that in fact are towns that have slightly smaller UHI's.   This leads to a result proving the original UHI to be small because  the whole concept of a meaningful comparison is flawed from the start.  See my pages on Atlanta / Macon and Sydney Australia for  examples of this.
• The many data breaks, jumps and steps caused by equipment and site changes, or non-climatic data drifts due to equipment and site deterioration,  all constitute inhomogeneities that contribute to a blurring of UHI definition in large datasets.
• A serious and very common factor in recent decades is the general  move from city centers to airports in  post WWII years.  This has usually lead to an immediate step down in the graphed record, see Atlanta in early 1960's  for this type of step.  So, for a time if the record remains uncorrected for the move, some UHI has been compensated for by the step but this is only temporary because airports can be shown to have vigorously growing UHI's of their own, due to the common industrial expansion in those precincts in the second half of this last century.  See also link to Sydney Australia above where the Airport UHI  has grown faster than the city.
• Climatic variations between regions,   between areas of high and low topography, between coast and inland can also cause a blurring of UHI definition in large datasets.  A rural area with many rural records may have been affected by years of subtle micro-climate change due to land clearing and reduction of forest areas.

Errant Warming Trends in Rural USSR Stations

While working on the recent USSR pages several examples of non-climatic trends from rural stations have come to light.

Kzyl Zar west of the Lake Balhash grid boxes is probably the most severe. Figure 3 shows the colossal errant warming in Kzyl Zar  of 3 to 4 degrees in 40 years,  when differenced to two nearby cities.  It is this sort of errant warming that is making the GHCN rural trend equal to the trend when urban stations are added, a possibility totally missed by Peterson et al 1999.

Just look at the magnitude of the abberations in Figs 4, 5, 6 and compare them to the rate of claimed "global warming".

Figure 4

Kokpetky is a rural station east of the Balhash grid box and when diffenced to two nearby cities, non-climatic warming  of 0.5 to 1 degree is seen in  Kokpetky over the 90 years, see Figure 5.

Figure 5

Ekimcan from the SW Sea of Ohotsk grid box is another rural station with a severe non-climatic warming trend of about 1.5 degrees over 70 years,  see Figure 6.

Figure 6

All these USSR examples are from the GHCN,  other datsets are probably similar.

You would have to say that the GHCN statistical tyre kicking process is pretty ineffective to let these grossly aberrant USSR trends through.

Any comments by the GHCN team or anyone with a relevant contribution can be posted.

Links to other pages dealing with data quality

USA and USSR inhomogenous data, usually anomalously warming,   from the V2 GHCN

Aberrant data from Peru

Examples of aberrant monthly data entries and worse found throughout the Jones data

Report comparing record quality in national datasets above 50 degrees north. Downloadable zipped Word 97 file
New report by Warwick Hughes and Alan McCallum
Data quality as measured by missing months has been computed for stations above 50 degrees north in three main global data sets, Jones 1991, Jones 1994 and V2 GHCN(adjusted files).
In recent decades station numbers are reduced in most countries,  data is deteriorating in terms of having more missing months and in the area of the old USSR where problems seem worse, remaining stations are more urban.
Just where the IPCC claim the most warming, in high latitudes, data is getting worse, stations are closing and not just in Russia.

Anyone needing  VBA code writing and work with MS Access queries should contact Alan McCallum through me.

Download 3.6 MB zip file for MS DOS of Jones 1994 station by station temperature data

Download 136 KB zip fil4e for MS DOS of Jones station by station data 1996 update.
 

You read this first here.

Page updated  11, October, 2000

© 2000   Warwick Hughes,

Back to Front  Page