- 34 climate stations centred on Moscow are divided into three groups: 1) Large cities over 1 million 2) large towns and small cities 20,000 to 1 million and 3) Rural 20,000 and less.
- The warming trend through the large cities is +2˚C since 1880. The warming trend through the rural records is +1.2˚C since 1880. The large towns and small cities lie close to but slightly warmer than the rural records. There is clear evidence for urban warming.
- It is quite clear that the rural records provide the most reliable evidence for warming from this region that spans a 765 km radius around Moscow. Including the large cities in a regression biases results upwards by +0.4˚C.
- Berkeley Earth report warming for Russia of +3.39˚C per century since 1960. The rural records I have selected yield 0.92˚C warming per century. Clearly a very large discrepancy that requires an explanation.
Are things heating up in Moscow?
My hunt for global warming is moving north. I am not yet done with the southern hemisphere and will return south next week with a summary that will include S Island New Zealand and Paraguay in addition to Central Australia, Southern Africa, Patagonia and Antarctica that have all been dealt with individually [1 to 5].
In this post I want to make a first pass at characterising urban warming that is distinctively different to CO2 forced global warming. Urban warming was formerly non-contentious, that was until Berkeley Earth published BEST and more or less declared urban warming dead .
Time series of the Earth’s average land temperature are estimated using the Berekley Earth methodology applied to the full dataset and the rural subset: the difference of these is consistent with no urban heating over the period 1950 to 2010, with a slope of -0.1±0.24/100yr (95% confidence).
This is a poorly formulated statement. For a start we have to guess that they mean 0.1±0.24˚C/100yrs. And we need to note they are comparing a rural sub-set with the whole data set, not an urban sub-set of records. And so what they mean is that urban heating does not impact their whole data set and not that urban heating does not exist. Reading the paper we learn that urban records are adjusted to a regional average, are de-weighted if they depart from that regional average and are further de-weighted by area weighting since urban areas account for only 0.5% of global land area.
Whilst I approve the methodology of comparing a record to a regional average (congruity) to judge the quality of a record, I believe it is absolutely wrong to adjust a record to some pre-conceived idea of what it should show. The average of a selection of good quality records will give the regional average. There is absolutely no need to apply an adjustment. If a record is good, it should be in, if it is bad for whatever reason, it should be out.
Dealing with urban records is nonetheless tricky. Ideally they should all be excluded to minimise any direct impacts of Man on the temperature record. But in many areas, the urban records are often the longest and most continuous and if you want a long record that goes back to 1880, it may be unavoidable to use the urban ones. In southern Africa, some urban records were flat, in line with rural records, whilst some showed clear warming.
Clicking around on the NASA GISS GHCNV2 map, I have found it difficult to find decent sets of congruous records from heavily populated areas. But in doing so Russia caught my eye for a number of reasons. First it comprises 11.35% of the global land area. Second, it has a large number of long continuous records from both urban and rural areas and third, as we shall see, the data appear to be of excellent quality. Thus I got distracted and set about doing an urban-rural comparison and what better place to start than with Moscow, a conurbation with over 8 million people.
I placed Moscow at the centre of a NASA GISS GHCNV2 search that returned the stations shown in Figure 1. This showed 6 cities with over 1 million people spread across W Russia, Ukraine and Belarus. I wanted to compare these with rural stations where I selected 9 towns of 20,000 or less. Flicking through the station charts it was clear the urban records showed warming, less so for the rural records. The results were in a way surprising hence I took time to add a third group comprising the remaining stations giving 34 records in total.
Figure 1 34 stations selected by a NASA GISS search placing Moscow at the centre.
Figure 2 The study area with Moscow, Kharkov, Kiev, Minsk and St Petersburg labelled. The dark green is forest, the paler brown is agricultural land. I’d be interested to know if the predominance of agricultural land to the south reflects deforestation or was this previously natural grassland?
Figure 3 The number of operational stations is extraordinarily good with many long records. All 34 stations operational from 1952 to 1970. And note that during WWII some stations went off line but most remained operational, which I find extraordinary. The main discontinuity comes in 1990 with the fall of the Soviet Union.
Anomalies (dT) are calculated by deducting the station average from the time temperature series for that station. Where I have compared this method with a fixed base period of 1963 to 1992 no material difference is evident and it is time consuming to do this both ways all the time. Throughout I use the GHCN V2 data that are only slightly adjusted from raw records. I have yet to find a user friendly source of raw records (V1) on line.
Figure 4 There is no doubt from this data that temperatures have been rising in these large cities fairly continuously since 1880. A regression through the data shows an increase of +2˚C since 1880 (+1.53˚C per century). The tops and bottoms are rising with similar gradient and the data bears all the hall marks of a rising trend. A feature of the bottoms are four cold spikes that break through the rising trend, most notably in 1941/42 which is a feature we will see in many European records and one that killed many Germans and Russians during WWII.
Figure 5 Nine rural records from the area surrounding Moscow. If you click on the charts they will open in a new browser window. If you do this and compare this chart with Figure 3, you will be hard pressed to see the differences between them. There is a very high degree of congruity. But doing a regression through these 9 rural records we do get a significantly reduced amount of warming of +1.2˚C since 1880 (+0.92˚C per century).
Thus my hunch from visual inspection proved to be correct although the amount of difference between the rural and urban was not as large as I expected. The high degree of congruity between these two groups led me to look at the remaining records from large towns and small cities to see if they lay in between the urban and rural groups.
Figure 6 The intermediate group of large towns and small cities do indeed give a trend intermediate to the rural and large cities groups although only marginally and insignificantly warmer than the rural group with a gradient of +1.3˚C since 1880.
Figure 7 And for good measure the three groups plotted together which really brings home the high level of congruity between the groups and the small differences in gradient that exist between them.
There is no doubt in my mind that of the three groups, the rural group is the one most likely to be representative of warming in this part of Russia although the towns and small cities group is almost the same. Adding in the 6 records for large cities will bias the results. There are two simple ways to assess this bias. The first is to give each station equal weight and to use simple arithmetic:
Sum = 47.5/34 stations = 1.4˚C
So simple arithmetic equal weight would bias the results upwards by +0.2˚C. The other way is to simply plot all the data together and run a regression:
Figure 8 Regressing all the data together gives a slope of +1.6˚C, significantly higher than the rural only result.
The rural only records from this part of Russia and neighbouring Ukraine and Belarus show warming of +1.2˚C since 1880 equivalent to +0.92˚C per century. Records from large towns and small cities yield a result biased 0.1˚C higher and are effectively the same as the rural records. The records from large conurbations over 1 million yield +2.0˚C since 1880 and clearly bias the results upwards if included in a regression.
Berkeley Earth (BEST) report a warming trend for Russia of +3.39˚C per century since 1960, almost 4 times the rate of the rural records reported here . From the data reported here there is no reason to select 1960 as a start date since the trend is linearly up since 1880. Looking else where in Russia, for example E Siberia (work in progress), two warming legs are recognised, one does begin in the 1960s, and there beginning a regression in 1960 may have some merit. But in E Siberia, it was as warm in 1943 as it was in 2007 and a case can be made for no warming at all.
This post is a first and simple pass at Russia and urban heating that I will build upon. I find clear evidence of urban heating. Urban stations that display warming trends that can be attributed to the urban location, should not be used in temperature reconstructions.
Time series of the Earth’s average land temperature are estimated using the Berkeley Earth methodology applied to the full dataset and the rural subset: the difference of these is consistent with no urban heating over the period 1950 to 2010, with a slope of -0.1±0.24/100yr (95% confidence).
The data presented here, though limited, do not support this statement from Berkeley Earth. Nor do the results presented here support the high annualised rate of warming reported for Russia by that group.
 Temperature Adjustments in Australia
 The Hunt For Global Warming: Southern Africa
 The Hunt For Global Warming: Southern Africa Part 2
 The Hunt for Global Warming: South America
 The Hunt For Global Warming: Antarctica
 Wickham et al 2013: Influence of Urban Heating on theGlobal Temperature Average using Rural Sites Identified from MODIS Classifications
 Berkeley Earth