ENVR E-210 : Assignment 1a - Handling sustainability data
Critical Analysis of Environmental Systems (Harvard Extension School)
David Fong
14th September 2019
Greenhouse gas emission data originally sourced from “CAIT Climate Data Explorer. 2015. Washington, DC: World Resources Institute”. Available online at: http://cait.wri.org
Data for Land-Use and Forestry indicator are provided by the Food and Agriculture Organization of the United Nations (FAO). WRI has been granted a non-exclusive, non-transferrable right to publish these data. Therefore, if users wish to republish this dataset in whole or in part, they should contact FAO directly at copyright@fao.org
Any use of the Land-Use Change and Forestry or Agriculture indicator should be cited as FAO 2014, FAOSTAT Emissions Database. Any use of CO2 emissions from fuel combustion data should be cited as CO2 Emissions from Fuel Combustion, ©OECD/IEA, 2014.
Q1a Total global Greenhouse Gas emissions (including land-use change and forestry)
Years 2000 to 2012
\(MtCO_2e\)
data <- read.delim("Emissions.tdf")
answer1 <- data %>%
group_by(Year) %>%
summarize(sum(TotalIncluding, na.rm = TRUE))
# removes 'NA' (empty) values from the sums
answer1## # A tibble: 13 x 2
## Year `sum(TotalIncluding, na.rm = TRUE)`
## <int> <dbl>
## 1 2000 35444.
## 2 2001 36148.
## 3 2002 36855.
## 4 2003 38158.
## 5 2004 39452.
## 6 2005 40560.
## 7 2006 41344.
## 8 2007 42432.
## 9 2008 42893.
## 10 2009 42578.
## 11 2010 44237.
## 12 2011 45379.
## 13 2012 46049.answer1 %>% filter(Year %in% c(2000, 2012))## # A tibble: 2 x 2
## Year `sum(TotalIncluding, na.rm = TRUE)`
## <int> <dbl>
## 1 2000 35444.
## 2 2012 46049.Q1b Greenhouse Gas Emissions including land-use change and forestry
United States and Brazil from years 2000 to 2012.
\(MtCO_2e\)
data %>%
filter(Country %in% c("United States", "Brazil")) %>%
filter(Year %in% 2000:2012) %>% # all years from 2000 to 2012 inclusive
group_by(Country) %>%
summarize(sum(TotalIncluding, na.rm = TRUE))## # A tibble: 2 x 2
## Country `sum(TotalIncluding, na.rm = TRUE)`
## <fct> <dbl>
## 1 Brazil 23794.
## 2 United States 81559.Q2a Greenhouse gas emission from land-use and forestry
LandUseForestry : Total CHG Emission for Land-Use Change and Forestry only (MtCO2e)
PercentageLandUse : Percentage of total Emissions Attributed to Land-Use Change and Forestry only
data <- data %>%
mutate(LandUseForestry = TotalIncluding - TotalExcluding,
PercentageLandUse = LandUseForestry/TotalIncluding * 100)
datatable(data) %>% formatRound(columns=c("TotalExcluding", "TotalIncluding",
"LandUseForestry", "PercentageLandUse"),
digits=2)Q2b What is the signficance of a negative number in Total GHG emissions for land-use change and forestry only?
In some countries and some years, land-use change and forestry results in net negative greenhouse gas emissions, i.e. net absorption of greenhouse gas from the atmosphere.
Q3 Brazil, China, European Union, United States greenhouse gas emissions
data3 <- read.delim("Emissions1990_2012.tdf")
data3 <- data3 %>%
filter(Country %in% c("Brazil", "China", "European Union (15)", "United States")) %>%
# only from these four country (groups)
mutate(Year = ISOdate(Year, 1, 1)) %>% # change to a formal 'date' format
filter(Year >= ISOdate(2000, 1, 1) & Year <= ISOdate(2012, 1, 1)) %>%
# only from years 2000 to 2012
select(Country, Year, TotalGHGIncludingLandUse) # only need some data fields/columns
datatable(data3 %>% mutate(Year = year(Year)))Plot of Total Greenhouse Gas Emissions (2010-2012)
ggplot(data3, aes(Year, TotalGHGIncludingLandUse)) +
geom_line(aes(color = Country)) +
labs(title = "Total Greenhouse Gas Emissions (2000-2012)",
subtitle = "including land-use change and forestry",
x = "Year", y = expression("Emissions ("~MtCO[2]~e~")"))
Greenhouse gas emissions relative to 2000 levels
data3 <- data3 %>%
group_by(Country) %>%
mutate(Emissions2000 = TotalGHGIncludingLandUse[Year == ISOdate(2000, 1, 1)]) %>%
# Emissions2000 : find the 'year 2000' emissions for each Country
ungroup() %>%
mutate(AbsoluteChange2000 = TotalGHGIncludingLandUse - Emissions2000,
PercentRelativeChange2000 = AbsoluteChange2000/Emissions2000*100)ggplot(data3, aes(Year, AbsoluteChange2000)) +
geom_line(aes(color = Country)) +
labs(title = "Total Greenhouse Gas Emissions (2000-2012)",
subtitle = "including land-use change and forestry, absolute change relative to 2000 levels",
x = "Year", y = expression("Emissions relative to 2000 levels ("~MtCO[2]~e~")"))
ggplot(data3, aes(Year, PercentRelativeChange2000)) +
geom_line(aes(color = Country)) +
labs(title = "Total Greenhouse Gas Emissions (2000-2012)",
subtitle = "including land-use change and forestry, percent change relative to 2000 levels",
x = "Year", y = expression("Emissions relative to 2000 levels (%"~MtCO[2]~e~")")) 
Hypothesis 3A The rate of growth in emissions from 2000-2012 for these four groups has been highest for China
The chart shown above supports the hypothesis that the rate of growth in greenhouse gas emission (including land-use change and forestry) from 2000 to 2012 was highest in China (compared to Brazil, ‘European Union (15)’ and the United States).
Hypothesis 3B The trend in the US and Europe is for increasing emissions but at a lower rate than China over the period from 2000-2012
The chart shown above does NOT support the hypothesis that the 2000-2012 trend in the US or Europe was for increasing emissions of greenhouse gas emissions. The chart suggests that in both the US and Europe, total greenhosue gas emissions did not increase, and may have even decreased slightly.
Linear regression model
data3 <- data3 %>% mutate(YearsAfter2000 = year(Year)-2000) # Set Year 2000 to 'zero' year
model3 <- lm(PercentRelativeChange2000 ~ Country + YearsAfter2000 + Country * YearsAfter2000, data3)
tab_model(model3)| PercentRelativeChange2000 | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 9.31 | 4.27 – 14.35 | 0.001 |
| China | -16.32 | -23.44 – -9.19 | <0.001 |
| European Union (15) | -5.81 | -12.94 – 1.31 | 0.117 |
| United States | -8.38 | -15.50 – -1.26 | 0.026 |
| YearsAfter2000 | -1.07 | -1.78 – -0.36 | 0.005 |
| CountryChina:YearsAfter2000 | 13.80 | 12.79 – 14.80 | <0.001 |
| CountryEuropean Union (15):YearsAfter2000 | 0.03 | -0.98 – 1.04 | 0.950 |
| CountryUnited States:YearsAfter2000 | 0.37 | -0.64 – 1.38 | 0.476 |
| Observations | 52 | ||
| R2 / R2 adjusted | 0.987 / 0.985 | ||
Q4 Methane gas emissions
data4 = read.delim("Emissions_byGas.tdf") %>%
filter(Country %in% c("China", "India", "United States")) %>%
mutate(Year = ISOdate(Year, 1, 1)) # change to formal date format
ggplot(data4, aes(Year, TotalCH4)) +
geom_line(aes(color = Country)) +
labs(title = "Total CH4 (methane) emissions (1990-2012)",
subtitle = "including land-use change and forestry",
x = "Year", y = expression("Methane emissions ("~MtCO[2]~e~")"))
Methane emissions relative to 1990 levels
data4 <- data4 %>%
group_by(Country) %>%
mutate(CH4Emissions1990 = TotalCH4[Year == ISOdate(1990, 1, 1)]) %>%
## 1990 methane emissions for each country
ungroup() %>%
mutate(AbsoluteChangeCH41990 = TotalCH4 - CH4Emissions1990,
PercentRelativeChangeCH41990 = AbsoluteChangeCH41990/CH4Emissions1990*100)ggplot(data4, aes(Year, AbsoluteChangeCH41990)) +
geom_line(aes(color = Country)) +
labs(title = "Total CH4 (methane) Gas Emissions (1990-2012)",
subtitle = "including land-use change and forestry, relative to 1990 levels",
x = "Year", y = expression("Methane relative to 1990 levels ("~MtCO[2]~e~")"))
ggplot(data4, aes(Year, PercentRelativeChangeCH41990)) +
geom_line(aes(color = Country)) +
labs(title = "Total CH4 (methane) Gas Emissions (1990-2012)",
subtitle = "including land-use change and forestry, relative to 1990 levels",
x = "Year", y = expression("Methane relative to 1990 levels (%"~MtCO[2]~e~")"))
Hypothesis : increase in methane production has been increasing more rapidly in the US than in China or India
The chart above does NOT support the hypothesis that methane production has been increasing more rapidly in the US than in China or India. During the 1990-2012 time period, both India and China appear to increase their total methane emissions. By contrast, the United States methane emissions stays relatively constant during the 1990-2012 period.
What is an alternative hypothesis for this different pattern in the data?
Although natural gas and petroleum production and enteric fermentation (livestock) are the source of the majority of total U.S. methane emissions, and it is given that these sources of methane have been increasing, reductions in other sources of methane emissions (e.g. landfills and coal mining) balanced increases in natural gas/livestock methane production over the 1990-2012 time period.
Linear regression model
data4_norm <- data4 %>% mutate(YearsAfter1990 = year(Year) - 1990) # Set Year 1990 to 'zero' year
model4 <- lm(PercentRelativeChangeCH41990 ~ Country + YearsAfter1990 + Country * YearsAfter1990, data = data4_norm)
# include interaction effects
tab_model(model4)| PercentRelativeChangeCH41990 | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | -4.89 | -7.54 – -2.25 | 0.001 |
| India | 2.76 | -0.98 – 6.50 | 0.153 |
| United States | 2.87 | -0.87 – 6.60 | 0.137 |
| YearsAfter1990 | 1.89 | 1.68 – 2.09 | <0.001 |
| CountryIndia:YearsAfter1990 | -0.34 | -0.63 – -0.05 | 0.027 |
| CountryUnited States:YearsAfter1990 | -2.06 | -2.35 – -1.77 | <0.001 |
| Observations | 69 | ||
| R2 / R2 adjusted | 0.944 / 0.939 | ||
plot_model(model4, type = "pred", terms = c("YearsAfter1990", "Country"))
Regression model with data points
ggplot(data4, aes(Year, PercentRelativeChangeCH41990, color = Country)) +
geom_point() +
geom_smooth(method = "lm") +
labs(title = "Total CH4 (methane) Gas Emissions (1990-2012)",
subtitle = "including land-use change and forestry, relative to 1990 levels",
x = "Year", y = expression("Methane relative to 1990 levels (%"~MtCO[2]~e~")"))
Individual country models
India
model4_India <- lm(PercentRelativeChangeCH41990 ~ Year,
subset = (Country == "India"),
data = data4)
summary(model4_India)##
## Call:
## lm(formula = PercentRelativeChangeCH41990 ~ Year, data = data4,
## subset = (Country == "India"))
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.6450 -0.9391 -0.1521 1.3965 2.3050
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.316e+01 1.723e+00 -19.25 8.03e-15 ***
## Year 4.916e-08 1.722e-09 28.55 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.729 on 21 degrees of freedom
## Multiple R-squared: 0.9749, Adjusted R-squared: 0.9737
## F-statistic: 815.2 on 1 and 21 DF, p-value: < 2.2e-16plot(subset(data4,
Country == "India", # choose only Country == "India"
select = Year, # just the 'Year' data
drop = TRUE), # changes the table into a list of numbers ('vector')
subset(data4,
Country == "India",
select = PercentRelativeChangeCH41990,
drop = TRUE))
abline(model4_India)
par(mfrow=c(2,2))
plot(model4_India)
United States
model4_USA <- lm(PercentRelativeChangeCH41990 ~ Year,
subset = (Country == "United States"),
data = data4)
summary(model4_USA)##
## Call:
## lm(formula = PercentRelativeChangeCH41990 ~ Year, data = data4,
## subset = (Country == "United States"))
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.6390 -2.7597 0.6052 1.8340 4.5480
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.459e+00 2.870e+00 0.509 0.6164
## Year -5.521e-09 2.869e-09 -1.925 0.0679 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.88 on 21 degrees of freedom
## Multiple R-squared: 0.15, Adjusted R-squared: 0.1095
## F-statistic: 3.704 on 1 and 21 DF, p-value: 0.06792plot(subset(data4,
Country == "United States", # just Country == "United States"
select = Year, # just the Year data
drop = TRUE), # changes table to a list of numbers ('vector')
subset(data4,
Country == "United States",
select = PercentRelativeChangeCH41990,
drop = TRUE))
abline(model4_USA)
par(mfrow=c(2,2))
plot(model4_USA)