Range of environmental temperature conditions in the United Kingdom For: Department of Transport Date: 17 May 2011 Authors: Matthew Perry and Nicola Golding
temperature_range_report_merged_v2.doc © Crown copyright 2008
-1–
Document History Version
Purpose
Date
1.1
Final results
28 April 2011
1.2
Final results with minor modifications
17 May 2011
Prepared by: Matthew Perry (Scientific Consultant) and Nicola Golding (Adaptation Scientist) Reviewed by: Simon Brown (Climate Extremes Research Manager), Catrina Johnson (Manager, Scientific Consultancy) and Hazel Thornton (Manager, Climate Adaptation Team) Authorised for issue by: Lindsey Smith (Account Manager)
DISCLAIMER: • • •
• • •
•
•
This document is published by the Met Office on behalf of the Secretary of State for Defence, HM Government, UK. Its content is covered by © Crown Copyright 2011. This document is published specifically for the readership and use of the Department for Transport and may not be used or relied upon by any third party, without the Met Office’s express written permission. The Met Office aims to ensure that the content of this document is accurate and consistent with its best current scientific understanding. However, the science which underlies meteorological forecasts and climate projections is constantly evolving. Therefore, any element of the content of this document which involves a forecast or a prediction should be regarded as our best possible guidance, but should not be relied upon as if it were a statement of fact. To the fullest extent permitted by applicable law, the Met Office excludes all warranties or representations (express or implied) in respect of the content of this document. Use of the content of this document is entirely at the reader’s own risk. The Met Office makes no warranty, representation or guarantee that the content of this document is error free or fit for your intended use. Before taking action based on the content of this document, the reader should evaluate it thoroughly in the context of his/her specific requirements and intended applications. To the fullest extent permitted by applicable law, the Met Office, its employees, contractors or subcontractors, hereby disclaim any and all liability for loss, injury or damage (direct, indirect, consequential, incidental or special) arising out of or in connection with the use of the content of this document including without limitation any and all liability: o relating to the accuracy, completeness, reliability, availability, suitability, quality, ownership, non-infringement, operation, merchantability and fitness for purpose of the content of this document; o relating to its work procuring, compiling, interpreting, editing, reporting and publishing the content of this document; and o Resulting from reliance upon, operation of, use of or actions or decisions made on the basis of, any facts, opinions, ideas, instructions, methods, or procedures set out in this document. This does not affect the Met Office’s liability for death or personal injury arising from the Met Office’s negligence, nor the Met Office’s liability for fraud or fraudulent misrepresentation, nor any other liability which cannot be excluded or limited under applicable law. If any of these provisions or part provisions are, for any reason, held to be unenforceable, illegal or invalid, that unenforceability, illegality or invalidity will not affect any other provisions or part provisions which will continue in full force and effect.
Contents Introduction..................................................................................................................... 3
Part 1: Range of Conditions Experienced..................................................................... 3
1.1. Identifying the area of interest ............................................................................... 3 1.2. Analysis methods ................................................................................................... 5 1.2.1 Data extraction and quality control....................................................................... 5 1.2.1.1 Temperature ................................................................................................. 5 1.2.1.2 Solar radiation............................................................................................... 6 1.2.2 Extreme Value Analysis....................................................................................... 7 1.3. Maximum temperature results ............................................................................... 8 1.4. Minimum temperature results ...............................................................................11 1.5. Mean temperature results .....................................................................................14 1.5.1 24 hour mean temperature .................................................................................14 1.5.2 Eight hour mean temperature .............................................................................19 1.6. Solar radiation results ...........................................................................................24 1.6.1 24 hour average insolation .................................................................................24 1.6.2 Eight hour average insolation .............................................................................26 1.6.3 Insolation on a vertical surface ...........................................................................26 1.7 Discussion of results..............................................................................................28
Part 2: Climate Change Information .............................................................................32
2.1 Introduction.............................................................................................................32 2.1.1 UKCP09 Projections...........................................................................................32 2.2 Climate change temperature information..............................................................37 2.2.1 Discussion of variables .......................................................................................37 2.2.2 Results..................................................................................................................38
1 © Crown copyright 2011
2.2.3 Discussion of results...........................................................................................45 2.2.3.1 Annual mean of daily mean temperatures .......................................................45 2.2.3.2 Summer mean of daily mean temperatures .....................................................45 2.2.3.3 Winter mean of daily mean temperatures ........................................................46 2.3 Climate change solar insolation information ........................................................47 2.3.1 Discussion of variables .......................................................................................47 2.3.2 Results..................................................................................................................48 2.3.3 Discussion of results...........................................................................................49 2.4 Wider context ..........................................................................................................50 2.4.1 Limitations of this methodology ..........................................................................52 Conclusions ...................................................................................................................53 References .....................................................................................................................54 Appendices ....................................................................................................................58 Appendix 1: Consideration of the uncertainty due to emissions scenarios....................58 Appendix 2: Maximum temperature for other seasons (not JJA):..................................59 Appendix 3: Minimum temperature for other seasons (not DJF): ..................................60 Appendix 4: Change in mean daily maximum total downward surface shortwave flux for other seasons (not JJA) ...............................................................................................62
2 © Crown copyright 2011
Introduction The transport of radioactive materials is tightly regulated and controlled in order to maintain very high levels of safety. Packages are required to be designed to account for the effects of temperature and insolation in accordance with the European Agreements for the transportation of dangerous goods via road rail and inland waterways (ADR, RID, AND), which in turn apply the International Atomic Energy Agency (IAEA) Regulations for the Safe Transport of Radioactive Material TS-R-1. However, for packages travelling solely within the UK, different temperatures and insolation values can be assumed, which may permit a reduced cost in design (DfT Technical Specification, 2010).
In recognition of this, and in response to the impact of climate change, the Department for Transport (DfT) has identified a requirement to study the range of temperature and insolation values likely to be experienced in the UK. The first part of this report provides information on the current range of temperature and solar irradiance values experienced around the UK, based on historical observed data. The second part provides an initial assessment of model projections of these two variables over the next few decades, and considers the changes possible in response to climate change.
Part 1: Range of Conditions Experienced 1.1. Identifying the area of interest The area of interest specified is all major public road and rail routes in the UK. Movements of radioactive material may cover substantial distances across the UK, so the whole of the UK is considered as one geographical area.
Geographical Information
System capabilities were used to identify climate stations representative of these routes. A buffer zone was created of 1 km either side of all main (‘A’) roads, motorways and railways. These areas were merged together, and all climate stations within this zone were identified. Figure 1 shows the merged transport route buffer zone together with all selected climate stations within this zone.
3 © Crown copyright 2011
Figure 1: Buffer zone (1 km) of UK transport routes together with selected climate stations within this zone.
4 © Crown copyright 2011
Any climate station which recorded air temperature at any time between 1961 and 2010 was included in the search, a total of 1506 stations of which 912 (61%) were within the buffer zone. Three of the selected stations were excluded: Snowdon Summit, which was only selected due to the railway going up Snowdon; Cairnwell which, at an altitude of 928m above sea level, is over 250m above the height of the A93 road, making it unrepresentative of the route; and Glen Ogle which at 564m above sea level is over 250m above the height of the A85 road, again making it unrepresentative of the route. Stations in the Channel Islands were also excluded. The remaining selected stations provide an excellent coverage of main UK transport routes. The highest station was Holme Moss at 520m above sea level, which is adjacent to the A6024 road in Yorkshire.
Most of the selected stations have not operated for the whole of the 50 year period from 1961 to 2010. Of the 909 stations, between 200 and 400 stations had daily temperature data available on any one day, with an average of 344 stations. Climate stations all record the 24 hour maximum and minimum air temperature daily at 0900. Some of these stations additionally record hourly air temperature, and a smaller subset record hourly solar radiation. For solar radiation, 100 of 144 stations were selected as being within the transport routes zone.
1.2. Analysis methods 1.2.1 Data extraction and quality control 1.2.1.1 Temperature 24 hour maximum and minimum temperatures were extracted from the Met Office database for the period 1961 to 2010, for all selected stations. This 50 year period was chosen to provide a long enough period to produce robust statistics, starting in 1961 because there is much less digitised data available prior to this date. For each day, the highest maximum temperature and the lowest minimum temperature recorded at any station was found. In addition, the average maximum and minimum temperatures across all stations on each day were calculated. Daily mean temperatures were calculated for each station every day by taking an average of the maximum and minimum temperatures for that day. Although this method is an approximation, it has the advantage of having many more stations available compared to using hourly data. Studies have shown that the root mean square error of this method compared to calculating an average from 24 hourly values is between 0.6°C to 1°C (Dall’Amico a nd Hornsteiner, 2006; Weiss and
5 © Crown copyright 2011
Hays, 2005). The highest and lowest mean temperature recorded at any station on each day was found, and the average mean temperature across all stations also calculated.
Although the data in the database have already been quality controlled, further quality checks were applied by checking the differences between the highest (or lowest) and second highest (or second lowest) station temperatures on each day.
For maximum
temperature, if the warmest station was at least 4°C warmer than the second warmest station on any day, then the warmest station was marked as suspect and excluded. Given the density of the station network, it is unlikely that local variations of maximum temperature of more than 4°C will occur between sta tions, so such differences are likely to be caused by incorrectly reported values or poorly exposed or maintained recording equipment. For minimum temperature, localised variations can be greater due to the formation of cold air pools in hollows. For this reason, a slightly higher threshold of 5°C was allowed between the coldest and second coldest station on each day. For daily mean temperature, differences of 3°C and 4°C were allowe d respectively between the warmest and second warmest stations and the coldest and second coldest stations on each day.
For eight hour temperatures, the average temperatures of the hottest and coldest eight hour periods of the day were calculated for each station on each day. The hottest and coldest periods were chosen by calculating the long term average over all selected stations by hour and month. As few stations had hourly data available prior to 1971, the data period analysed was 1971 to 2010. From 1971 at least 22 stations were available each day, increasing to 50 by 1988 and 100 by 2009. Similar quality control was carried out as for the 24 hour mean temperatures.
1.2.1.2 Solar radiation Solar irradiance or insolation is a measure of the power or rate of energy per unit area. Solar radiation is the total energy received per unit area over a period of time. The Met Office database contains hourly values of total global solar radiation received on a horizontal surface in KJ / m2. This is recorded using pyranometers at meteorological stations which are inspected regularly to ensure that they meet the required standard. The term ‘global’ means the total solar energy reaching the surface, i.e. the sum of direct and diffuse radiation.
Diffuse radiation is radiation that has been scattered by the
atmosphere or reflected from the ground.
6 © Crown copyright 2011
Solar radiation has only been widely recorded across the UK in more recent years, so the data period chosen was from 1993 to 2010, a length of 18 years. Of the 100 stations selected, between 19 and 72 stations were available on each day. From 2000 onwards there were at least 50 stations available.
Daily totals of global solar radiation on a horizontal surface were extracted from the Met Office database for these stations, and were converted into average daily irradiance rates in W / m2. The averages over 24 hours (midnight to midnight) will be much less than the peak rate during the middle of the day, as the averaging period includes the night time when there is no radiation received at the surface. Also extracted were radiation totals over an eight hour period when the irradiance rate would be highest on a cloudless day.
Quality control was carried out on the data to ensure that erroneous values were removed.
This was done by calculating the number of standard deviations that the
maximum and minimum station values were away from the mean on each day, and checking any values more than 5 standard deviations away. In addition, maximum values for each month were checked by inspecting hourly values and comparing values with neighbouring stations.
1.2.2 Extreme Value Analysis For each day in the 50 year data series (or 40 years for the eight hour mean), data from all available stations representative of major UK transport routes was pooled to find the site with the most extreme (hot or cold) temperature on that day.
These data were
analysed using extreme value analysis techniques in order to estimate the shape of the tail of the distribution of these statistics. This enables estimates to be made of the values of each statistic expected to occur over a range of return periods up to 1:10,000 years.
The method chosen was a peaks-over-threshold approach using the Generalised Pareto Distribution, because this method makes better use of all available data compared to the annual maximum approach (Coles, 2001).
This method assumes that threshold
exceedances are independent, and this is not the case for daily temperature data because hot and cold temperatures tend to occur in spells of several days. One method for dealing with this is declustering, and Coles describes a method of achieving this called ‘runs declustering’. In this method, a cluster is considered active until r consecutive values fall below the threshold u. When this happens the cluster is terminated, a new cluster being
7 © Crown copyright 2011
initiated on the next exceedance of u. The choices of u and r are a balance between bias and variance, together with sensitivity checks of the results to the choices made. There are also diagnostic plots available to help with the choice of the threshold value and to assess the quality of fit of the model to the data.
The computations described in Coles (2001) are available as a set of functions in the R statistical computing language (R Development Core Team, 2007), and these have been developed into a graphical interface called ‘extRemes’, as described in Gilleland and Katz (2006), which includes a tool for runs declustering. This tool was used to carry out the extreme value analysis, the results of which are described in sections 1.3, 1.4 and 1.5. 95% confidence limits for the return levels were estimated using the profile likelihood method (Coles, 2001). The level of uncertainty associated with the return level estimates increases as you extrapolate to probabilities beyond the range of the data, leading to wider confidence limits, especially for the 1:10,000 year event.
Extreme value analysis is not appropriate for the solar radiation data for two reasons. Firstly, the 18 years length of data is not long enough to obtain reliable results. Secondly, the physical limits on solar radiation mean that values very close to the maximum possible will already have been recorded in the period of data studied.
This will occur on a
cloudless day close to the time of year (or month) when the day is longest and the sun is highest in the sky.
1.3. Maximum temperature results The highest air temperature recorded at any of the selected stations during the period 1961 to 2010 was 38.5°C, at Faversham in Kent on 10 th August 2003. This is also the record highest temperature ever recorded in the UK. Table 1 shows monthly statistics for the series of temperatures from the warmest station on each day. For example, on an average July day the temperature would reach 25.2°C in the hottest part of the UK representative of a major transport route.
Extreme value analysis was carried out on the series of values for each month, as well as for the year as a whole. The series were declustered into hot spell maxima using a runs separation r of 3 days. The threshold was chosen so that between one and two events per year occurred for the monthly models, and approximately three events per year for the whole year model. The results were found to be insensitive to small changes in the
8 © Crown copyright 2011
threshold within a sensible range. Inspection of diagnostic plots showed that the fitted models were generally a good fit to the observed data, and the plots for the annual model are shown in Figure 2. The quantile plot shows a good fit, although the four hottest spells tend away from the fitted model. It is assumed that these events, particularly the hottest event from August 2003, are particularly unusual events which have occurred in this time period. In practise the whole year analysis takes in hot spells above the threshold of 29°C from across the summer months rather than modelling the annual cycle of the whole year.
Figure 2: Model diagnostic plots for the daily maximum temperature (hottest station) over the year as a whole. Probability and Quantile plots give an indication of how well the observations (circles) fit to the fitted model (blue line). The Quantile Plot shows temperature in °C. The Return Level plot shows the return level temperature (°C) against the return period in years, with the black line showing the fitted model, the blue lines showing an indication of upper and lower 95% confidence limits, and the circles again being the observations. The Density plot shows a histogram of the observations against the fitted model (blue line).
9 © Crown copyright 2011
Average
Hottest
Day
Day
Low
Best
High
Low
Best
High
Jan
11.3
18.3
14.3
14.4
14.5
16.2
16.6
17.3
Feb
11.5
19.6
14.5
14.7
14.9
16.9
17.5
18.4
Mar
13.7
25.6
17.2
17.4
17.5
21.2
22.1
23.5
Apr
16.7
27.2
21.1
21.1
21.1
24.5
25.2
26.2
May
20.4
31.9
25.3
25.3
25.4
29.3
29.7
30.5
Jun
23.4
35.6
27.9
28.1
28.4
31.7
32.6
33.8
Jul
25.2
36.5
28.5
29.6
29.8
33.5
34.2
35.1
Aug
24.8
38.5
28.4
28.6
28.8
32.8
33.9
35.8
Sep
21.9
31.6
25.8
25.9
25.9
29.1
29.7
30.7
Oct
17.9
29.5
21.1
21.1
21.2
25.1
26.1
27.3
Nov
14.0
20.8
17.2
17.3
17.4
19.7
19.9
20.3
Dec
12.0
18.1
15.1
15.1
15.4
17.2
17.4
17.8
All
17.8
38.5
31.0
31.3
31.7
34.5
35.2
36.4
Month
1 year return value
10 year return value
Table 1: Monthly and whole year statistics from daily maximum temperatures (°C) for 1961 to 2010, at the hottest station representative of UK transport routes: average and hottest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.13
17.5
18.3
20.2
18.6
20.4
23.3
Feb
-0.15
18.6
19.6
22.0
19.8
22.0
25.7
Mar
-0.19
23.8
25.2
29.3
25.7
28.5
34.2
Apr
-0.35
26.1
27.1
30.0
17.1
28.3
32.3
May
-0.39
30.9
31.4
33.2
31.9
32.4
35.8
Jun
-0.23
34.0
35.2
38.8
35.5
37.6
42.3
Jul
-0.37
35.2
36.2
38.5
36.9
37.4
41.2
Aug
-0.13
35.9
37.9
42.9
38.6
43.0
50.6
Sep
-0.30
30.8
31.6
35.1
32.0
33.0
37.5
Oct
-0.27
27.2
28.7
32.6
29.5
30.9
35.9
Nov
-0.50
20.4
20.8
21.7
20.8
21.1
23.0
Dec
-0.45
18.0
18.2
19.1
18.3
18.5
20.3
All
-0.18
36.6
37.8
40.5
38.5
40.6
44.2
Table 2: Monthly and whole year statistics from daily maximum temperatures for 1961 to 2010, at the hottest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
10 © Crown copyright 2011
The estimated return values for annual probabilities of 1 (1 year event), 0.1 (10 year event), 0.01 (100 year event) and 0.0001 (10,000 year event) are shown in Tables 1 and 2 together with the lower and upper bounds of 95% confidence limits.
The shape parameter is consistently negative for all of the monthly models. The important effect of the negative shape parameter is shown in the return level plot in Figure 2, with the fitted model curving towards the horizontal upper limit as the return period increases. This is expected for temperature variables which have physical limits. However, the range of values of the shape parameter from -0.13 to -0.50 has no physical explanation. The shape parameters for the chosen models for each month are also shown in Table 2. This monthly variation can lead in some cases to implausible results for the low probability events (1:100 and 1:10,000 year return levels). For example, the month of August has a less negative shape parameter which leads to higher estimated return levels for the 1:10,000 year return period than for the year as a whole. At these probabilities, the whole year analysis is the most reliable.
1.4. Minimum temperature results The lowest temperature recorded at any of the selected stations during the 1961-2010 period was -27.2°C on 10 th January 1982 at Braemar, which is located on the A93 road in Aberdeenshire, Scotland.
It also equals the lowest temperature recorded in the UK.
During this cold spell, both Braemar and Grantown-on-Spey had five consecutive days with minimum temperature below -22°C. Table 3 show s monthly statistics for the series of daily minimum temperatures at the coldest station representative of UK transport routes.
Extreme value analysis was carried out on the series of values for each month, as well as for the year as a whole, in the same way as for maximum temperature. The fitted models were generally found to be a good fit to the observed data. Diagnostic plots for the whole year model are shown in Figure 3, with the quantile plot showing a good fit. The whole year model takes cold spells from across the winter months below the threshold of -12°C
The shape parameter is again consistently negative for all of the monthly models (see Table 4). There is a clear seasonal distinction, with the summer months from April to September having more strongly negative shape parameters, leading to fitted models which curve more quickly to an upper bound as the return period increases. The winter months generally have less negative shape such as that portrayed by the return level plot
11 © Crown copyright 2011
for the whole year model in Figure 3. The exception to this is February which has a more negative shape parameter leading to 1:10,000 year return level estimates which are warmer than those for either March or November. This is mainly due to the lack of observed very cold extremes in this month, but there is not physical explanation for this month-to-month variation, so the estimates should be interpreted with caution.
Figure 3: Model diagnostic plots for the daily minimum temperature (coldest station) over the year as a whole.
See Figure 2 for details.
Note that temperatures are negated so that negative
temperatures are shown as positive.
12 © Crown copyright 2011
Average
Coldest
Day
Day
Low
Best
High
Low
Best
High
Jan
-5.6
-27.2
-11.7
-11.5
-11.2
-24.5
-21.1
-19.2
Feb
-5.6
-22.0
-11.5
-11.3
-11.3
-20.9
-19.3
-18.5
Mar
-3.8
-21.7
-8.3
-8.2
-8.1
-19.6
-16.5
-14.7
Apr
-2.4
-11.1
-6.9
-6.7
-6.5
-10.8
-9.9
-9.4
May
0.1
-7.7
-4.1
-3.9
-3.9
-7.3
-6.7
-6.5
Jun
3.0
-4.8
-1.3
-1.0
-0.9
-4.3
-3.7
-3.2
Jul
5.0
-2.8
0.8
0.8
0.7
-2.0
-1.5
-1.2
Aug
4.6
-4.4
-0.2
0.0
0.2
-3.2
-2.6
-2.2
Sep
2.4
-6.1
-2.7
-2.3
-2.2
-5.9
-5.2
-4.7
Oct
-0.1
-9.9
-4.9
-4.6
-4.5
-8.9
-8.2
-7.9
Nov
-3.2
-20.9
-8.6
-8.0
-7.6
-16.8
-14.6
-13.3
Dec
-5.5
-27.0
-11.7
-11.1
-10.6
-22.5
-19.7
-18.0
All
-0.9
-27.2
-15.6
-16.3
-17.1
-22.0
-23.5
-26.2
Month
1 year return value
10 year return value
Table 3: Monthly and whole year statistics from daily minimum temperatures (°C) for 1961 to 2010, at the coldest station representative of UK transport routes: average and coldest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.16
-38.3
-27.9
-24.3
-51.3
-35.9
-23.6
Feb
-0.47
-27.6
-22.0
-20.9
-30.5
-23.2
-22.0
Mar
-0.14
-32.3
-22.5
-19.3
-33.9
-30.0
-18.2
Apr
-0.34
-14.3
-11.4
-10.7
-15.8
-12.4
-11.1
May
-0.42
-9.4
-7.7
-7.5
-10.8
-8.3
-8.0
Jun
-0.35
-6.7
-4.9
-4.3
-8.1
-5.7
-4.9
Jul
-0.33
-3.7
-2.5
-2.1
-5.2
-3.2
-2.8
Aug
-0.28
-5.4
-4.0
-3.4
-7.3
-5.1
-4.4
Sep
-0.43
-8.1
-6.3
-5.6
-9.6
-6.9
-6.1
Oct
-0.41
-11.1
-9.6
-9.2
-13.0
-10.4
-9.9
Nov
-0.14
-24.8
-19.4
-17.2
-33.2
-25.4
-20.7
Dec
-0.13
-33.5
-26.0
-23.0
-45.1
-34.1
-27.3
All
-0.14
-25.7
-28.7
-34.6
-28.6
-35.3
-43.6
Table 4: Monthly and whole year statistics from daily minimum temperatures for 1961 to 2010, at the coldest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
13 © Crown copyright 2011
1.5. Mean temperature results For mean temperature, both the hottest and coldest stations representative of major UK transport routes were considered.
For both of these, results for the daily mean
temperature over both 24 hours and 8 hours will be shown. For the hottest station, the warmest 8 hour period for the time of year is used, while for the coldest station the coldest 8 hour period for the time of year is used.
1.5.1 24 hour mean temperature The highest 24 hour mean temperature recorded at any of the selected stations during the 1961-2010 period was 30.7°C at London Weather Centr e on 10th August 2003, the same day as the highest maximum temperature was recorded nearby at Faversham. Table 5 shows average and maximum values of 24 hour mean temperature at the hottest station, by month and for the year as a whole.
Extreme value analysis was carried out on the series of values for each month, as well as for the year as a whole, in the same way as for maximum and minimum temperature. The fitted models were generally found to be a good fit to the observed data. Diagnostic plots for the whole year model are shown in Figure 4, with the quantile plot showing a good fit. The August 2003 event stands out as being an unusual event within the data period analysed.
The shape parameter is again consistently negative across all months. February and August have the least negative shape parameters, leading to higher estimated 1:10,000 year return levels than would be expected, the August estimate being slightly higher than the estimate from the whole year model.
14 © Crown copyright 2011
Average
Hottest
Day
Day
Low
Best
High
Low
Best
High
Jan
8.6
15.1
11.3
11.4
11.5
13.0
13.4
14.0
Feb
8.4
15.3
11.1
11.2
11.3
12.9
13.4
14.2
Mar
9.7
17.5
12.5
12.7
12.9
15.1
15.7
16.5
Apr
11.6
19.3
15.2
15.2
15.2
18.0
18.2
18.7
May
14.8
24.4
18.6
18.6
18.8
22.2
22.4
23.2
Jun
17.7
28.1
21.3
21.3
21.5
24.6
25.6
26.6
Jul
19.6
27.9
22.9
23.1
23.4
25.9
26.4
27.5
Aug
19.4
30.7
22.0
22.3
22.6
25.8
26.6
28.1
Sep
17.2
24.2
19.6
19.8
20.1
22.2
22.7
23.7
Oct
14.3
22.4
16.8
17.0
17.3
19.3
20.2
21.0
Nov
11.1
17.1
14.2
14.2
14.3
15.9
16.2
16.7
Dec
9.4
15.7
12.3
12.3
12.4
14.4
14.5
15.0
All
13.5
30.7
24.1
24.4
24.7
27.0
27.6
28.7
Month
1 year return value
10 year return value
Table 5: Monthly and whole year statistics from 24 hour mean temperatures (°C) for 1961 to 2010, at the hottest station representative of UK transport routes: average and hottest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.15
14.1
14.8
16.5
15.1
16.5
18.9
Feb
-0.05
14.3
15.3
17.5
15.7
18.6
22.4
Mar
-0.22
16.6
17.5
20.0
17.6
19.2
22.5
Apr
-0.49
18.9
19.2
20.4
19.3
19.6
21.8
May
-0.40
23.8
24.1
26.2
24.7
25.0
28.2
Jun
-0.27
26.4
27.8
31.2
27.9
29.6
34.0
Jul
-0.27
27.4
28.3
31.2
28.1
29.8
33.4
Aug
-0.13
28.3
29.8
33.5
30.6
33.9
39.2
Sep
-0.23
23.6
24.5
27.2
23.8
26.1
29.6
Oct
-0.25
20.7
21.9
24.2
21.9
23.5
26.4
Nov
-0.37
16.6
17.1
18.5
17.0
17.6
19.5
Dec
-0.36
15.3
15.5
16.5
15.9
16.1
17.9
All
-0.13
28.8
30.0
32.5
30.6
33.1
36.8
Table 6: Monthly and whole year statistics from 24 hour mean temperatures for 1961 to 2010, at the hottest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
15 © Crown copyright 2011
Figure 4: Model diagnostic plots for high extremes of 24 hour mean temperature at the hottest station, over the year as a whole. See Figure 2 for details.
The lowest 24 hour mean temperature recorded at any of the selected stations is -18.8°C on 29th December 1985 at Fyvie Castle, near to the A947 road in Aberdeenshire. The station recorded a minimum temperature of -21.7°C a nd a maximum of -15.9°C on this day. This latter value is the lowest daily maximum temperature on record for the UK. Table 7 provides statistics of 24 hour mean temperature at the coldest of the selected stations by month and for the year as a whole.
16 © Crown copyright 2011
Average
Coldest
Day
Day
Low
Best
High
Low
Best
High
Jan
-1.0
-18.7
-4.7
-4.9
-5.3
-11.0
-12.6
-15.5
Feb
-0.7
-14.8
-3.7
-4.0
-4.4
-9.4
-10.7
-12.9
Mar
1.3
-10.7
-1.5
-1.9
-2.4
-6.3
-7.4
-9.5
Apr
3.4
-4.7
0.5
0.5
0.1
-2.9
-3.1
-3.9
May
6.1
-0.6
3.3
3.5
3.1
0.5
1.1
-0.5
Jun
8.8
2.3
6.0
5.9
5.9
4.1
3.7
3.0
Jul
10.6
5.6
8.1
8.1
8.0
6.7
6.4
6.1
Aug
10.5
5.1
7.7
7.7
7.5
5.9
5.8
5.4
Sep
8.4
0.0
5.5
5.3
5.1
2.9
2.3
1.4
Oct
5.5
-2.0
2.4
2.2
1.7
-0.9
-1.1
-1.8
Nov
1.7
-12.8
-1.8
-2.2
-2.7
-6.9
-8.1
-10.0
Dec
-0.7
-18.8
-4.4
-4.8
-5.4
-10.4
-12.0
-14.8
All
4.5
-18.8
-7.9
-8.5
-9.2
-13.5
-14.8
-17.1
Month
1 year return value
10 year return value
Table 7: Monthly and whole year statistics from 24 hour mean temperatures (°C) for 1961 to 2010, at the coldest station representative of UK transport routes: average and coldest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.11
-15.4
-18.6
-26.6
-19.0
-27.0
-39.6
Feb
-0.18
-12.9
-15.1
-21.6
-15.0
-20.0
-29.1
Mar
-0.10
-9.4
-11.9
-16.9
-12.0
-18.2
-25.9
Apr
-0.41
-4.3
-4.6
-6.4
-5.0
-5.3
-8.2
May
-0.18
-0.3
-1.3
-4.1
-0.9
-3.3
-7.1
Jun
-0.19
2.9
2.3
0.2
2.2
0.7
-2.2
Jul
-0.36
5.9
5.7
4.9
5.5
5.2
3.8
Aug
-0.43
5.2
5.1
4.0
4.9
4.7
3.1
Sep
-0.19
1.3
0.4
-2.1
0.1
-1.7
-5.1
Oct
-0.49
-1.9
-2.2
-3.8
-2.3
-2.6
-5.6
Nov
-0.20
-10.3
-12.3
-17.0
-13.1
-17.3
-23.9
Dec
-0.04
-15.0
-18.5
-25.4
-19.8
-29.9
-41.8
All
-0.09
-17.1
-19.9
-24.5
-20.8
-27.6
-34.4
Table 8: Monthly and whole statistics from 24 hour mean temperatures for 1961 to 2010, at the coldest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
17 © Crown copyright 2011
Return values for return periods from 1 year to 10,000 years are shown in Tables 7 and 8. Diagnostic plots for the whole year model of low extremes of mean temperature are shown in Figure 5. The shape parameter is less negative than for hot extremes, and the return level plot shows the effect of this, with only a slight curve in the fitted model. December has the least negative shape parameter of the monthly models, leading to estimated 1:10,000 year return levels which are colder than those for the whole year model. Again there is greater confidence in the whole year model.
Figure 5: Model diagnostic plots for cold extremes of 24 hour mean temperature at the coldest station, over the year as a whole. See Figure 2 for details. Note that temperatures are negated so that negative temperatures are shown as positive.
18 © Crown copyright 2011
1.5.2 Eight hour mean temperature Eight hour mean temperatures were calculated over two periods, one representing the warmest part of the day and the other the coldest period.
For the warm (day time) period, mean temperatures were calculated over the period from 11 to 18 GMT except for the months from September to November, for which the period from 10 to 17 GMT was found to be the warmest on average. The highest eight hour mean temperature was 36.3°C at Wisley on 10 th August 2003. Several other stations in south-east England also had eight hour mean temperatures of over 35°C on this day, which is the same day as the highest 24 hour maximum and mean temperatures were recorded.
Statistics were calculated for the daily series of values at the hottest station representative of major UK transport routes, and these are shown in Table 9. As expected, the values are higher than those for the 24 hour mean temperature but lower than those for the 24 hour maximum temperature.
Extreme value analysis was carried out as for 24 hour mean temperature for each month and for the year as a whole. The models generally fitted well to the data, and diagnostic plots for the whole year model are shown in Figure 6. This shows that the August 2003 hot spell was an unusual event as it stands well above the next hottest event. However, the model for November had a very low shape parameter, and the results for this month are not considered robust.
19 © Crown copyright 2011
Average
Hottest
Day
Day
Low
Best
High
Low
Best
High
Jan
9.0
15.7
12.2
12.3
12.4
13.9
14.3
14.9
Feb
9.1
17.1
12.1
12.2
12.3
14.0
15.2
16.2
Mar
10.9
19.8
15.2
15.2
15.3
17.8
18.3
19.2
Apr
13.7
24.9
18.4
18.5
18.7
22.4
22.8
23.7
May
17.3
30.5
22.4
22.4
22.6
27.3
27.8
29.0
Jun
20.2
33.5
25.6
25.7
26.0
29.4
30.6
32.1
Jul
22.4
33.6
27.1
27.2
27.6
31.0
31.7
32.8
Aug
22.1
36.3
26.0
26.2
26.5
30.3
32.1
34.0
Sep
18.8
27.6
22.5
22.6
23.1
26.2
26.5
27.5
Oct
15.0
25.9
17.8
18.2
18.6
21.8
22.8
24.3
Nov
11.4
17.4
15.1
15.1
15.1
16.8
16.9
18.2
Dec
9.7
16.6
13.1
13.1
13.2
14.9
15.4
16.1
All
15.0
36.6
28.7
29.1
29.6
32.3
33.1
34.5
Month
1 year return value
10 year return value
Table 9: Monthly and whole year statistics from 8 hour mean temperatures (°C) for 1961 to 2010, at the hottest station representative of UK transport routes: average and hottest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.25
14.9
15.4
17.2
15.7
16.5
18.7
Feb
-0.21
15.2
17.0
20.0
16.4
18.9
22.9
Mar
-0.32
19.1
19.8
22.8
20.0
20.9
24.5
Apr
-0.39
24.0
24.5
26.8
24.9
25.5
29.3
May
-0.39
29.3
30.0
32.7
30.5
31.3
35.9
Jun
-0.28
31.5
33.2
37.5
33.7
35.3
40.7
Jul
-0.39
32.5
33.4
36.4
33.4
34.5
38.5
Aug
-0.21
32.9
35.7
41.1
35.5
39.3
46.5
Sep
-0.46
27.5
27.9
30.7
28.1
28.5
32.5
Oct
-0.22
24.2
25.5
29.6
25.8
28.2
33.6
Nov
-0.56
17.2
17.4
17.5
17.5
17.6
19.3
Dec
-0.28
16.0
16.6
18.8
16.8
17.6
20.3
All
-0.17
34.4
35.7
38.7
36.3
38.8
42.7
Table 10: Monthly and whole year statistics from 8 hour mean temperatures for 1961 to 2010, at the hottest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
20 © Crown copyright 2011
Figure 6: Model diagnostic plots for 8 hour mean temperature (11 - 18 GMT) at the hottest station, over the year as a whole. See Figure 2 for details.
For the cold period, mean temperatures were calculated over the period from 02 to 09 GMT for January, 01 to 08 GMT for November, December and February, 00 to 07 GMT for March, April, September and October and 23 to 06 GMT for May to August. The lowest eight hour mean temperature was -21°C at Alt naharra, on the A836 road in northern Scotland on 8th January 2010. Another notably cold night occurred on the 3rd March 2001 when an 8 hour mean temperature of -19.5°C was recorded at Kinbrace, also in northern Scotland. And on 30th December 1995 there was an 8 hour mean temperature of -20.3°C at Aviemore.
Statistics were calculated for the daily series of values at the coldest station representative of major UK transport routes, and these are shown in Table 11.
21 © Crown copyright 2011
Average
Coldest
Day
Day
Low
Best
High
Low
Best
High
Jan
-0.7
-21.0
-6.6
-6.6
-6.7
-13.7
-15.7
-19.5
Feb
-1.0
-18.2
-5.4
-5.9
-6.5
-11.7
-13.4
-16.5
Mar
0.1
-19.5
-4.3
-4.7
-5.3
-10.0
-11.9
-16.0
Apr
1.3
-6.4
-2.6
-2.7
-2.9
-5.0
-5.3
-5.8
May
4.1
-3.8
-0.1
-0.2
-0.2
-3.0
-3.1
-5.1
Jun
7.1
-1.7
3.5
3.4
3.1
0.7
0.2
-0.8
Jul
9.1
1.6
5.6
5.6
5.4
3.3
2.9
2.1
Aug
8.8
0.1
4.6
4.5
4.4
2.0
1.5
0.7
Sep
6.2
-3.3
0.8
0.8
0.7
-1.4
-1.8
-2.6
Oct
4.0
-7.1
-1.4
-1.7
-1.9
-4.3
-5.0
-6.2
Nov
1.3
-17.9
-4.1
-4.4
-4.9
-8.9
-10.5
-13.8
Dec
-0.6
-20.3
-6.9
-7.3
-7.8
-13.5
-15.5
-19.7
All
3.3
-21.0
-10.7
-11.5
-12.4
-17.1
-18.9
-22.6
Month
1 year return value
10 year return value
Table 11: Monthly and whole year statistics from 8 hour mean temperatures (°C) for 1961 to 2010, at the coldest station representative of UK transport routes: average and coldest day and 1 and 10 year return values (best estimate and lower and upper limits of 95% confidence intervals).
Month
Shape
100 year return value
10,000 year return value
parameter
Low
Best
High
Low
Best
High
Jan
-0.17
-18.4
-21.8
-33.1
-21.3
-28.8
-45.1
Feb
-0.11
-16.0
-19.2
-26.8
-19.3
-27.1
-38.7
Mar
0.05
-15.0
-20.0
-28.8
-21.1
-39.3
-58.3
Apr
-0.44
-5.9
-6.2
-7.5
-6.4
-6.7
-8.9
May
-0.62
-3.6
-3.8
-6.2
-3.8
-4.0
-4.1
Jun
-0.29
-0.7
-1.5
-4.2
-1.8
-2.8
-6.2
Jul
-0.35
2.2
1.7
-0.7
1.5
0.9
-1.9
Aug
-0.38
0.8
0.2
-2.1
0.2
-0.6
-3.6
Sep
-0.32
-2.4
-3.1
-5.3
-3.2
-4.0
-6.7
Oct
-0.21
-6.0
-7.1
-10.4
-6.4
-9.2
-13.6
Nov
0.07
-13.3
-17.7
-24.9
-20.0
-36.3
-52.6
Dec
-0.05
-18.4
-22.8
-33.1
-21.9
-34.9
-52.7
All
-0.09
-20.7
-24.8
-32.2
-23.3
-33.5
-44.5
Table 12: Monthly and whole year statistics from 8 hour mean temperatures for 1961 to 2010, at the coldest station representative of UK transport routes: shape parameter of the fitted extreme value model and 100 and 10,000 year return values (best estimate and 95% confidence interval).
22 © Crown copyright 2011
Extreme value analysis was carried out as for 24 hour mean temperature for each month and for the year as a whole. Diagnostic plots for the whole year model are shown in Figure 7. The shape parameters for the monthly models vary considerably, partly due to seasonality and partly to random effects. This makes the monthly return values for 100 and 10,000 year events inconsistent and should be interpreted with caution. In particular, the months of March and November had unusually cold events within the observed record, and this contributes to them having positive shape parameters and more extremely cold return levels for low probability events than expected. May has a very negative shape parameter which also leads to unrealistic estimates of the 1:100 and 1:10,000 year return levels. The whole year model gives the most reliable estimation of the severity of low probability cold spells.
Figure 7: Model diagnostic plots for 8 hour mean temperature (00 - 07, 01 - 08 or 02 - 09 GMT) at the coldest station, over the year as a whole. See Figure 2 for details. Note that temperatures are negated so that negative temperatures are shown as positive.
23 © Crown copyright 2011
1.6. Solar radiation results 1.6.1 24 hour average insolation Table 13 shows the maximum daily average global solar insolation on a horizontal surface recorded at any of the selected stations representative of UK transport routes during each month of the year. Also provided are the 95th and 99th percentile values, which give an indication, based on the 1993 to 2010 period, of the highest values to be expected on major UK transport routes on one in 20 and one in 100 days respectively.
These insolation rates will be much lower than the peak insolation during the middle of the day because they are averaged over a full 24 hours (midnight to midnight) including the night time.
Table 13 shows the locations and dates of the maximum recoded daily
radiation total in each month, and Figure 8 plots out the hourly insolation rates for these events for four of the months. The smooth profile of these days indicate, as expected, that these are days with very little or no cloud.
Only in the April event is there an
indication of a small amount of cloud in the afternoon. The maximum hourly average insolation rate on the day with the highest total radiation (at Liverpool on 27th June 2005) was 976 W / m2 for the hour ending 13:00 GMT. Within this hour the maximum insolation would have been slightly higher. Locations in the south of the UK are more likely to receive the highest radiation amount in the winter.
In summer the longer days
experienced in the north means that northern locations are able to receive similar or even higher daily radiation amounts compared to locations in the south.
24 © Crown copyright 2011
Month
Station with greatest radiation th
th
Maximum value
Average
95 %ile
99 %ile
Max
Location
Date
Jan
50
78
85
90
Cardinham Bodmin
29 January 2006
Feb
90
132
144
158
Cardinham Bodmin
29 February 2004
Mar
158
213
231
249
Hawarden (Chester)
30 March 2000
Apr
239
288
301
312
Camborne
30 April 2001
May
301
353
366
374
Stornoway
26 May 2008
Jun
321
364
373
392
Liverpool Museum
27 June 2005
Jul
297
349
358
363
Stornoway
5 July 2008
Aug
252
306
318
330
Dunstaffnage (Oban)
5 August 1995
Sep
183
236
246
257
Liverpool Museum
2 September 2005
Oct
112
157
167
184
Cardinham Bodmin
8 October 1998
Nov
61
90
100
110
Southsea
1 November 2003
Dec
41
59
72
79
Brooms Barn, Suffolk
31 December 2001
All
176
341
359
392
Liverpool Museum
27 June 2005 2
Table 13: Monthly and whole year statistics of daily (00 to 00) global solar irradiance (W / m ) at the location with the highest value on each day, from 1993 to 2010. Also shown is the location and date of the maximum value.
1000 June
Solar irradiance (W / sq m)
900
April 800
October
700
December
600 500 400 300 200 100 0 1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0
Time of day (GMT) 2
Figure 8: Diurnal cycle of hourly average irradiance values (W / m ) for the days and locations with maximum total radiation in June, April, October and December (see Table 9 for details of locations and dates).
25 © Crown copyright 2011
1.6.2 Eight hour average insolation The profiles in Figure 8 also show that the maximum radiation for an 8 hour period occurs between 08 and 16 GMT for all months. Statistics of the average insolation rates over this period at the station with the maximum value on each day are shown in Table 14. As expected, the values are greater than those averaged over 24 hours, especially for the winter when the 8 hour period includes almost all of the radiation received. The location and date of the maximum values for each month are generally the same except that the Scottish locations no longer feature as their summer diurnal cycles are spread over a slightly longer period of daylight, but with a lower peak insolation.
Month
Station with greatest radiation th
Maximum value
Average
95 %ile
99 %ile
th
Max
Location
Date
Jan
146
229
252
261
Cardinham Bodmin
29 January 2006
Feb
256
369
408
435
Cardinham Bodmin
29 February 2004
Mar
424
565
595
638
Hawarden (Chester)
30 March 2000
Apr
595
702
729
751
Camborne
30 April 2001
May
699
803
823
836
Liverpool Museum
29 May 2005
Jun
726
817
836
861
Liverpool Museum
27 June 2005
Jul
684
787
808
812
Liverpool Museum
13 July 2004
Aug
612
716
749
766
Southsea
2 August 2003
Sep
480
604
631
649
Hawarden (Chester)
10 September 1999
Oct
316
431
461
515
Cardinham Bodmin
8 October 1998
Nov
179
267
293
321
Southsea
1 November 2003
Dec
123
179
215
238
Brooms Barn, Suffolk
31 December 2001
All
437
773
811
861
Liverpool Museum
27 June 2005 2
Table 14: Monthly and whole year statistics of 8 hour (08 to 16) global solar irradiance (W / m ) at the location with the highest value on each day, from 1993 to 2010. Also shown is the location and date of the maximum value.
1.6.3 Insolation on a vertical surface Solar radiation is usually only recorded on a horizontal surface, but the amount of radiation received on inclined planes is often of interest. The amount of radiation received on inclined planes varies according to the position of the sun in the sky and how this relates to the plane of interest. The position of the sun in the sky varies with the time of day and the time of year, as well as the latitude of the location.
26 © Crown copyright 2011
In this case, the plane of interest is a vertical surface, because of the rectangular shape of packages. The orientation of the plane will vary at different times during transportation, but the most exposed orientation is considered, this being south-facing. Various models are available to estimate the solar radiation received on inclined surfaces from that recorded on a horizontal surface. These models require both global (total) radiation and diffuse radiation as inputs (e.g. Badescu, 2008; SolarTool, 2008). Hourly diffuse radiation is only available from two stations in the UK; Camborne and Lerwick, and only up to 1998. Camborne is located in Cornwall and is a location that can receive close to the maximum possible solar radiation in the UK at most times of the year. Lerwick is located on the Shetland Islands, and has the longest days in the UK during the summer.
The series of equations from Solar Tool (2008) have been used to estimate the insolation on a south-facing vertical surface at Camborne and Lerwick from total and diffuse insolation values.
This has been done for the hour from 12 to 13 GMT on the days of
each month of the year when the radiation for this hour was the greatest (and diffuse radiation data was available).
These hours generally would have had clear sky
conditions, but due to the restricted time period of data available some of the hours may have had some cloud, which will affect the results. The results are shown in Table 15.
Month
Camborne
Lerwick
Horizontal
Vertical
Ratio
Horizontal
Vertical
Ratio
January
369
791
2.15
167
644
3.86
February
551
853
1.55
364
668
1.83
March
756
827
1.09
583
866
1.48
April
870
723
0.83
756
824
1.09
May
946
606
0.64
842
763
0.91
June
954
576
0.60
868
699
0.81
July
926
569
0.61
847
710
0.82
August
870
622
0.72
732
731
1.00
September
783
746
0.95
615
799
1.30
October
614
823
1.34
431
661
1.53
November
393
786
2.00
217
490
2.26
December
287
825
2.88
115
343
2.98
2
Table 15: Horizontal and estimated vertical insolation (W / m ) for the hour from 12 to 13 GMT at Camborne and Lerwick, on the days of each month when radiation was greatest for that hour. The ratio of vertical to horizontal insolation is also shown.
27 © Crown copyright 2011
The results show a clear annual cycle of the ratio between vertical and horizontal insolation, with the greatest ratios occurring in the winter, and the lowest in the summer. The ratios are higher at Lerwick than at Camborne. This means that the annual cycle of insolation on a vertical surface varies much less than that for a horizontal surface. Maximum values of insolation on a vertical surface tend to occur in the spring and autumn months.
1.7 Discussion of results A range of temperature and insolation statistics have been presented. The approach taken has been to identify areas of the UK which are representative of major transport routes, and to consider the most extreme conditions (worst case) recorded at any stations within that area on each day. This is in line with the stated requirement to consider the whole of the UK that is representative of major transport routes as one area, because a journey involving the transport of a package may cover a substantial distance across any part of the UK. Another possible approach would be to take the spatial average across all stations within the representative zone on each day. This is not considered likely to be of use because this would not reveal the extreme conditions which are of interest.
For high and low temperature extremes, three different statistics have been considered based on different durations of exposure. The peak temperature at any time of day, the temperature averaged over the hottest eight-hours of the day, and the 24 hour average temperature (the average of the maximum and minimum temperatures). The choice of statistics may be based on how quickly packages respond to changing temperature, and how this affects the design of packages. The impact of the choice of statistic on the results is illustrated in Figure 9, using the 1:10 year return levels as an example. The uncertainty of these estimates is also shown on these charts in the form of 95% confidence limits.
28 © Crown copyright 2011
-12 Temperature (deg C)
Temperature (deg C)
38 36 34
Max
32
8-hour
30
24-hour
28
-16 24-hour -20
8-hour Min
-24 -28
26
Figure 9: 1:100 year return levels of a) hot temperature extremes and b) cold temperature extremes, showing the impact of time averaging duration.
Upper and lower limits of 95%
confidence intervals are shown as solid lines in the same colour as the best estimate (dashed line).
For each temperature statistic representing different duration of exposure, there are ranges of statistics presented to represent different probabilities of occurrence. As well as the average and most extreme day in the observed record, extreme value analysis has been carried out in order to estimate the values associated with probabilities of occurrence ranging from 1:1 year to 1:10,000 years. Here, the choice of statistic is likely to be based on an acceptable level of risk or a balance between the cost of design and the cost of times when packages could not be transported due to the weather conditions. For example, if designing to a 1:10 year return level, you would expect on average one spell every 10 years to be more extreme than the temperature designed to. Although the hot or cold spell may last for several days, the return level designed to is only likely to be exceeded on the hottest or coldest day of the spell. The impact of the choice of statistic on the results is illustrated in Figure 10, using the example of the eight hour temperatures, and showing the range of probabilities associated with this statistic.
This figure also
shows how uncertainty increases when extrapolating to lower probability events. Although the confidence intervals do overlap in some cases, in reality the order of the values would not change.
For the hot extreme models, there is a tendency for the hottest events (especially the August 2003 event) to be hotter than predicted by the fitted model. It is assumed that these events belong to the same population but are particularly unusual events which have occurred in this time period. However there could be a different population for very hot events with a less negative shape parameter. This could come about through a division of hot days into those where evaporative cooling takes place and those where it does not because it is too dry (Brabson et al, 2005).
29 © Crown copyright 2011
Temperature (deg C)
44
40 1:1 1:10
36
1:100 1:10,000
32
28 -10
Temperature (deg C)
-15 -20
1:1 -25
1:10
-30
1:100 1:10,000
-35 -40 -45
Figure 10: Eight hour mean temperatures for a range of annual probabilities of occurrence: a) hot extremes and b) cold extremes. Upper and lower limits of 95% confidence intervals are shown as solid lines in the same colour as the best estimate (dashed line).
The temperature results are summarised in Tables 16 (high extremes) and 17 (low extremes), showing the best estimate values for a range of return periods and durations of exposure.
Statistic
1 year
10 years
100 years
10,000 years
Maximum
31.3
35.2
37.8
40.6
8 hour mean
29.1
33.1
35.7
38.8
24 hour mean
24.4
27.6
30.0
33.1
Table 16: Summary of results for high extremes of temperature, showing the best estimate temperatures (°C) for a range of return periods and durations of exposure.
30 © Crown copyright 2011
Statistic
1 year
10 years
100 years
10,000 years
Minimum
-16.3
-23.5
-28.7
-35.3
8 hour mean
-11.5
-18.9
-24.8
-33.5
24 hour mean
-8.5
-14.8
-19.9
-27.6
Table 17: Summary of results for low extremes of temperature, showing the best estimate temperatures (°C) for a range of return periods and durations of exposure.
As well as the whole year results used so far in this section, results have been provided for each month of the year. These results may be useful where movements of packages are to be restricted to particular times of year. There is less confidence in the results for individual months compared to the whole year results, so that results for probabilities extrapolated beyond the range of the data (1:100 and 1:10,000 year probabilities) should be used with caution.
Different statistics have been provided for solar insolation because extreme value analysis is not appropriate for these data. Averages over both 24 hours (00 to 00) and eight hours (08 to 16) have been provided. The eight hour averages are higher and likely to be most useful, because the 24 hour average values include the hours of darkness when no solar radiation is received. As for temperature, the average and maximum value across all days in each month have been provided (again based on the series of values from the station with the highest value on each day).
The maximum values will be close to the
physical limit on insolation. This will occur on a cloudless day near to the time of the month or year and the location when and where the combination of the position of the sun in the sky and the length of day leads to a maximum potential insolation. Instead of return values for probabilities down to 1:10,000 years, 95 and 99 percentile values are provided. The 99 percentile values are those which occur once every 100 days on average, which is approximately equivalent to once every three years for a monthly value or three times per year for the whole year value. This gives an indication of the expected frequency of occurrence of such days with high insolation levels.
The results presented in Part 1 of this report have been based on analysis of historical observed data over the last 50 years. The analysis has not taken temporal trends into account, but Part 2 of this report provides further information on the impact of climate change on temperature and insolation values.
31 © Crown copyright 2011
Part 2: Climate Change Information 2.1 Introduction Over the last century the Earth’s climate has changed (Houghton et al., 2001; Levitus et al., 2001). The mean global temperature has risen by approximately 0.74 ˚C, and in addition there have been reductions in snow cover and sea ice extent, and observations show the sea level and the heat held by the oceans have increased (Zwiers, 2002). All model projections indicate that this warming is likely to continue into the future, regardless of the ‘emissions scenario’ used. This warming will also affect other climate variables.
Part 1 of this report has provided information on the current range of temperature and solar radiation values experienced around the UK. This information has been given for areas within 1km both sides of all main roads and railway lines across the UK, and has identified the locations of extreme values of temperature and solar insolation. Because the results in Part 1 have considered UK transport routes as a single area, they cannot be directly compared to the results in Part 2. The results presented in Part 2 will, however, complement those from Part 1 by showing how the range of conditions may change in the future.
This report will provide an initial assessment of model projections of the two variables over the next few decades, and will consider the changes possible in response to climate change. It will use the UKCP09 projections to establish maximum, mean and minimum temperatures for the 2020s and 2030s and the uncertainty in these projections at a 25km grid resolution. These will be graphically presented, showing the relevant percentiles of future distributions (10th, 50th and 90th percentiles) to represent the uncertainty in the projections. Change in total short wave radiation flux (equivalent to ‘solar radiation’, and relative to the 1961-1990 average) will also be presented alongside the relevant uncertainty ranges. Future absolute radiation flux projections are not available; however this information can be used to give an indication of future radiation strengths through comparison with the observed radiation properties presented in the first report.
2.1.1 UKCP09 Projections The UK Climate Projections (UKCP09) are the most recent set of climate change scenarios released by the UK Climate Impacts Programme. UKCP09 describes how the
32 © Crown copyright 2011
UK climate might change during the 21st Century using leading climate science developed at the Met Office Hadley Centre (for more detail see Jenkins et al. 2009). The main difference from its predecessor (UKCIP02, Hulme et al. 2002)) is the systematic incorporation of some of the uncertainties associated with the climate system (see Figures 11 and 12).
From UKCIP02 only a single representation of change was available
With IPCC AR4, multiple models allowed quantification of structural uncertainty
In UKCP09, a statistical framework accounting for known uncertainties allowed probabilistic projections.
Figure 11: An illustration of the shift in methodology from the single model projections of UKCIP02 to the probabilistic projections of UKCP09.
The Intergovernmental Panel on Climate Change
Fourth Assessment Report (IPCC AR4) used models from the many different modelling centres around the world.
Modern, complex climate models are an essential tool for simulating and understanding the present-day and future climate.
Confidence in model estimates of future climate
evolution has been enhanced significantly in the last few years since the IPCC Fourth Assessment Report (AR4).
Climate models have been demonstrated to reproduce
observed features of recent climate and past climate changes and there is considerable confidence that Atmosphere-Ocean General Circulation Models (AOGCMs) provide credible quantitative estimates of future climate change, particularly at continental and larger scales (Randall et al., 2007). However, due to resolution constraints, caused by limitations to computing power, and uncertainties inherent in the modelling process, simulations cannot be expected to perfectly replicate reality. The quantification of these uncertainties and their conversion into probabilistic projections formed the basis of the UKCP09 findings (Figure 12). Multiple simulations of the Met Office Hadley Centre’s
33 © Crown copyright 2011
global climate model were created using perturbations of uncertain model parameters and switches. These simulations covered the present-day and the future. Using a Bayesian statistical framework, these were combined with output from several other international global climate models, observations and further high resolution (25km) regional climate simulations to generate the final projections. (Murphy, et al., 2009).
UKCIP02 Single projection
UKCP09 Central estimate (50%)
Very unlikely to be more than (90%)
Summer Rainfall 2080’s
Very unlikely to be less than (10%)
Figure 12: Moving from uncertainty to probability, an example of the difference between results for UKCIP02 and UKCP09 for summer mean rainfall across the UK for the 2080s for the high emission scenario (Hulme et al., 2002; Jenkins et al., 2009).
The probabilistic projections derived from this framework ‘represent the relative degree to which each climate outcome is supported by the evidence currently available, taking into account our understanding of climate science and observations, and using expert judgement’ (Murphy, et al., 2009). The ‘lower 10%’ estimate gives changes which are very likely to be exceeded whilst the ‘upper 90%’ estimate gives changes very unlikely to be exceeded. Providing information in this probabilistic way is useful for planning and policy development as users can start to assess expected losses, costs and benefits that might occur over a range of likely outcomes (for further discussion see Zwiers, 2002).
UKCP09 handles uncertainty arising from differing greenhouse gas emissions by considering separate pathway scenarios used by the IPCC (Meehl et al., 2007). These
34 © Crown copyright 2011
scenarios are not intended to act as a comprehensive set of possible emission futures and currently no scenario is viewed as more likely than any other. Projections are given for three scenarios; low, medium and high, referred to as ‘B1’, A1B’ and ‘A1FI' (Nakicenovic and Swart, 2000) by the IPCC. The low emissions scenario follows a pathway of clean technologies replacing fossil fuel use. The medium is based on a mixture of future fossil fuel and non-fossil fuel whilst the upper is reflective of a pathway predominantly fossil fuel based (see Figure 13). The A1B scenario replaces the two medium emissions scenarios used in UKCIP02 (medium-high and medium-low) that corresponded to the IPCC SRES B2 and A2 so that only one medium emissions scenario is given in UKCP09.
Figure 13: The three SRES emissions scenarios used in UKCP09 are A1FI (high), A1B (medium) and B1 (low).
Where an analysis is concerned with only the first half of the 21st century it is usual to consider only the one emissions scenario, as until the 2050s there is little divergence in the modelled climate variables between scenarios.
This is largely because a good
proportion of the change in the next few decades will result from the climate (in particular the ocean) adjusting to the change in the atmosphere that have already occurred (Zwiers, 2002). For this report we will therefore focus on results from the medium emissions scenario (A1B) for the 2020s and 2030s. We will, however, present some results from the latter part of the century (2080s) for illustrative purposes (see for instance Appendix 1).
The UKCP09 projections are provided for seven overlapping thirty year time-periods with reference to the 1961-1990 baseline. Each future time period is referred to by its central decade, so the 2020s refers to 2010-2039 and the 2030s refers to 2020-2049 and so on. This report will use UKCP09 probabilistic projections for the decades 2020s and 2030s, where the decade is represented by the 30 year period centred on that decade.
35 © Crown copyright 2011
The spatial resolution of the model has improved since UKCIP02, and grid boxes are now 25km rather than 50km allowing better representation of finer scale patterns. Results are aggregated for 16 administrative regions (Figure 14), 23 river basins, and 9 marine regions.
Figure 14: An illustration of the resolution of the UKCP09 climate projections showing the 25km grid (left) and the 16 administrative regions (right).
This report provides a summary of the projected UKCP09 changes across the UK in mean daily temperature, daily temperature extremes, and maximum solar insolation values. Temperature and insolation changes are presented for annual mean, and also for seasonal mean changes where relevant.
A summary considering the usefulness of projecting the changes presented on climate change timescales is also given, as well as an assessment of additional methods that may provide further robustness to this initial analysis.
36 © Crown copyright 2011
2.2 Climate change temperature information 2.2.1 Discussion of variables Future mean daily maximum, mean and minimum temperatures for each decade are available at a 25 km grid resolution across the UK.
It is important to note that the
temperature information given is a temporal mean of the variable, for example the maximum temperature data available is the mean of daily maximum temperatures across the 30 year period at a location, and therefore individual days with extreme maximum temperatures will not be captured using this information. Different sources of information are available to consider how future temperatures could change, but in this assessment UKCP09 probabilistic information is used. For a full assessment which may capture more of the individual peaks a combination of information would be necessary. In particular to explore extreme events regional climate model data would be more appropriate but is outside the scope of this project. However, a more complete representation of the known uncertainties is captured by the probabilistic information given here than use of the regional climate model data alone would give and therefore the projections are still of value to help understand future temperature extremes.
Given that extremes of temperature are of greatest interest here we have selected information about the annual mean, minimum and maximum across the 2020s and 2030s. We have then considered the mean daily temperature and the mean daily maximum temperature for the summer months (June, July, August) for both decades (see Appendix 2 for maximum temperatures in other months), and also mean daily temperature and the mean daily minimum temperature for the winter months (December, January, February) for both decades (see Appendix 3 for minimum temperatures in other months).
Further to this, plots are included of the change in maximum temperature of the warmest day in summer (this can be thought of as the change in the extreme value of the season, although strictly we have used the 99th percentile of the daily distribution of maximum temperature over the 30 year period) and the change in the minimum temperature of the coolest day in winter (again the change in the extreme value of the season, or the 99th percentile of the daily distribution of minimum temperature over the 30 year period). This should allow for the seasonal variations to be captured and will give an indication of the range of temperatures we might expect in the future.
37 © Crown copyright 2011
Please note the upper and lower estimates (90th and 10th percentile) given throughout are not representative of the entire range. We indicate here that only 10% of results fall above the value of the 90th percentile, and 10% fall below the value of the 10th percentile. These estimates should therefore not be used as an absolute limit on temperatures and solar flux in any way, but as an indication of the range considered very likely to be within based on the uncertainties included in the UKCP09 analysis.
2.2.2 Results Annual mean daily mean temperature, daily maximum temperature and daily minimum temperature th
th
th
Figure 15a) 2020s Annual mean daily mean temperature – 10 , 50 , 90 percentile.
th
th
th
Figure 15b) 2020s Annual mean daily minimum temperature – 10 , 50 , 90 percentile.
38 © Crown copyright 2011
th
th
th
Figure 15c) 2020s Annual mean daily maximum temperature – 10 , 50 , 90 percentile.
39 © Crown copyright 2011
th
th
th
Figure 15d) 2030s Annual mean daily mean temperature – 10 , 50 , 90 percentile.
th
th
th
th
th
th
Figure 15e) 2030s Annual mean daily minimum temperature – 10 , 50 , 90 percentile.
Figure 15f) 2030s Annual mean daily maximum temperature – 10 , 50 , 90 percentile.
40 © Crown copyright 2011
Summer (JJA) mean daily mean temperature, daily maximum temperature and warmest day maximum temperature. th
th
th
Figure 16a) 2020s Summer (JJA) mean daily mean temperature – 10 , 50 , 90 percentile.
th
th
th
Figure 16b) 2020s Summer(JJA) mean daily maximum temperature – 10 , 50 , 90 percentile.
th
Figure 16c) 2020s Change in the maximum temperature of the warmest day in Summer – 10 , th th 50 , 90 percentile.
41 © Crown copyright 2011
th
th
th
Figure 16d) 2030s Summer (JJA) mean daily mean temperature – 10 , 50 , 90 percentile.
th
th
th
Figure 16e) 2030s Summer(JJA) mean daily maximum temperature – 10 , 50 , 90 percentile.
th
Figure 16f) 2030s Change in the maximum temperature of the warmest day in Summer – 10 , th th 50 , 90 percentile.
42 © Crown copyright 2011
Winter (DJF) mean daily mean temperature, daily minimum temperature and coolest day temperature. th
th
th
Figure 17a) 2020s Winter (DJF) mean daily mean temperature – 10 , 50 , 90 percentile
th
th
th
Figure 17b) 2020s Winter (DJF) mean daily minimum temperature – 10 , 50 , 90 percentile.
th
Figure 17c) 2020s Change in the minimum temperature of the coolest day in Winter – 10 , th th 50 , 90 percentile.
43 © Crown copyright 2011
th
th
th
Figure 17d) 2030s Winter (DJF) mean daily mean temperature – 10 , 50 , 90 percentile
th
th
th
Figure 17e) 2030s Winter (DJF) mean daily minimum temperature – 10 , 50 , 90 percentile.
th
Figure 17f) 2030s Change in the minimum temperature of the coolest day in Winter – 10 , th th 50 , 90 percentile.
44 © Crown copyright 2011
2.2.3 Discussion of results 2.2.3.1 Annual mean of daily mean temperatures The projections for annual mean of all daily temperature, daily maximum temperature and daily minimum temperature for the years 2010-2039 (2020s) and 2020-2049 (2030s) retain the current spatial characteristics of the UK climate, with temperatures still highest in the south of England and lowest in central Scotland (Figures 15a-f). For the majority of the UK projections of annual mean daily temperatures are between 6°C and 12°C for the 2020s and between 6°C and 15°C for the 2030s (Figur es 15a and 15d). Projections of mean temperatures are fairly consistent across the UKCP09 analysis range.
Projections of annual mean daily minimum temperatures (Figures 15b and 15e) suggest similar results to the mean with the majority of the UK in the 2020s projected temperatures of 3°C to 9°C (for the 2030s the figure is also 3°C
to 9°C). Exceptions to this include
central Scotland (range of 0°C to 6°C for mean dail y mean temperature and minimum temperature) and coastal England and Wales (range of 3°C to 12°C for mean daily minimum temperature).
Projections of annual mean daily maximum temperature (Figures 15c and 15f) display a lower degree of confidence (greater difference across the UKCP09 analysis range), with temperatures across the majority of the UK ranging from 9°C to 18°C for the 2020s (for the 2030s the figure is also 6°C to 18°C). This ra nge of temperatures is due to both uncertainty in the UKCP09 analysis range and spatial variation across the UK.
The
greatest maximum temperatures are projected for southeast England (12°C to 18°C).
2.2.3.2 Summer mean of daily mean temperatures The highest values of summer mean daily mean temperature and daily maximum temperature are projected for the southeast of England, with daily mean of 15°C to 21°C, and daily maximum 18°C to 27°C for the 2020s (Figur es 16a and 16b). Results for the 2030s are similar (Figures 16d and 16e), although a greater proportion of the UK is projected to experience the higher temperatures of up to 27°C (in the 90 th percentile projections).
Projections of the change in maximum temperature of the warmest day in summer in the 2020s show much greater variation across the UKCP09 analysis range, with change in temperature of the warmest day ranging from -4°C to +6°C (Figure 16c).
45 © Crown copyright 2011
The 10 th
percentile runs project a negative change across the whole of the UK, whereas the 50th and 90th percentiles project positive changes to temperatures of up to 2°C and 6°C respectively. These changes are spatially consistent across the UK.
Projections for the 2030s are less spatially coherent (Figure 16f), and project further increases in temperature of up to an 8°C increase i n maximum temperature of the warmest day in summer for northern England, southern Scotland and parts of central southern England in the 90th percentile runs. 10th percentile estimates are still negative across the UK (a possible decrease in temperature of up to 2°C).
2.2.3.3 Winter mean of daily mean temperatures Across the majority of England and Wales winter mean daily temperature and daily minimum temperature are projected to be between 0°C to 9°C and 0°C to 6°C respectively (Figures 17a, b, d and e). The lowest daily temperatures are reached in central Scotland with mean temperatures ranging from -3°C to +6°C and minimum temperatures from -3°C to +3°C. These projections change very little between the 2020s and 2030s with most of the range of results coming from spatial variation across the UK and differences across the UKCP09 analysis range.
Projections of the change in minimum temperature of the coolest day in winter for the 2020s (Figure 17c) show spatial consistency across the UK, but with large variation across the UKCP09 analysis range (projections for the UK as a whole range from -2°C to +4°C. Negative changes are projected across the UK by the 10th percentile, and the 50th and 90th percentiles project increases in the temperature of the coolest day right across the UK. Again there is no discernible change from these results by the 2030s (Figure 17f).
46 © Crown copyright 2011
2.3 Climate change solar insolation information 2.3.1 Discussion of variables Total downward surface shortwave flux is a measure of the amount of shortwave radiation received by a unit area per unit time at the Earth’s surface. The total downward surface shortwave flux is equivalent to global irradiance or insolation. The UKCP09 projections provide information on the 30-year average of monthly, seasonal and annual total downward surface shortwave flux. It is important to note that there is significantly less scientific understanding of solar radiation than future temperatures, due to uncertainties associated with future cloud distributions and amounts (see for e.g. Wu et al., 2004).
Future absolute flux projections are not available, and therefore we provide plots of change in total short wave radiation flux (relative to the 1961-1990 average) which can be used to give an indication of future irradiance strengths through comparison with the observed insolation properties presented in Part 1 of this report.
Again, as extremes of solar insolation are of greatest interest here we have selected information about change in maximum total short wave radiation on an annual basis across the 2020s and 2030s. We have then considered the change in insolation values for the summer months (June, July, August) for both decades as this is when maximum solar insolation occurs (see Appendix 4a-c for other months).
47 © Crown copyright 2011
2.3.2 Results Change in annual mean daily maximum total downward surface shortwave flux th
th
th
th
th
th
Figure 18a) 2020s Annual mean daily maximum flux – 10 , 50 , 90 percentile.
Figure 18b) 2030s Annual mean daily maximum flux – 10 , 50 , 90 percentile.
48 © Crown copyright 2011
Change in Summer mean daily maximum total downward surface shortwave flux th
th
th
th
th
th
Figure 18c) 2020s Summer (JJA) mean daily maximum flux – 10 , 50 , 90 %ile
Figure 18d) 2030s Summer (JJA) mean daily maximum flux – 10 , 50 , 90 %ile
2.3.3 Discussion of results The projections of change in annual mean daily maximum solar flux (Figures 18a and 18b) indicate that changes may be small, and there is little to distinguish between changes expected in one decade and the next (again the emission scenario chosen has little bearing on the results this early in the century and therefore only the medium scenario is presented here). Projections at the 10th percentile for both the 2020s and 2030s show a small decrease in solar radiation (of up to -10 Wm-2), whilst 50th and 90th percentiles project a small increase (up to 10 Wm-2). The changes projected are spatially
49 © Crown copyright 2011
consistent across the UK. The changes projected are negligible in the scale of changes experienced seasonally at present (Murphy et al., 2009).
More significant changes are projected in the change in summer mean daily maximum solar flux (Figures 18c and 18d), particularly in the upper end projections. At the 90th percentile for the 2020s an increase in solar radiation of up to 30 Wm-2 is projected for the southwest of England and parts of Wales. For the 2030s there is a slight expansion in areas projected to receive between 10 Wm-2 and 20 Wm-2 in the 50th percentile projections and areas receiving up to 30 Wm-2 in the 90th percentile now cover the whole of the southwest region and Wales.
The projections here highlight an increase in the expected maximum solar flux on an average summer’s day. These changes are expected due to a reduction in average cloudiness across the summer months (Murphy et al., 2009). However, the highest solar flux values at any point during the summer would occur in clear sky conditions, for instance during a heat wave. Given that the total incoming shortwave radiation from the sun will not change on these timescales, days with clear skies (and therefore maximum solar flux) in the future will experience levels of solar flux no higher than the present maximum (although results shown here and in the wider literature (see section 2.4) suggest that clear sky days may occur more frequently).
It will be particularly important to consider the joint occurrence of maximum temperatures and maximum solar flux when using the data presented here. Greatest increases in solar flux associated with decreases in cloud cover are unlikely to occur during times of projected maximum temperature, when cloud cover is already low. Therefore it is not appropriate to combine the highest projected values of temperature from section 2.2 of this report with the greatest changes in solar flux projected in section 2.3.
2.4 Wider context The projections shown here give results broadly consistent with those used in the wider peer-reviewed literature. The incidence of heat waves in Europe in summer increased during the 20th century (Schär et al., 2004; Beniston and Stephenson, 2004; McGregor et al., 2005). Regional climate model studies conclude that this trend is likely to continue through the next century, and extreme temperature events such as the European heat wave in 2003 may become more common (e.g. Zwiers and Kharin, 1998; Huth et al.,
50 © Crown copyright 2011
2000; Kharin and Zwiers, 2000; Miehl et al., 2000; Stott et al., 2004; Meehl et al., 2007). However the IPCC note the need to improve the accuracy and regional detail of projections as the current data and analyses are not sufficient to draw robust conclusions. The wider literature, for instance, do not discuss in detail the incidence of warm periods outside the summer seasons as we have done here.
Some authors make suggestions of climatic zones moving northward (one estimate gives a shift of 400-500km by the end of the twenty first century), potentially giving the UK an occurrence of days with extreme high temperatures (above 30°C) similar to what parts of southern France may experience at present (Beniston et al., 2007). Gosling et al. (2009) used the Hadley Centre model HadCM3 to consider the impacts of changing temperatures on heat-related mortality in cities across the world including London and found that higher mortality in future projections was attributed both to increases in the mean temperatures but also in increased variability of temperatures.
This has implications for the
interpretation of projections shown here and elsewhere.
Further studies looking at regional scale models over the UK consider the uncertainty of projections. Rowell (2006) notes that whilst the regional scale model ‘adds significant value to the user, it inevitably contributes a further source of uncertainty’. The author finds that the uncertainty in UK mean climate projections tends to be greatest in the summer season, and of a similar magnitude to large-scale natural climate variations, however confidence is greater for temperature predictions than for precipitation. The dominant source of uncertainty is found to arise from the structure and physics of the GCMs used for downscaling rather than the RCMs themselves.
Rivington et al. (2008) also considered the skill of the Hadley Centre regional climate model HadRM3 in representing weather characteristics across the UK. It was found that the model performs very well over some regions and variables, but poorly for others. In particular maximum temperature estimates were generally represented well, however cold extremes and minimum temperatures were overestimated. Solar radiation was also found to be overestimated in general, but good results were produced for some regions.
Several other authors have used UKCIP analysis for a range of applications. Semenov (2011) for instance uses the UKCP02 analysis to consider the impacts of extreme temperatures among other variables on the production of wheat in England and Wales. Their results suggest that predicted increases in maximum temperatures by 2050 with a
51 © Crown copyright 2011
high emissions scenario are between 2°C and 4°C wit h highest values reached in August. Using UKCIP02-based projections for the UK, Semenov (2007) demonstrated that by the end of the 21st century the frequency of heat waves across the UK could increase substantially (by an order of magnitude), and also the length and severity would increase, with higher peak temperatures reached.
A recently published review of the UKCP09 methods and results concluded that the data used in this analysis represent a large step beyond UKCIP02. However the review did caution careful use of the data and awareness of assumptions made. They also point to the importance of the skill of the global climate model and state that errors in this overlying model ‘cannot be compensated by any downscaling (as used in UKCP09) and will be reflected in uncertainties on all scales’ (UKCP09 Steering Group, 2011).
Ongoing
research is considering structural uncertainties of GCMs, such as the resolution of the stratosphere and the influence this may have on mid-latitude storm systems (Scaife et al., in press) such as those experienced in the UK. Whilst this has been shown to have a potentially significant impact on projections of winter winds, rainfall and flooding, further research is being done to establish whether this may influence other variables such as temperature extremes. Scaife et al. conclude that making these types of changes to the underlying structure of the GCM could potentially ‘represent a first-order correction to climate projections for the mid-latitudes’. There is research ongoing to identify precisely how much influence this would have on projections such as those analysed in this report.
2.4.1 Limitations of this methodology Despite the benefits of the probabilistic information presented here, this method does have some limitations. This assessment gives a change across a 30 year averaged period rather than the maximum changes seen during those 30 years. It may, for instance be useful to consider not just the 1 in 1 year event as described here (warmest day in summer and coolest day in winter figures), but also more extreme values such as the 1 in 10 or 100 year event as done in the first part of this report. In addition it may be useful to consider changes to the frequency of extreme insolation values, in addition to the change in maximum values as we have described.
To better capture the temperature and solar flux extremes a more detailed analysis could be undertaken. This analysis would make use of the Met Office Hadley Centre regional
52 © Crown copyright 2011
climate model projections for the UK, combined with statistical analysis to better quantify the changes in UK temperature and solar insolation into the future.
Conclusions Part 1 of this report has shown the range of temperatures which has been observed close to major transport routes in the UK. As well as the maximum and minimum temperatures, results have been presented for the mean temperature over eight hour and 24 hour periods. The observed data has been analysed in order to estimate temperatures which are expected to occur with annual probabilities of 1, 0.1, 0.01 and 0.0001. In addition, eight hour and 24 hour average insolation values have been provided, and the maximum observed values have been shown to be close to the maximum possible.
The results presented in Part 2 of this report concur with the wider literature and find that annual and seasonal mean and maximum temperatures across the UK can be expected to rise under scenarios of continued climate change.
The majority of the UKCP09
analysis range also projects a UK-wide increase in the mean temperature of the warmest summer day, which will have implications for the transportation of temperature-sensitive materials. Minimum temperatures across the UK in winter and annually, as well as the winter mean temperature of the coolest day are also very likely to rise, thereby reducing cold extremes.
The UKCP09 analysis range presented here does however note the
possibility of a reduction in coolest day temperatures and in warmest day temperatures. As the scientific understanding improves through time updated estimates of mean temperature changes will be available.
53 © Crown copyright 2011
References Badescu V (2008). Modeling solar radiation at the earth surface. Springer. Beniston, M (2004) The 2003 heat wave in Europe: a shape of things to come? Geophys Res Lett 31:L02202 Beniston, M. Stephenson, D.B., Christensen, O.B., Ferro, C.A.T., Frei, C., Goyette, S., Halsnaes, K., Holt, T., Jylhä, K., Koffi, B., Palutikof, J., Schöll, R., Semmler, T., Woth, K. (2007). Future extreme events in European climate: an exploration of regional climate model projections. Climatic Change 81:71-95 Brabson BB, Lister DH, Jones PD, Palutikof JP (2005). Soil moisture and predicted spells of extreme temperatures in Britain. Journal of Geophysical Research Atmospheres 110. Coles, SG (2001). An introduction to statistical modelling of extreme values. Springer. Dall’Amico M, Hornsteiner M (2006). A simple method for estimating daily and monthly mean temperatures from daily minima and maxima. International Journal of Climatology 26: 1929 – 1936. Gilleland E and Katz RW (2006). Analyzing seasonal to interannual extreme weather and climate variability with the extremes toolkit (extRemes), 18th Conference on Climate Variability and Change, 86th American Meteorological Society Annual Meeting, Atlanta. Gosling, S.N., McGregor, G.R., Lowe, J.A. (2009) Climate change and heat-related mortality in six cities Part 2: climate model evaluation and projected impacts from changes in the mean and variability of temperature with climate change. Int J Biometeorol 53:31-51 Houghton, J.T. et al. (eds) (2001), Climate Change 2001: The Scientific Basis (Cambridge University Press, 2001). Hulme M, Jenkins GJ, Lu X, Turnpenny JR, Mitchell TD, Jones RG, Lowe J, Murphy JM, Hassell D, Boorman P, McDonald R and Hill S (2002) Climate Change Scenarios for the United Kingdom: The UKCIP02 Scientific Report, Tyndall Centre for Climate Change Research, School of Environmental Sciences, University of East Anglia, Norwich, UK.
54 © Crown copyright 2011
Huth, R., Kysely, J., Pokorna, I. (2000) A GCM simulation of heat waves, dry spells, and their relationships to circulation. Clim Change 46: 29–60 Jenkins, G.J., Murphy, J.M., Sexton, D.S., Lowe, J.A., Jones, P. and Kilsby, C.G. (2009). UK Climate Projections: Briefing report. Met Office Hadley Centre, Exeter, UK. Kharin, v.v., Zwiers, F.W. (2000) Changes in the extremes in an ensemble of transient climate simulations with a coupled atmosphere-ocean GCM. J. Climate 13: 3760-3788
Levitus, S. et al. (2001), Anthropogenic warming of the Earth’s climate system, Science 292: 267-270. McGregor, G.R., Ferro, C.A.T., Stephenson, D.B. (2005) Projected changes in extreme weather and climate events in Europe. In: Kirch W, Menne B, Bertollini R (eds) Extreme weather and climate events and public health responses vol. 13–23 Dresden. Springer, Berlin Heidelberg New York, p 303 Meehl GA, Zwiers F, Evans J, Knutson T, Mearns L, Whetton P (2000) Trends in extreme weather and climate events: issues related to modelling extremes in projections of future climate change. Bull Am Meteorol Soc 81: 427–436 Meehl, G.A., T.F. Stocker, W.D. Collins, P. Friedlingstein, A.T. Gaye, J.M. Gregory, A. Kitoh, R. Knutti, J.M. Murphy, A. Noda, S.C.B. Raper, I.G. Watterson, A.J. Weaver and Z.C. Zhao, (2007) Global Climate Projections. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Murphy, J.M., Sexton, D.M.H., Jenkins, G.J., Booth, B.B.B., Brown, C.C., Clark, R.T., Collins, M., Harris, G.R., Kendon, E.J., Betts, R.A., Brown, S.J., Humphrey, K.A., McCarthy, M.P., McDonald, R.E., Stephens, A., Wallace, C., Warren, R., Wilby, R., Wood, R.A. (2009), UK Climate Projections Science Report: Climate change projections. Met Office Hadley Centre, Exeter, UK. Nakicenovic, N. & Swart, R. (Eds.) (2000). Special Report on Emissions Scenarios. A Special Report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, NY, USA, 599pp
55 © Crown copyright 2011
R Development Core Team (2007). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org. Randall, D.A., R.A. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V. Kattsov, A. Pitman, J. Shukla, J. Srinivasan, R.J. Stouffer, A. Sumi and K.E. Taylor, (2007) Cilmate Models and Their Evaluation. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Rivington, M., Miller, D., Matthews, K.B., Russell, G., Bellocchi, G., Buchan, K. (2008) Evaluating regional climate model estimates against site-specific observed data in the UK. Climatic Change 88: 157-185. Rowell, D. (2006) A demonstration of the uncertainty in projections of UK climate change resulting from regional model formation. Climatic Change 79: 243-257 Semenov, M.A., (2011) Impacts of climate change on wheat in England and Wales. J.R.Soc. Interface 6: 343-350. Semenov, M.A. (2007) Development of high resolution UKCP01-based climate change scenarios in the UK. Agric. Forest Meteorology 144: 127-138 Schär, C., Vidale, P.L., Lüthi, D., Frei, C., Häberli, C., Liniger, M., Appenzeller, C. (2004) The role of increasing temperature variability in European summer heatwaves. Nature 427: 332–336 SolarTool (2008). Generating solar radiation at inclined planes. EMD International, Aalborg, Denmark. Stott, P., Stone, D.A., Allen, M.R., (2004) Human contribution to the European heatwave of 2003. Nature 432: 610-614 UKCP09 Steering Group, (2011) Responses to the UKCP09 Science Peer Review March 2011 http://ukclimateprojections.defra.gov.uk/content/view/1140/664/
56 © Crown copyright 2011
Weiss W, Hays CJ (2005). Calculating daily mean air temperatures by different methods: implications from a non-linear algorithm. Agricultural and Forest Meteorology 128: 57-65. Wu, T., Wang, Z., Liu, Y., Yu, R. and Wu, G. (2004) An evaluation of the effects of cloud parameterization in the R42L9 GCM. Adv. Atmospheric Sciences 21: 153-162. Zwiers, F.W., Kharin, V.V. (1998) Changes in the extremes of the climate simulated by CCC GCM2 under CO2 doubling. J Climate 11: 2200–2222 Zweirs, F. (2002), The 20-year forecast, Nature 416: 690-691.
57 © Crown copyright 2011
Appendices
Appendix 1: Consideration of the uncertainty due to emissions scenarios. 1a) 2030s Annual mean daily mean temperature – Low, medium and high emissions scenarios th (50 percentile only)
1b) 2080s Annual mean daily mean temperature – Low, medium and high emissions scenarios th (50 percentile only)
58 © Crown copyright 2011
Appendix 2: Maximum temperature for other seasons (not JJA):
th
th
th
2a) 2020s Winter (DJF) daily maximum temperature – 10 , 50 , 90 percentile.
th
th
th
2b) 2020s Spring(MAM) daily maximum temperature – 10 , 50 , 90 percentile.
59 © Crown copyright 2011
th
th
th
2c) 2020s Autumn (SON) daily maximum temperature – 10 , 50 , 90 percentile.
Appendix 3: Minimum temperature for other seasons (not DJF): th
th
th
3a) 2020s Spring (MAM) daily minimum temperature – 10 , 50 , 90 percentile.
60 © Crown copyright 2011
th
th
th
3b) 2020s Summer(JJA) daily minimum temperature – 10 , 50 , 90 percentile.
th
th
th
3c) 2020s Autumn (SON) daily minimum temperature – 10 , 50 , 90 percentile.
61 © Crown copyright 2011
Appendix 4: Change in mean daily maximum total downward surface shortwave flux for other seasons (not JJA)
th
th
th
4a) 2020s Winter (DJF) mean daily maximum flux – 10 , 50 , 90 percentile.
th
th
th
4b) 2020s Spring(MAM) mean daily maximum flux – 10 , 50 , 90 percentile.
62 © Crown copyright 2011
th
th
th
4c) 2020s Autumn (SON) mean daily maximum flux – 10 , 50 , 90 percentile.
63 © Crown copyright 2011
Met Office FitzRoy Road, Exeter Devon EX1 3PB United Kingdom
Tel (UK): 0870 900 0100 (Int) : +44 1392 885680 Fax (UK): 0870 900 5050 (Int) :+44 1392 885681
[email protected] www.metoffice.gov.uk
