XU 1

Siqi.Xu
Sustainability FRINQ
Professor Perkins
2/4/2014
The graphs are in the excel work sheet attached with this word document.
1. (a) What is the hottest year on record in each of the sites and how hot was it in each
city that year? (You wont be able to get the temperatures directly off the chart
(except for Boise City), but can use the chart to point you to which row of data you
should look at).

(b) What are the benefits and problems (pros and cons) of using running means?
Benefits:
Clear and Continuous.
If there is a missing data in a month or some months, it will cause the missing data of
the annual temperature. If I use the annual temperature data, the line of graph will be
discontinuous. Moreover, it is not good for comparing each years temperature. But
City Year
Highest
temperature
Measured
temperature
degrees beyond measured
temperature
Boise City
OK 2004 56.9192 54.1 2.8192
Liberal KS 1954 57.938 56.8 1.138
McMinnville
OR 2004 53.8664 51.4 2.4664

XU 2

if I use running means, this phenomenon can be avoided.

Avoid the effect of mistakes, reliable
When there is a data which is far away from the real temperature, I can use running
mean to decrease the effect of the data which is measured maybe by mistake. For
example: you get numbers 50, 52, 53, 10, 51 for a group of annual temperature.

Make the graph smooth and easy to catch trend of period
49.5
50
50.5
51
51.5
52
52.5
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
t
e
m
p
e
r
a
t
u
r
e

year
Comparision with use or not use running mean
not use running mean use running mean
0
10
20
30
40
50
60
1886 1888 1890 1892 1894 1896 1898
t
e
m
p
e
r
a
t
u
r
e

year
The compare with use running mean and not
use
not use running mean use running mean

XU 3

If you do not use the running mean, the line will be up and down. But if you use
running mean, the line will be smoother than the former one. And the trend will be
easier to find.

(I copy the image below from the website to prove the smooth and not smooth)

Reflect period as you want
Running mean focus on a period of trend instead of current figure. It depends on how
long you would like to choose. For example, I can choose 2 year running mean, 5
year running mean, 10 year running or even more. If I focus on a long period running
mean, the graph can reflect the trend of a given period more clearly and may reflect
future trend.

XU 4

Problems:
Need lots of data
When you use running mean, you should get a huge number of data to calculate the
average of a given period. If the data is not so much, the advantage of using running
mean will not be so obvious even no advantage.
Not in time
Using annual temperature as points will reflect the change for every years
temperature in time while using running mean may show that trend later.
Not accurate
If the figures have some complex relationship, the running mean is not very accurate
to reflect the complex relationship.
Use only for the trend very obvious
If the difference between every data is not large, the running mean cannot reflect the
trend of these figures. For example:
Temperature: 50,51,52,51,50,51,51,52,51,50
5 year running mean: 51, 51, 50.8, 51, 51, 51, 51, 51, 51, 51

XU 5


It seems no change in 5 year running mean data.

2. (a) Compare the 1920s and 1930s temperature anomalies in the three cities. Were
the 1930s anomalously warm compared to the 1920s in all three locations?
Yes. Because:





But for Boise City and Liberal, every years temperature anomaly in 1930s is greater than
1920s. Whereas for McMinnville, not every years temperature anomaly data is greater than
1920s.

49.5
50
50.5
51
51.5
52
52.5
1886 1888 1890 1892 1894 1896 1898
t
e
m
p
e
r
a
t
u
r
e

Year
annual temperature running mean
Average temperature anomaly
Boise City Liberal McMinnville
20s -0.51889 -1.81452 0.86736
30s 0.721311 0.10176 1.02216

XU 6

(b) Compare the temperature anomalies since the 1950s (i.e., from 1960 on) in the three
cities. Do you see the same trend in the three locations?
Approximately the same but not absolutely the same.
Generally, both them three have the trend of first go up then down then up. However, to
be more accurate.
There is also a little different for temperature anomaly trend for Boise City. It has a
down trend from 1952-1953 and an up trend from 1953-1954. It means that the trend
for Boise City is up, down, up down and up. It has additional trough than other two
city.
The peak and trough for three cities do not exist at the same time.

(c) If you had only the Liberal, KS temperature data available to you, would you
conclude that the Earth is experiencing a warming trend?
No, that is not accurate. To make a conclusion, we should do enough times of
experiments to make sure the sample size is big enough. Moreover, the sample should be
selected randomly to make sure every individual have the same chance to be chosen. The
population is the cities in the earth while the sample is cities in the earth. We only choose one
city as our sample. So we cannot say that if we only choose one city of the earth, the whole earth
is experiencing a warming trend.
3. (a) Plot the global temperature anomaly time series and compare it to the time series
for the individual stations. Compare the global time series with the series for Boise
City, OK. Are the trends the same?

XU 7

(I use two kinds of data (one for Boise City and another for the global) to make
two lines to compare them. The graph is graph f in excel.)
Approximately the same but not absolutely the same.
Generally, they share the same trend for some part.
However, to be more accurate, there are some differences. If we extrude the line
of Boise city from 1909-1930. The trend during that period maybe closer to the global
trend. From year 1976-1982, the trend between the two is absolutely different. Besides,
the extent that Boise City is greater than global but the troughs and peaks for global line
are more than Boise City line.
(b) A meaningful answer to the second part of the last question requires statistical analysis.
While we will not do significance testing, we can begin to analyze trends for each data set.
Begin by individually selecting the 1960 data set for Boise City in the plot you made for
part 2 above (just click on one) and then use Excel's Add Trendline tool to fit a linear trend
line to the data. Excel does this using a least squares method. The options will allow you to
add the equation and regression coefficient for each line to your chart. Do the same for the
entire Boise City time series and the global data.

(The graph is graph b.)
+1.What are the slopes of the lines for the global temperature anomaly and the Boise City,
OK temperature anomaly?
The slope for Boise City OK is 0.0197. (For the whole line)
The slope for global temperature is 0.0106.
+2.What do they represent?

XU 8

Generally, for an equation, the slope means the rate at which the y values are changing
for every unit change in the x values.
Right here, for the slope of Boise City, it means that every year (the number of year is
going up), Boise Citys temperature anomaly will be estimated to increase 0.0197 unit. While for
the slope of global line, it means that every year, global temperature anomaly will be estimated
to increase 0.0106 unit.
It reflects that the changing rate of Boise City is greater than the changing rate of global
line. It also means the temperature anomaly of Boise city is less steady than the temperature
anomaly of global.
+3.How does the slope change for the Boise City time series data depending on whether you
select all of it or only data from 1960 on?
When you select the whole line, the slope is 0.0197 while when you only select the data
from 1960 on, the slope is 0.0547.
+4.What does this suggest?
This indicates that temperature anomaly for the data from 1960 on is changing greater
than the whole line. It means that the temperature anomaly changing rate from 1960 on for Boise
City is greater than the changing rate from 1900s. The former is less steady than the latter.