CO2, emissions, and global warming 3. the evidence Duke University

The Mauna Loa data provide an opportunity to gain experience applying basic exploratory data analysis.

Timeline for vignette

• day 1: Course overview, form working groups, install Rstudio, introduce R vignette

• day 2: Discussion of science and media issues on greenhouse warming; Gain experience with R using Mauna Loa C02 data

• days 3-4: in groups Introduce basics for computing in R; explore data on CO2 and climate

• day 5: Discussion of CO2 data

• day 6: in groups on vignette containing examples with NC data; can consider how local climate differs from global; how microclimate differs from ‘climate’; extreme events: drought, flood, fire, intensifying hurricanes

• day 7: in groups draft answers to questions on R vignettes

• day 8: debate litigation

Activities today

1. Introduce basics for computing in R
2. Explore data on CO2 and climate

• Group analysis of Mauna Loa data, including questions posted to Sakai

For next class

• Field trip to collect forest inventory and ebird data.

Based on your first day with the Mauna Loa vignette, provide a short summary for the following:

1. What time of year is CO2 highest/lowest and why?
2. At the current rate of CO2 increase in the atmosphere, how many years of change in the mean is equal to the seasonal change in CO2.
3. The Mauna Loa CO2 series is an example of passive surveillance data that became one of the important debates of modern times. List three ways that this data set bears on the litigation issues we discussed on big oil.
4. The ice-core data tell us that CO2 has been fluctuating for a long time. How does that perspective inform the interpretation of contemporary increases? Specically, does it bear on the question of responsibility for the societal cost of climate change?

Post your answers to Sakai.

CO2 at Mauna Loa

Here are a few thoughts as you work with R.

• what type of an object is mauna_loa_weekly)

• there is a problem with Halverson’s interpretation of the uncertainty in the trend. I illustrate with a simple experiment, let’s decrease the data to 10 observations.

plot(
  mauna_loa_weekly$date,
  mauna_loa_weekly$co2ppm,
  type = 'l',
  xlab = 'Date',
  ylab = 'CO2 Concentration PPM',
  main = 'Mauna Loa Weekly Carbon Dioxide Concentration'
)
abline(trend, col = 'dark blue')

# summary of linear fit
s1  <- summary(trend)$coefficients 

# redo with only 10 observations
keep <- sample(nrow(mauna_loa_weekly), 10)

y <- mauna_loa_weekly$co2ppm[keep] 
x <- mauna_loa_weekly$date[keep]

fit2 <- lm( y ~ x)

points(x, y, cex=2, col=2)
abline(fit2, col = 'brown', lty=2)

s2 <- summary(fit2)$coefficients

print(s1)

##                           Estimate   Std. Error   t value Pr(>|t|)
## (Intercept)           3.180373e+02 1.500848e-01 2119.0511        0
## mauna_loa_weekly$date 4.897732e-03 1.329748e-05  368.3204        0

print(s2)

##                 Estimate  Std. Error   t value     Pr(>|t|)
## (Intercept) 3.195451e+02 1.539115832 207.61599 3.242212e-16
## x           4.667591e-03 0.000151861  30.73593 1.364148e-09

• What is the difference in these two fits, and how should it affect our uncertainty about rising CO2? How does this relate to the differences in regressions using monthly and annual data?

• How does he use the diff function?

• Why might sea-surface temperatures NOT look like atmospheric CO2 concentrations?

• What are the wiggles in the composite ice-core record?