Lab A – Instrumental Record of Earth’s Climate
Location of 8 globally distributed meteorological stations for computation
of global temperature change since 1840. Result of the computation is
shown below:
Decadal average temperature anomaly (relative to 20th century mean) for
8 globally distributed meteorological observation stations shown on the
above map plotted as a function of decade. Dots represent values. Note
the almost 2 degree C since 1840.
-2-
A1.1 Introduction
In this lab, we shall examine the history of temperature at the Earth’s surface
compiled by human observers using thermometers. The record of thermometer readings
compiled since the time thermometers were invented is referred to as the “instrumental
record” of climate change. Our look into this data serves several important purposes:
First, we will use the instrumental record to formulate a sense of what distinguishes
climate from weather. The earth’s weather is extremely variable, and this will be well
displayed in the instrumental record. Climate represents the average state of the
temperature (and other variables, but we consider temperature here) around which this
variability produces no net change within roughly a decade of time. Slow change in the
average state is what is referred to as climate change.
The second purpose of the lab activity is to understand how the instrumental record
plays into the scientific inquiry of climate change. Specifically, we shall investigate how
the instrumental record is used by the IPCC Working Group I in their AR5 report. In
brief, the instrumental record shows a persistent, globally widespread warming since
about a century ago that (a) seems to correlate with changes in atmospheric chemical
composition (notably greenhouse gasses), and (b) can only be explained by predictive
climate simulation models if these models account for changes in atmospheric
composition. To a lesser degree, we will pursue the notion that some of the climate
change signal that has been expected is “missing” from the instrumental record, and that
there may be an anomalous reduction in the warming trend over the last 5 years or so.
Two purposes:
•
•
Understand how climate signals differ from weather signals
See how the “instrumental record” is used in the IPCC AR5
In addition to having a “hands on” experience with “real” data—data that is exactly
the same as what climate scientists use—students who complete this lab will gain
familiarity with two aspects of modern scientific practices: First will be the fact that
knowledge about complex systems is often gathered through the analysis of extremely
large data sets with extraordinarily creative analysis measures that cope with the large
size. The instrumental record fills a 50 Megabyte text file, and without modern
computational data analysis practices, it would be impossible to use it effectively. The
second aspect of familiarity to be gained by students is the usage mode of modern
numerical analysis software. Whereas beakers and test tubes formed the primary
equipment set in the days when the Atmospheric composition was first discovered,
today’s analysis of how this composition impacts the climate requires modern digital
numerical methods. Students will be given an “open ended” introduction to what is
available on campus, including MatLab, Excel, Numbers and other software.
Develop awareness of:
•
•
•
How big datasets are used
Modern numerical analysis software
Use of graphical analysis to understand complex systems
-3-
Global lawbreaking.
A1.2 Thermometers and Weather Records
The instrument for measuring the heat content of matter (and air in particular) arose
from a device called a thermoscope invented in the early 1600’s for displaying relative
changes in temperature over a period of time at a particular place. The thermoscope
evolved into a thermometer once a definitive calibration scale was established by people
such as Christian Huygens (1665), Isaac Newton (1701), Daniel Fahrenheit (1724) and
Anders Celsius (1742). Thermometers register monotonic increases in their scale (known
as the temperature) as the heat content of the matter they measure is increased.
Anders Celsius, 1701 – 1744, father of thermometry.
We are familiar with how some thermometers work. The mercury thermometer relies
on the thermal expansion of a volume of liquid mercury into a vacuum column of a glass
-4tube. Some thermometers work using radiation emitted by the matter whose temperature
they measure. We shall spend a great deal of time considering these types of
thermometers because of the necessity of understanding black body radiation and its
relationship to temperature.
Regular recording of temperature measurements around the world as a means of
understanding weather phenomena began (along with efforts to measure other
parameters, such as atmospheric pressure using a barometer) in the mid 19th century
throughout the world. It began in the 1870’s in the U.S. as a result of the effort by the
then president Ulysses S. Grant to serve military purposes.
Today the instrumental record consists of data from over 6000 weather stations (in
addition, there are over 7500 precipitation stations and over 2000 barometric pressure
stations). Of these stations, more than 1650 have records greater than 100 years, ~200
have records greater than 150 years and the earliest measurements go back to 1697 A.D.
The longest continuous record is for the city of Berlin, which began recording
temperature measurements in 1701.
Of course, there are problems and questions regarding the quality of the data. One
must consider the fact that thermometers have become more accurate with time (today,
temperature is routinely measured to the nearest 1/100th of a degree Centigrade). Some
parts of the world have more meteorological stations than others (Antarctica, for example
has only a few). Nevertheless, the fact that thermometers are calibrated according to the
strictest rules created by the early scientists of the 17th and 18th centuries means that
temperature was measured according to a fixed standard from roughly the start of the era
known as the industrial revolution (put on display at the recent opening ceremonies of the
2012 London Olympiad).
Plaque in Stonehaven, Scotland, citing the establishment of a public
barometer in 1852 as a means of promoting greater safety of life at sea.
There are roughly 2000 stations contributing to an instrumental record
of atmospheric pressure since about the time this public barometer was
installed. I ran into this while on a recent trip and, short of having a
photograph of an old thermometer station, I decided to illustrate the lab
-5write-up with this. Yes, there was a barometer there, but also a
thermometer.
A1.3 Temperature Measurement Technique
Everyone who has used a thermometer to take the temperature of their body, food
they are cooking or to gage what to wear on a given day has a pretty good idea of how
temperature is measured. In the present day, temperature is measured on an hourly basis
by automated systems that then report daily averages and maximum and minimums. I, for
example, once installed automatic weather stations on icebergs and ice shelves in the
Antarctic. These stations had very robust computers (called data loggers that looked like
sardine cans, i.e., had no regular human interface) that woke up once every 20 minutes to
take the temperature of a thermometer (an electrical resistance thermometer called a
thermister) that was housed in a special shelter that would eliminate the localized
warming associated with direct sunlight. In the past, measurements were taken by a
variety of methods, including by observers who had the job of doggedly keeping to a
schedule of measurement and recording.
Setting up an automated weather station (AWS) in Antarctica. This
station, and others like it, transmits hourly temperature readings to the
University of Chicago and other universities for use in studying the
climate of Antarctica. This picture was taken at about 2:00 AM on an
iceberg called B15A.
The timing and methods of temperature measurements, i.e., whether once per 20
minutes or once per hour, and whether by human or computer, has changed significantly
over the duration of the instrumental record. As a result of this, it has been decided that
the most “atomistic” (reduced to its smallest component) form of temperature data that
can be studied in a meaningful way is related to the monthly mean. Considering daily
-6means or even hourly means is not possible with such a diverse data source. Conversely,
considering only yearly or decadal means gives up some of the richness of behavior seen
robustly and with confidence in the monthly mean data.
Our work shall thus focus on using monthly mean temperature as the medium with
which to explore the instrumental record of climate change. Another reason for using data
organized on a monthly schedule is that it is fully compiled and publically available as a
result of the U.S. National Climate Data Center (located in Asheville, North Carolina,
and founded in 1934). The data we shall use was compiled by the NCDC since a time
long before global warming was a issue of concern within the nation, so there is little
chance that it has been tampered with, or degraded for darker purposes.
A1.4 Glossary, Useful Information, and Interesting Trivia
Temperature: A thermodynamic state variable that is determined by the heat
content of matter. Temperature is a measure of how much energy is contained
within “internal” microscopic motions of atoms and molecules. Heat flows from
high temperatures to low temperatures in an effort to spread these microscopic
motions equally through all parts of the universe.
Entropy: A fairly mysterious term that describes how, when heat flows from hot
to cold, getting the heat to flow back again is improbable. In situations where
entropy increases, it is equivalent to saying (figuratively) “the horse is out of the
barn”… Entropy is a function of temperature.
Temperature (practical definition): a well-calibrated scale describing the
degree to which air feels hot or cold.
Temperature anomaly: Departure or difference between the actual recorded
temperature and some reference temperature. Usually the reference temperature
is a time-average of the temperature. For example, the monthly temperature
anomaly might be the temperature difference between the observed value for a
given month and the average of all similar months (e.g., the June temperature
anomaly is the difference between each June’s temperature and the mean of all
available June temperatures).
Seasonal cycle: In this lab, we shall define the seasonal cycle as the 12 average
monthly temperatures that most accurately describe the annual cycles of
temperature at a given place. We shall determine the seasonal cycle for many
stations by averaging monthly temperatures.
Instrumental record: This is the “compact language” used to signify historical
meteorological measurements of temperature. The instrumental record is
problematical, because it goes back only about a century and a half, and it is full
of difficulty due to missing data, urbanization effects and problems with
standardization of equipment and measurement technique. Look it up:
http://en.wikipedia.org/wiki/Instrumental_temperature_record
NaN: a computer term for “not a number.” Often missing data are replaced with
NaN’s in text, tables and arrays. Be careful to handle NaN’s appropriately when
crunching the data for this lab and others.
-7Mean: A statistical quantity created by adding a set of related numbers and then
dividing the result by the number of numbers that were added.
Timeseries: the standard set of temperature data consisting of paired numbers. In
our lab a timeseries will be N pairs of numbers, with the first element of the pair
being the year (often a decimal year to signify the month, increasing by
increments of 1/12) and the second element of the pair being the average monthly
temperature recorded by the station.
Array: (also known as a matrix or spreadsheet) a rectangular table of numbers.
Running Mean: an average of a time-series where the value of the running mean
is equal to a mean of data spanning a time interval surrounding the time in
question.
Urban Heat Island Effect: One of the challenges to using historical, instrumentrecorded climate data is the fact that many (if not most) of the stations where
observations are made are located in cities and urban areas. It is a well known
meteorological phenomena that cities tend to be "hotter" (warmer on a day-to-day
basis) than the surrounding countryside. This raises the unpleasant but serious
possibility that what is seen as "warming" in the instrumental record is really just
the increase of urbanization over time (i.e., warming of the local meteorological
conditions surrounding the observation station due to the increase in the amount
of concrete around the site). The data from individual cities that we use in our lab
activities have not been corrected for the urban heat island effect.
IPCC Working Group 1: “The Intergovernmental Panel on Climate Change
(IPCC) has been established by WMO and UNEP to assess scientific, technical
and socio-economic information relevant for the understanding of climate change,
its potential impacts and options for adaptation and mitigation. It is open to all
Members of the UN and of WMO. The IPCC Working Group I (WG I) assesses
the physical scientific aspects of the climate system and climate change.” -IPCC
See the website for WG1 here: http://www.climatechange2013.org/ and
https://www.ipcc-wg1.unibe.ch/
Qin Dahe: A fellow glaciologist and friend who co chairs the WG1
-8-
That’s Qin Dahe at base camp near Mt. Everest.
Smoothing filter: a process which removes wiggles (rapid ups and downs) from
a graph of a given data variable. In this lab we will use a “running mean” of
neighboring data values as a means of smoothing the data.
Histogram: A very specific kind of bar plot that is useful for displaying the
variability of temperature data. The x-axis (horizontal) of the plot is typically a
range of temperature values (or “temperature bins” as the language used to
describe histograms often uses), and the y-axis (vertical) represents the number of
instances in which the data actually records a measurement within each bin. A
histogram often looks like a bell-shaped curve with the most frequent occurrence,
or peak, located near the mean value of the data. The “mode” is the value of the
data that is most probable, i.e., defining the peak of the histogram. The mode and
the mean are not necessarily the same, but they can be.
Probability: The number of times a result occurs divided by the total number of
times results are defined. For example, the probability of a temperature anomaly
being warmer than 2 °C in a given temperature data set would be a count of all
occurrences of the temperature above 2 divided by the total number of elements
(measurements) in the data.
GUI: Graphical User Interface. A computer program that presents buttons and
text boxes to a user so that the user may run the computer program without having
to know the details of how the program was developed.
Spreadsheet Program: A software package that is useful for performing
arithmetic and mathematical analysis on large sets of data. The data in such a
program is organized into a tabular, or array, format called a spreadsheet. Excel
and Numbers are the Windows/Apple versions of the typical spreadsheet
program.
-9-
MatLab: A software package that allows programming and other forms of
mathematical and numerical analysis of data.
Python: A package similar to MatLab.
A1.5 Berlin’s Temperature Record
The place on Earth that has the longest instrumental record of temperature is the city
of Berlin. In particular the Dahlem neighborhood of the borough called SteglitzZehlendorf. The temperature record starts in 1701, has a few breaks, and becomes quite
reliable in the early 1800’s. It is probable that the data from the 1700’s have been
modified and re-copied (with possible errors), and this is thought to mean that only the
period beyond about 1850 is reliable enough to take as a faithful record of conditions.
Location map of Dahlem, in the outskirts of Berlin, Germany.
We shall devote the first activity of the LAB to analysis of the Berlin data. Following
the analysis of Berlin, you will be asked to continue your analysis by considering other
meteorological stations from around the world to get a picture of what the instrumental
record has to say about global warming.
- 10 -
Lab Activities: Part 1. Analysis of Berlin-Dahlem Climate.
Computers are like Old Testament Gods; lots of rules and no mercy.
Joseph Campbell
It [the computer] takes these very simple-minded instructions—‘Go fetch a
number, add it to this number, put the result there, perceive if it’s greater
than this other number’––but executes them at a rate of, let’s say,
1,000,000 per second. At 1,000,000 per second, the results appear to be
magic.
Steve Jobs, 1985
Step 1. Think and discuss.
The activities for this part of the lab involve quantitative analysis of temperature
observations (data). Before beginning the activity, it is important to reflect on what an
“analysis” means, what can/cannot be done with temperature data, and what significance
the activities may ultimately yield. Please discuss among fellow students the following
questions:
- What is the difference between weather and climate?
- Starting from weather data (daily or monthly temperatures), how would one investigate
whether the climate has been steady or not?
- Averages of long lists of numbers (data) will be taken in this lab. What is the meaning
of an average (or of a mean)?
- What is one strategy for dealing with missing data? Might there be more than one
strategy? If so, describe several.
- What kind of computer programming or numerical analysis experience do people in the
class generally have? Who has the least amount of experience? What advice might that
person be given at the outset?
- How do people learn the specifics about using a computer?
Step 2. Acquire the data and examine it.
Choose which data analysis software or strategy you plan to follow. You are welcome to
use any numerical analysis software you can find and operate, and you are even invited
(if you have the time for it) to use a hand calculator. Our recommendation is that you use
either MatLab or one of the two basic spreadsheet packages (Excel or Numbers).
Load the form of the Berlin-Dahlem data that most suits the methodology you plan to
use for its analysis. Load two data sets: one for “timeseries” and one for “table” data. In
the “timeseries” version of the data, there will be two long columns. The first is the year
(decimal year in increments of 1/12 of a year, representing “months”), and the second
column is the monthly average temperature observed over the increment of time centered
on the decimal year value appearing in the first column. To represent the data for a single
year, 12 consecutive rows of the second column must be accessed.
- 11 The monthly average temperature represents the statistical mean of all hourly temperature
observations made during the month in question. To compute a mean, you must sum all
the member temperature observations (member = belonging to the month in question),
and divide the sum by the number of members.
The “table” form of the data possesses 13 columns and many rows. Each row represents
one year. The first column of the row lists the year (AD or Common Era), and each of the
columns numbered 2 – 13 represent the value of the monthly average temperature from
January to December, respectively.
Where there is no data, either NaN’s will appear, or the file will have a blank cell. Do
you notice any missing data?
The units for time are years (in decimal years) and for temperature are degrees
Centigrade.
- In what way might a data table with 12 months of data organized as columns, with each
row being a separate year, be more suitable for determining information about the
seasonal cycle than the “timeseries” form of the same data?
- Describe what averages over a column vs. averages over a row in a data table might
provide?
Step 3. Set up your lab report (“notebook”).
Familiarize yourself with a word processing program that is capable of accepting
insertion of figures and other elements of non-typewritten material. Word and Pages are
the two most popular programs on campus, but there are others. (If you are really intense
about getting good formatting, try using LaTeX.)
Create a document that will become your personal lab notebook. Each student must have
their own individual notebook (this would be a word processing document that is later
printed to hand in on completion of the lab activities). Even if you collaborate with other
students, please create your own personal notebook (document). Wherever you insert text
that is not composed by yourself, please indicate that the text is a quotation and use an
appropriate method of citation (exact format of the citation is up to you to choose). Do
not copy materials written or otherwise composed by other students in the course.
In composing your lab notebook, please use complete sentences and clear labels. Messy
notebooks will be counted down when final grades are applied.
- Do you have a strategy for saving your notebook? (It is unlikely to reappear in future
lab sections if the lab-room computers are erased on a regular basis.)
- Do you know how to paste figures (or insert figure files) that display graphs into your
notebook? Do you know how to paste tables of data into your notebook?
Step 4. Devise a strategy for completing the analysis of Berlin-Dahlem temperature
data.
Look over the 4 stages of activity indicated below, reading the questions and thinking
about the complexity of the tasks at hand. Discuss a strategy for completing the analysis
with fellow lab section members and with the TAs who are in the lab section to assist
you. Select a computer software package (or several packages if you want to spread the
- 12 analysis out among more than one) to use as the computational engine for the analysis.
(Students are welcome to also perform the analysis “by hand” without a computer. This
may prove far more difficult and tedious than the effort to learn the necessary skills
required to perform the analysis on a computer.)
Questions you might want to consider:
- What differentiates one type of computer software package from another?
- What is the best way to make use of TA help and knowledge?
- How do you recover from computer freeze-ups? How do you ensure safety of your
work?
Step 5. Begin!
Computer code: Students will have to use numerical analysis software that will operate
either on the lab computers, the computers they have in their own possession, or on
USITE computers. The university provides free access to both commercial and open
access software: including MatLab (my preferred software), Python, Excel and Numbers.
There is also an open source program for climate data analysis provided at this link:
http://clearclimatecode.org/gistemp/
If you are using MatLab, please be prepared to kluge together programs that involve use
of the following suggested commands:
mean()
nanmean()
for n=1:245; a group of statements; end
variable_name = [1850:10:2010];
if ( ); a group of statements; end
address_in_table=find( );
figure(2)
clf
plot(x,y,’r-‘)
xlabel(‘my x-label’)
ylabel(‘my y-label’)
title(‘my title’)
hold on, more plot commands
It will be helpful to know how to look up commands in the MatLab documentation. There
are several ways to do this.
How to deal with stressful computer interactions: Some students will find using
computers to process data, and the data itself, to be stressful. Often this happens because
students have low expectations of their own performance, and are “expecting” themselves
- 13 to encounter embarrassing difficulty. To counterbalance this stress, it is useful to realize
that the act of learning a computer system is much like the act of learning language: it is a
bootstrap process that requires small, painfully acquired increments (first words) to rise
along the learning curve. If you find that you are frustrated or embarrassed, realize that
there are two elements in your favor: other people near you are experiencing similar
frustration, and your progress will improve with time and experience. The TA’s (graduate
student lab leaders) are there to help you, so feel free to ask questions or seek help.
There are 4 activities below and various questions to be answered within each. Activities
and answers should culminate with materials being inserted or composed within
your lab notebook (word processing document).
1. Create a plot of a single year’s data: select an arbitrary year and plot the 12 monthly
temperatures. Put informative labels on both axes and a title on the top. Place the figure
(the lab TA will show you how to save the figure as a jpeg file, and place it into your lab
report template) into your lab report, write a caption for it and answer the following
questions:
a. What are the maximum temperature, minimum temperature and what is the
temperature range of the data plotted for the single year?
b. What is the coldest month? What is the warmest month? When do the Earth’s
seasons experience the two solstices? Do the warmest and coldest months fall on
a solstice? If not, explain why. If you don’t know what a solstice is, look it up in
Wikipedia.
2. Compute the average seasonal temperature cycle for the entire data set, compute by
taking an appropriate average. You will find that this computation is done most
conveniently with the “table” form of the data, as all you have to do is compute the mean
of each of columns 2 – 13 (corresponding to January – December). (Be prepared to deal
with “NaN’s”.) Recreate the plot in problem 1 above, but superimpose the average
seasonal cycle (my term for the 12 averages just computed) and present a legend or labels
that indicate which plotting elements represent the average seasonal cycle and which the
temperature of the single year. Place this graph in your lab report with an appropriate,
informative caption. Answer the following questions:
a. By how much does the individual year differ from the average seasonal cycle
(give a range in degrees C).
b. Are the deviations between the actual month’s temperatures and the average
seasonal cycle patterned randomly or is there a systematic pattern? Speculate on
reasons for what you observe.
c. Is there evidence for the particular year in question (selected in 1.) to have had
a warmer/colder than average summer or winter?
3. Compute the decadal temperature anomaly as a function of decade. This is
accomplished in several steps:
Step 1. For the entire data set, subtract the appropriate part of the average seasonal
temperature cycle from each of the columns in the table form of the data (remember to
use columns 2 to 13, as column 1 is reserved for the year number).
- 14 Step 2. Create a new table that contains one number for each year. This number will be
the average of the 12 numbers derived from Step 1 above. The result of this step is called
the “annual average temperature anomaly.”
Step 3. For each decade within the data set, compute the average of the 10 years’ annual
average temperature anomaly. This is the “decadal average temperature anomaly.”
Step 4. Create a plot that graphs the decadal average temperature anomaly for each
decade as a function of decade. If possible, try to fit a line through this data (using some
kind of line fitting routine, or do so by eye, e.g., using Matlab’s ginput() command to
place two dots and connect them with a line). This graph should look something like the
graph on the first page of this lab introduction (see above). Place this graph in your lab
report with an appropriate, informative caption. Answer the following questions:
a. Is there a trend in the temperature anomaly by decade (what you just plotted)
for Berlin/Dahlem? If so, please describe it. Give quantitative information as well
as qualitative information.
b. Consider the decadal temperature anomaly data since 1950. Is there a warming
trend? If so, please quantify it with a single rate, e.g., degrees per decade. Give
your answer in complete sentences. How does this trend compare with any trends
you see in the data before 1950?
c. Referring back to the graph that was created in problem 2, how does the
temperature trend determined above compare with the natural variability of the
monthly temperature. In other words, is the temperature change over a decade
implied by the trend observed above larger or smaller than the typical variability
of an arbitrary month’s temperature when compared with the average seasonal
cycle?
d. Is there evidence of climate warming in the instrumental record of Berlin?
Please feel free to express an opinion, even though the analysis of the data may be
ambiguous.
4. Please read the appropriate parts of the IPCC AR5 WG1 report. Examine the
IPCC’s analysis of the historical instrumental record (it may be tough to find). The report
is available here: http://www.climatechange2013.org/
a. Describe the section of the report you think is most relevant for comparison
with the analysis of the Berlin temperature record you conducted here. Indicate
why you chose to consider this section (give your rational for identifying it as “of
interest” in light of the work you conducted in this lab so far). What is the
purpose of the section you chose in the eyes of its authors (the WG1 of the IPCC
AR5).
b. To what degree does Berlin- conform to the pattern of the global analysis
conducted by the IPCC? Be descriptive and detailed in your answer. Place images
or figures extracted from the AR5 WG1 report into your lab report to help
illustrate the points you are making.
- 15 -
Lab Activities: Part 2. Analysis of Global Instrumental Record.
Graphical User Interface (GUI): In the second part of the lab activities, students will be
invited (but don’t have to necessarily use) one of several graphical user interfaces
provided on the lab computers or available over the internet. Two interfaces were
specifically designed for this class: the MatLab routine called “data_gather”, and the
web-browser routine called GHCNM. There are also various data analysis routines that
have been specifically developed for exploring the instrumental record (these can be
found using google search and also by reference to the IPCC WG1 AR5 report).
Instrumental Record Data: We shall be using data compiled by the U.S. National
Climatic Data Center run by the National Oceanic and Atmospheric Administration
(NOAA). The website for the CDC is: http://www.ncdc.noaa.gov/ and the website where
the data is provided is here: http://www.ncdc.noaa.gov/ghcnm/
Location of instrument stations for computation of global temperature
change. Color of dots represents the number of years the station has been
recording data.
Step 1. Devise a strategy for completing the analysis of Global Instrumental Record.
Look over the 3 stages of activity indicated below, and consider how best to complete the
analysis and to document it within your personal lab notebook (word processing
document). Some questions to consider and discuss among fellow lab participants:
- What are the relative merits of the various tools available for dealing with the global
instrumental record? Is one tool more simple to use? What features of each tool do you
like/dislike?
- How has the analysis of the global instrumental record been used/misused in the press
and in public policy debates?
- 16 - What is the size and extent of the underlying data? What aspects of the analysis are
challenging?
- What kinds of derived data (e.g., output of the GUI) will be collected during the course
of the lab activity? How will this new derived data be used?
- What constitutes an effective “station” to use in the analysis? Should all stations be
used?
- What is the WMO? How many stations are contributing data to the global instrumental
record?
- What is the format of the data on the GHCNM website/dataserver? What is the format
of the data provided in the lab?
- Is the data from all stations equally applicable to the analysis? If not why? Should
stations with missing data be eliminated? If so, how much missing data is allowable?
- Why does the analysis reference 1850 as the start date of the instrumental records “say”
on climate change?
- What kinds of effects might the instrumental record have that could skew the outcome
of the analysis (e.g., increased urbanization).
- Is it possible to “game” the analysis to produce a result that is distinctly different from
the result a more “objective” analysis would produce?
- How does the IPCC use the instrumental record in its AR5 report? Is this use essential?
- What would be the state of debate about climate change if the instrumental record did
not exist?
- Decide on how best to collaborate with fellow students and how to use TA help
effectively.
Step 2. Set up your lab report (“notebook”).
Restart your word processing program and load the document (notebook) created for Part
1 of the lab activities. Set up the document for receiving the results of the analysis
performed in Part 2.
Step 3. Begin!
There are 3 activities below and various questions to be answered within each.
Criteria for selecting stations to be analyzed should meet the following requirements:
•
•
•
Use stations that have record lengths that are as long as possible (students
should settle amongst themselves what minimum record length, i.e., time span
or start year, is adequate for the analysis of global climate trends).
Use stations that have a minimum of missing data (again students will settle
the level of missing data that is considered unacceptable)
Seek stations that are either distributed globally (in the case of studies of wide
geographic extend) or that are distributed relatively uniformly within subregions of the globe of interest. The TA will provide help with developing
- 17 -
•
maps of station location. People can also use google earth to determine station
location.
Seek stations that are not impacted by events or processes that could cause
bias in the record. (Students should discuss this and offer criteria, if any, to be
followed.)
A list of suggested “special” geographic focus areas for special focus is provided as
follows:
1. Tropical stations
2. High-latitude (polar) stations
3. Northern Hemisphere stations
4. Southern Hemisphere stations
5. Ocean islands
6. Big cities
7. Continental USA only
8. Uninhabited (low human population) regions only
9. Developing countries
10. Spanish speaking countries
11. Areas of military conflict
All analysis should use at least 10 stations to contribute to any objective result of the
analysis.
1. Global Analysis: Conduct an analysis of between 10 and 50 stations in the data set
(anyone wanting to use all 6000 stations should consult the TA first) to determine what
climate trends are visible in the instrumental record. Choose the stations carefully and
attempt to address the time period since 1850, but be prepared to shorten the time period
to the 20th and 21st centuries if necessary. Try to distribute your stations over the globe
and over a wide variety of latitudes and continents. Answer the following questions with
as much detail as possible, including the insertion of graphs and captions associated with
graphs that you create in lab:
- Create a well formatted table that lists the names of the stations you use. In this table,
provide information about what the station attributes are (e.g., what continent, what
elevation, city or rural?, etc.). Avoid putting “all” metadata information in this table, just
put in information that interests you.
- Determine a global decadal temperature anomaly for the time period considered (this
could be starting in 1850 or in 1900). Create a graph of this anomaly and insert it (with
appropriate description) in your lab notebook.
- Determine a “climate trend” from the decadal temperature anomaly for the globe. (This
trend could indicate climate warming/cooling/or no change.) Illustrate this trend on the
graph you uploaded for the previous question, and describe the trend with a few
sentences in the response to this task.
- How does the “climate trend” you find compare with one or more fellow students in the
course? Please discuss why there are similarities or differences?
- 18 - Upload a well captioned map of your stations to the lab notebook.
- save two copies of your decadal temperature anomaly analysis, one in digital form and
one as a text table (this would be decade,temp_anomaly pairs of points) for use in part 3
of this lab.
2. Special Geographic Area Analysis: Conduct an analysis similar to the one conducted
above for detection of global trends, but select a sub-area of the earth to focus the
analysis. Answer the following questions with as much detail as possible, including the
insertion of graphs and captions associated with graphs that you create in lab:
- Describe the sub-set of data you chose to analyze. What attributes do the stations have
that restrict the analysis to a “special” area? Upload a map (and caption) to your lab
notebook indicating the special area and the stations used.
- Create a plot of the decadal temperature anomaly for your area. Upload this to your
report with a caption. Indicate any climate trends that are visible in this analysis.
- How do any climate trends visible in the above plot differ from the climate trends
identified in the global analysis above?
- Compare your analysis with that of one or more fellow students to determine how your
special regional analysis may differ from other regional analyses. Describe what you find.
3. Please read the appropriate parts of the IPCC AR5 WG1 report. Examine the
IPCC’s analysis of the historical instrumental record (it may be tough to find). The report
is available here: http://www.climatechange2013.org/
- Describe the section of the report you think is most relevant for comparison with the
global analysis you conducted here. This may be the same section that you used in part 1
of the lab activities. If so, consider revising your response in part 1 to include
considerations made more clear by the global analysis you just completed.
- To what degree does your analysis conform to the pattern of the global analysis
conducted by the IPCC? Be descriptive and detailed in your answer. Place images or
figures extracted from the AR5 WG1 report into your lab report to help illustrate the
points you are making.
- 19 -
Lab Activities: Part 3. Atmospheric CO2 History.
Ice Core CO2 Record: Since about 1960, glaciologists have been using ice-core drills to
penetrate the Greenland and Antarctic ice sheets for samples of very old ice. Some of
these samples contain air bubbles that were trapped within the ice (as it densified from
snow and became gradually buried). These air bubbles give a history of the atmospheric
gas composition going back for hundreds of thousands of years. Some specialized ice
cores (e.g., Law Dome in Antarctica) come from areas of high accumulation rates (high
snowfall rates), and are thus appropriate for giving a history of atmospheric CO2
concentration over the last 2 millennia. We shall be using this ice core derived data in the
lab activities associated with part 3.
Mauna Loa CO2 Record: Since about 1958, scientists (most notably David Keeling)
have measured the atmospheric concentration of CO2 atop one of the highest volcanoes
in the Pacific Ocean (a dormant volcano) where there is little local influence on the
atmospheric composition. This record is thought to be a faithful representation of the
atmospheric CO2 concentration for the entire globe. We will also use this data in the lab
activities for part 3. The Mauna Loa data is available here (and on the chalk site):
http://www.esrl.noaa.gov/gmd/ccgg/trends/
Units of data: decimal years for time and parts per million (molar) for CO2 concentration.
Step 3. Begin!
There are 4 activities below and various questions to be answered within each.
1. Recent CO2 trends: Plot the atmospheric record of CO2 concentration for the last 2
millennia in three graphs: (a) one that shows the entire data set, (b) one that shows only
the period prior to 1958, and (c) one that shows only the period after 1958. Upload (with
captions) these three figures to your lab notebook (word processing document).
- What are the primary differences between CO2 data determined by ice core and data
determined by measurement at Mauna Loa? What is the primary cause of the increased
variability seen in the Mauna Loa data?
- By how much has CO2 changed over the last two millennia? What time period has this
change been most pronounced? Remember to give units. Also express the change as a
percentage of the initial value. Give complete sentences.
- Is the trend in CO2 changing (i.e., accelerating) since 1958?
2. Recent CO2 trends by decade: Compute the decadal average CO2 for each of the
decades since 1850. Save this data for use in 3 below.
3. Climate sensitivity: For each decade of 2 above, determine the global decadal
temperature anomaly associated with each decade by referring back to part 2 of this lab.
If necessary consult the table you constructed in part 2 of the lab. Create a graph that
shows “decadal average CO2” on the x-axis (horizontal axis) and “decadal temperature
anomaly” on the y-axis (vertical axis) by plotting the derived anomaly and CO2 for each
decade (note that time will no longer be part of this graph). Attempt (if possible) to fit a
- 20 line to the data points, both for the entire data set and for the decades since, and inclusive
of, 1960. Upload (with caption) this figure to your lab notebook (word processing
document).
- Based on the graph, determine the sensitivity of the climate to CO2? The sensitivity is
defined as the decadal temperature change per unit of change in the CO2 concentration.
- Based on the graph, predict how much the global temperature will change as a result of
CO2 “doubling” (assume doubling will mean CO2 goes from its present value to twice its
present value). When will the CO2 double if current trends in CO2 continue? Compose
your answers in complete sentences and frame them in a paragraph that explains your
conclusion.
3. Please read the appropriate parts of the IPCC AR5 WG1 report. Examine the
IPCC’s analysis of climate sensitivity to atmospheric CO2 concentration. The report is
available here: http://www.climatechange2013.org/
- Describe the section of the report you think is most relevant for comparison with the
CO2 analysis you conducted here.
- To what degree does the sensitivity you determined here agree with the analysis
conducted by the IPCC? Be descriptive and detailed in your answer. Place images or
figures extracted from the AR5 WG1 report into your lab report to help illustrate the
points you are making.
- 21 -
Extra for the interested: a few situations where weather has
influenced human history:
Year without a summer: The year 1816 is commonly referred to as the year
without a summer because, shortly before, the eruption of a volcano, Mt. Tambora
in Indonesia, depressed summer temperatures in the Northern Hemisphere for a
period of time. The effect of volcanoes on global climate is well known. The
eruptions of Mt. Pinatubo in 1991 and El Chichón in 1982 both caused noticeable
cooling (on the order of a degree C) for about the following year.
Napoleon’s retreat from Moscow: After capturing Moscow in the autumn of
1812, Napoleon’s army was forced to retreat back to France under extremely
difficult circumstances in the winter of 1812/13.
Night Bivouac of the Grand Army during retreat from Russia, 1812.
Graph depicting the advance (tan) and retreat (black) of the Grand
Army in 1812. Temperature plotted on the lower panel.
- 22 -
Operation Barbarossa: Hitler and the Nazi regime of Germany ordered the
German army to attack the Soviet Union on June 22, 1941. The initial battle plan
was code named Barbarossa. This attack bogged down in the winter cold.
According to Wikipedia (see site): On 27 November 1941, General Eduard
Wagner, the Quartermaster General of the German Army, reported that "We are
at the end of our resources in both personnel and material. We are about to be
confronted with the dangers of deep winter." Students should be able to see the
unusually cold winter temperatures of 1941 and 1942 in the data provided.
NAO index: A measure of the NAO, or North Atlantic Oscillation. The NAO is
a “see-saw” atmospheric pressure pattern which alternately intensifies and
weakens the low-pressure cell that typically sits over Iceland. When the NAO is
strong, storms in the North Atlantic are intensified, and it is more difficult for
westward sailing across the North Atlantic. This see-saw is thought to be
responsible for certain weather and "local climate" anomalies in western and
central Europe, such as the period of cold during the early years of WWII.
The Siege of Leningrad: Among the many terrible battles of the German
invasion of the Soviet Union was the siege of Leningrad, where there was much
suffering within the civilian population. The siege began in September 1941 and
did not end until January 1944. The cold winter weather was both an ordeal and a
help to the Soviet people. A road was built across the lake ice on Lake Ladoga,
which bordered the city, and this allowed the population to be resupplied during
the winter (the road was called "Doroga Zhizni" or the road of life). It is estimated
that 800,000 of 3,000,000 inhabitants of the city died during this terrible battle.
Leningrad is now (and was previously) known as St. Petersburg. One of the odd
things about the Doroga Zhizni is that many trucks and heavy vehicles broke
through the ice in a manner that inspired a lot of research into the weight-bearing
capacity of a floating plate of ice. I use this research to help understand how
Antarctica’s ice shelves respond to the weight of new meltwater lakes on their
surfaces, and how these loads can contribute to ice-shelf disintegration.
- 23 -
Depiction of the siege of Leningrad from a diorama in the Moscow
museum of WW2.
Dimitri Shostakovitch: His 7th symphony was dedicated to the city of
Leningrad, and reflects some of the suffering that occurred as a result of the
winter siege. Pick up a copy of the Chicago Symphony Orchestra conducted by
Leonard Bernstein playing this symphony at the iTunes store (here).
The Russian composer, Dmitry Shostakovich, served throughout the
siege of Leningrad as a firemen. He composed and performed his 7th
Symphony during the siege.
The Chicago heat wave of 1995: During the period of 12 - 17 July, 1995, the city
of Chicago suffered a terrible heat wave that led to the deaths of over 600 people.
This event is considered to be one of the worst weather-related events in US
history and ranks alongside the several terrible hurricane disasters that have also
caused great damage and loss of life. The African American population of the city
suffered most grievously because of the vulnerabilities associated with
impoverished conditions.
The snowstorm election: Michael Bilandic served as the mayor of Chicago from
1976-1979. In 1979 he failed to win re-election and was replaced by Mayor Jane
Byrne, serving from 1979 to 1983. Snow removal was one of the main issues in
the 1979 election campaign, and Bilandic was widely criticized for not salting and
plowing the city's streets well enough during his term.
- 24 -
Former Mayor Michael Bilandic.
Current Mayor of Chicago, Rahm Emanuel
- 25 -
Appendix I. Useful things to know about MATLAB
Overview:
Among the most challenging things to learn is a new computer programming language.
Many of you will be learning how to program a computer for the first time. Many of you
will be “old hands” at performing this type of work. We emphasize the use of computers
as computational instruments in this course, because the alternative, i.e., processing large
amounts of data by paper, pencil and adding machine, is horrifyingly tedious and unlikely
to be even possible. A subtle viewpoint will be created by the experience you will receive
in learning MATLAB (and/or one of the spreadsheet programs) is that of computers as
“computing machines.” Indeed, the original invention of computers was literally driven
by the need to conduct analysis of data and numbers for the purpose of science. Today,
this purpose is more or less submerged underneath all of the spin-off purposes for
computers: texting messages, surfing for inappropriate website imagery, watching
movies, playing games, reading assigned papers, etc.
In the lab, your TA will introduce the tools you will need to perform MATLAB
computations. The TA will not perform the analysis, or compose the logical progression
of computations, he/she will only provide the tools. Some of you will find this to be
extremely uncomfortable. Please do not be alarmed by this discomfort, as it will subside
quickly as you gain proficiency in performing quantitative analysis with the use of a
computer.
Some things you will need to know about MATLAB:
•
When you launch MATLAB on your computer, you will be given a “workspace”
that consists of a collection of variables. These variables will be scalars, vectors
and tensors (vectors of higher order, also known as arrays or matrices). If you go
on in mathematics, you might make use of the fact that these variables can be real
or complex. The variables are numbers (or collections of numbers in the case of
vectors and matrices) that have limited precision. MATLAB works in 64-bit
arithmetic, so it means that there is no such thing as a perfect numerical answer,
only a very precise numerical answer, to the final step of your computation.
•
MATLAB provides a “user’s work screen” consisting of a pallet of windows and
sub-windows. This pallet provides a “command window” (where you type
commands a la carte and where you receive printed output from the computer), a
view of your workspace (this view can be enquired to actually see the arrays of
numbers), an editor window where you can compose scripts or programs, and a
few less useful windows (that can be clicked away), such as the “command
history” window.
•
MATLAB provides exquisite documentation to all of its functions. This
documentation is accessible by hitting the ‘?’ mark on the menu bar. When
struggling, when suffering, when in doubt, the help window provides the most
reliable path to salvation.
- 26 Useful commands for use during this week’s and future labs. These commands are
presented in random order, but if you can figure out how to use them in a proper order,
you can manage to complete the lab assignments for this week.
output = mean(input)
for n=1:10
%(a list of commands goes here)
end
% -- this is the symbol for creating a non-executable comment in a script
% Here is a typical sequence of commands designed to create a figure or graph
figure (2)
clf
plot(xdata,ydata,’ko’,’markersize’,15)
hold on
plot(otherdata_x,otherdata_y,’b.’,’markersize’,20)
title(‘title text’,’fontsize’,14)
xlabel(‘text to go on horizontal axis’)
ylabel(‘text to go on vertical axis’)
xlim([lower upper])
ylim([lower upper])
set(gca,’fontsize’,14)
disp(‘write something that you want displayed on the screen’)
legend(‘first plot symbol’, ‘second plot symbol’)
distance=zeros(length(deaths),length(pumps));
Npumps=length(pumps);
Ndeaths=length(deaths);
for n=1:Npumps
distance(:,n)= sqrt( (deaths(:,1)-pumps(n,1)).^2 + (deaths(:,2)-pumps(n,2)).^2 );
end
bins=[0:.25:15];
[count,index]=histc(distance(:,7),bins);
bar(bins,count,'BarWidth',1)
- 27 Specifics:
One of the benefits of the activities of the lab will be to introduce you to modern
computational software useful for numerical analysis. MatLab is one such software
package. Here are some tips that you may find useful.
How to get the data: The data can be loaded into MatLab in many
different ways (probably about 10). However, the easiest way is to load
MatLab “native” data format files which end with *.mat . These files (for
timeseries data and table formatted data) are available on the Chalk
website.
Commands that will be useful:
Only about 10 or 15 commands in MatLab’s language are needed to build
powerful data analysis programs that will do the task of “crunching” the
Berlin-Dahlem data. Students are expected to be familiar with these
commands, as they are practically universal in all computer programming
languages.
For clearing data from the workspace and for clearing the screen:
clear all
clc
For suppressing “printing” to the screen:
… end all statements with a semicolon “;”
How to load a data file:
load filename.mat
Cells: if you want to create bits of code that execute separately within a
larger script, you can use the cell structure in the script. Cells are separated
by two percent signs, e.g., here is a program that will plot a graph of a
long series of x,y pairs loaded into an array called “d”… d(:,1) represents
all rows of the first column of d and d(:,2) represents all rows of the
second column of d. This is what the “:” signifies, “all”. d(:,1) represents
the x value and d(:,2) represents the y value. Also shown, are commands
to set the font size, set the figure window to “1”, clear the previous figure
(if it exists in the window), and print titles and axis labels. Finally, there is
a command to “hold” the figure if desired.
%% add a comment <-this command is the cell break,
figure(1)
clf
set(gca,'fontsize',14)
plot(d(:,1), d(:,2), 'k-')
title(data_name)
xlabel('year')
- 28 ylabel('C')
hold on
%% next cell starts here <- another cell break
To compute a seasonal cycle, you might want to use a statement like this:
T_cycle=nanmean(table(:,2:13));
Note that nanmean( ) represents the mean of a data series where all nan’s
have been ignored. Also note that “table” would be the data you load in as
table formatted .mat file . Chances are you would have to use a data array
name that is different from “table”. But this can be changed by specifying
a new name:
Table = old_data_name;
Here’s an easy way to create a running mean (in this case, a 10 year, or 10
x 12 month running mean):
T_smooth=smooth(T_anomaly,12*10);
Here’s a nice way to superimpose various graphical entities on top of each
other:
hold on
plot(d(:,1), T_smooth, 'r-','linewidth',2)
plot([d(1,1) d(end,1)],zeros(1,2),'b-')
Here’s a way to compute decadal means:
decade = 1850;
decadal_mean=T( time(:,1)>=decade & time(:,1)<(decade+10) );
Here’s another way:
%% compute decadal means of the temperature anomaly
decadal_mean = zeros(fix(length(d(:,1))/120),2);
for n=1:fix(length(d(:,1))/120)
decade=2000-10*n;
decadal_mean(n,2)=nanmean(T(d(:,1)>=decade &
d(:,1)<(decade+10)));
decadal_mean(n,1)=decade;
end
- 29 -
Note that the for/end statement configuration represents one of the primary
reasons why computers are useful in computation: they can iterate.
Here’s how to make a bar graph:
figure(2)
clf
set(gca,'fontsize',18)
bar(bins,c,'BarWidth',1)
xlim([-15 5])
xlabel('Annual Temperature Anomaly (C)')
ylabel('Cumulative probability')
title(['Probability analysis for ' data_name])
hold on
bar(bins(bin_number:end),c(bin_number:end),'BarWidth',1,'FaceColor'
,'r','EdgeColor','k')
legend('anomaly < 1 C','anomaly > 1 C')
plot(zeros(2,1),[0 1]','k-')
Appendix II. Useful things to know about spreadsheet programs
Spreadsheet programs can also be used to process the data in this lab activity.
Generally speaking, spreadsheets are more “intuitive” than what is presented in MatLab
or Python, however, they are also more cumbersome and require more human interaction.
With a spreadsheet, you can by hand compute decadal means from table data by
highlighting various segments of the table and issuing formulas in separate cells that
address the content of these segments.
There are some wonderful tutorials on how to use spreadsheet programs such as
Excel and Numbers (Apple’s version of Excel), and these can be found using google, or
within some of the links provided below.
For example, here is a tutorial on how to create bar graphs and histograms in Excel:
http://www.ncsu.edu/labwrite/res/gt/gt-bar-home.html