Tracking the changes in CO2 emissions from the energy sector.
Introduction
In recent years the discourse around climate change has gained significant momentum. Every sector of the economy has been impacted with governments placing major targets to curb CO2 emissions. The one key sources of CO2 emissions has been from energy usage. As industries seek alternative carbon neutral energy sources, in this project the CO2 emissions form specific energy sectors will be investigated.
Provided by the Food and Agriculture Organisation of the UN, the dataset reviewed in this project has records covering approx. 50 years from 1970 to 2019. It holds a breakdown of CO2 emissions for a number of energy sectors from a myriad of countries. As it will become apparent, CO2 emissions from energy industries fluctuates over time and country to country. It is also notable that the energy types used in the fishing sector are particularly polluting. Moreover, the energy reliance differ from nation to nation and therefore the CO2 emissions from such sectors also vary.
This project will evaluate the data at a global level and shift its attention to two groups of nations G7 and South American nations to understand the variations in CO2 emissions.
G7
G7 consists of some of the wealthiest countries in the world. As such, the populations in these nations are likely to have the biggest demand for energy through the use of luxury and leisure items. On the other hand, the wealth in these nations are likely to enable energy industries to invest in more effective methods of producing energy. Therefore the industries may be able to keep CO2 emissions low.
South American
Juxtaposing the G7 are the South American countries. They have developing economies and are also in the southern hemisphere. Therefore, these countries are likely to have differing energy needs. In the final section, limitations of this dataset are explored. The data has several weaknesses which affect the conclusions which can be drawn.
Overview
The dataset requires some data progressing. For further details on this, please refer to the appendix. It should be noted that as the appendix will emphasis, the data has some mathematical outliers which are persistent. Subsequently, the median average has been used though this project as it is less likely to be skewed by extreme outliers. Furthermore, due to some null values, the data has been capped to include only data from 1971 to 2018, rather than the full scope of the data.
Following these steps, the data is as follows:
Global energy emissions
Beginning at the global level, below is the average CO2 emissions from the energy sector. It illustrates that there are significant fluctuations in the CO2 emissions. This may be due to aspects such as good weather requiring individuals to use less energy for heating or major events placing extra demand on energy. Additionally, since 2016 the CO2 emissions have been decreasing and is more or less on par with the level of emissions recorded in 1971. It, however, remains higher than the extreme low recorded in 1976.
Author's notes: For interactive plotly graphs, please visit the Kaggle notebook.
Cumulative total shows the scale of CO2 emissions from the energy industry over approx. 50 years, emphasising the scale of the issue. It is particularly troubling if it is assumed that no new product has been created to absorb even a part of this emission. In fact, reports of deforestation suggest that the existing entities which absorb CO2 may have diminished during the same time period.
Dividing the CO2 emissions to each item indicates that “Gas-diesel oils used in fisheries” had the highest average emissions and is over three times more polluting than the next energy sector with the highest level of CO2 emissions.
Furthermore, Electricity has an average CO2 emission compared to the other energy types.
Taking a snapshot of the 1971 data, stresses that Fuel oil used in fisheries was the most polluting, emitting over three times more CO2 than the next polluting energy type, Gas-Diesel oil. In addition Electricity’s role in the CO2 emissions from the energy sector appears to be relatively low.
Contrasting the above with 2018 data, the CO2 emissions from the electricity have jumped up significantly. An aspect of this is likely to be due to increased demand due to the popularity of items such as PCs to phones and smart devices.
That being said, the most polluting energy type is “Gas-diesel used for fisheries”. Further investigation will be required in understanding the exact cause of this high level of emission such as the level of supply throughout the sector to determine whether the energy type for fisheries is extremely polluting.
The overlay of the 1971 and 2018 data highlights the extent of the difference. It shows that the CO2 emissions by Fuel oil used in fisheries has decreased significantly. This is extremely positive in the context of climate change. However the cause of the decrease is unclear from this dataset.
Several energy types such as Motor gasoline, LNG and LPG’s CO2 have remained relatively unchanged in the approx. 50 year time period.
Energy fluctuations over time
As a cross section review indicated the energy usage has changed significantly over time. This is reinforced in the below graphs which outline the fluctuations in CO2 for each item over time. It is clear that the “Gas-diesel used for fisheries” and “Fuel oil used in fisheries” has undergone significant changes. “Fuel oil used in fisheries” has seen significant CO2 emission decreases. “Gas-diesel used for fisheries”, on the other hand, increased in the 1990s but has since also recorded year on year falls.
The fluctuations in other energy types have undergone minor changes.
The dynamics of the energy fluctuations are also presented in a heatmap below, demonstrating the fluctuations in a more visual manner. Please see appendix, section 2 for an interactive graph which overlays the fluctuations directly on one graph for direct comparison of the CO2 changes.
G7
In this section the data specifically for the G7 countries will be examined. As the information from the UK government highlights, the list of G7 nations are as follows:
UK,
USA,
Canada,
Japan,
Germany,
France,
Italy,
EU
As the EU is a collection of nations and some key member countries are also separately in the G7, they have not been included in this graph.
For G7 countries, the overall CO2 emissions from the energy sector is generally around 400 to 600 kilotons. However in 1978 and in 1990, there were significant spikes in the emission levels. Causes of these spikes are beyond the scope of this project.
Following is a graph illustrating the overall average emissions for the energy sector for the G7 countries. It highlights that unlike the global trends, Gas-Diesel oil was the most polluting sector for the G7 countries, nearly six times more polluting compared to Coal, the least polluting sector.
Comparing the G7 1971 data against the 2018 indicates that Gas-Diesel oil CO2 emissions have greatly increased in the approx. 50 years. This is likely to be due to the push for Diesel cars. Also contrary to global trends the CO2 emissions from “Motor Gasoline”.
Looking at the energy sectors for the G7 countries individually, presents some fascinating trends. Electric energy CO2 emissions, though having some significant fluctuations, have found a stable limit at 2000 kilotons. In contrast, significant work is needed to bring down the CO2 emission from Gas-Diesel as it dominates the missions compared to every other energy type.
Finally the heatmap clearly demonstrates the high rate of CO2 emission by the Gas-Diesel oil sector. Electricity’s CO2 grows throughout the approx. 50 years but not to the extent of Gas-Diesel. LNG oddly has a spike in 1990s which may be the cause of the overall increase in CO2 for the G7 in 1990s. However, it is still unclear what caused this increase, although 1990 was the year of the oil crisis.
South America
To directly contrast the G7 data, is the data for the South American nations. Many of the South American nations are regarded as being a part of the developing nations. As such, their reliance on energy is likely to differ from the G7 nations. Moreover, Brazil for instance has had some significant economic growth during the period in concern (The World Bank, 2022 ). Therefore it may have significant energy needs that differ from nations experiencing differing economic development.
Firstly it is clear that the overall level of CO2 emission for the South American countries is significantly lower than the G7. The highest average level in South American is just over 92 kilotons. The G7’s highest value was approx. 1,190. Secondly in the late 1980s, there was a major drop in the CO2 emissions before the trend reverting and peaking in 1997. Since then, there has been a plethora of fluctuations but maintain a general downwards trend.
Similar to the G7, the most polluting energy sector is Gas-Diesel for the approx 50 year period. However, Gas-diesel oils used in fisheries is also prominent, and the scale of difference should be noted between the two groups of countries.
Reviewing the 1971 data against the 2018 data for South American indicates that following the global trend “Fuel oil used in fisheries” has undergone the biggest transformation. The changes with other energy sectors also generally follow the global trends previously identified.
The individual items and its trends for South American shows that it has significantly more fluctuation in a number of energy sectors compared to G7 and even the global trends. As the scales appear to be the same for particularly the global trend, it can be assumed that the presentation of the data is accurate. However, this then leads to inquiries relating to the cause of the fluctuations. One aspect may be the increased demand for energy due to the growth in the economy. However further research examining economic trends against this data is needed to conclusively state this.
Finally, the heatmap presents the fluctuations in each type of energy sector compared to one another. As was discussed previously, the increase in CO2 emissions in Gas-Diesel is undeniable and major changes will be required for South American nations if they are to bring CO2 emissions and climate change under control. LPG CO2 emissions is particularly low throughout the period concerned.
Limitations of the dataset
The dataset has numerous limitations that need to be taken into account when drawing conclusions from this dataset. Some of these limitations are outlined below.
Firstly, the dataset does contain information on the changes in production. For instance, has there been any improvements in the production of one energy type over the other? As such, it cannot be concluded whether the CO2 decrease from 1970 to 2019 for “Gas-diesel oils used in fisheries” is due to decrease in demand/production or due to efficiencies in production.
Secondly, due to the focus on CO2 emissions, the role of sustainable energy sectors such as wind and solar, and their role is unclear.
Thirdly, the calculations behind CO2 emission values are likely to have some limitations. Gas, for instance, needs to be burnt to produce the energy they hold. Therefore emitting additional CO2 after they have been produced. Electricity, on the other hand, does not emit further CO2 at the point of usage. Consequently, this dataset could damp down the level of pollution of some energy types.
Furthermore, linked to the previous point is the extent of calculations. For example the machinery used to create such energy sources are likely to have CO2 emissions in its production. Likewise, transporting energy to the individuals needing it will also incur CO2 emissions. However it is unclear if these aspects have been included in this dataset.
Moreover, CO2 is not the only negative aspect of the energy sector. Oil spillage and acid rain are some aspects which are not captured in this dataset. This further emphasises that conclusions drawn from this dataset must be critically reviewed in the context of the wider discourse concerning climate change.
Electricity can be produced through the use of fuel such as coal and Gas, i.e. energy listed separately in the dataset (Hausfather, 2019). Once again it is clear how this has been accounted for in the dataset. Finally as The Economist (2011) highlights even the GDP is known to have issues with accuracy especially in poorer nations. If this is to be expanded, it can be assumed that these figures may not be fully accurate regardless of the other issues highlighted above. Having said that, the ability to secure more accurate data is likely to face greater issues.
Conclusion
This has been an extensive examination of the global CO2 emissions for the energy sector for roughly a 50 year period. It highlights the fluctuations in CO2 overall as well as for specific energy industries. Energy centred upon fisheries is particularly polluting though its level of pollution appears to have decreased in the 2000s and onwards. In addition, different countries recorded varying levels of CO2 emissions from different energy types. However due to a myriad of limitations with the data, further investigation is required to understand the cause of such fluctuations.
Appendix
Understanding the dataset
This section will examine individual elements of the dataset and provide explanations concerning base assumptions made throughout the project.
On the surface, there does not appear to be any null values.
Domain, Element and Unit
Firstly, the Domain Code" and "Domain" contain only one variable, GN/Energy use. As such they are unlikely to provide any insight. Similarly, "Element Code" and "Element" also contain one variable. Therefore these can be removed from the dataset with no loss of understanding/ distorting the overall dataset.
Whilst the unit column contains one variable, dropping this column will lead to a reduction in the readability of the data. As such, this variable will be retained during the body of the project.
Area
The Area variable holds a list of country names. It should be noted that entries such as "Saint Kitts and Nevis" though may appear to refer to a region are in fact one country as classified by the UN.
For added readability, the term “Country” has been used for added clarification. In majority of the cases this term holds true under the below definition of country:
“In broad terms, a country is a group of people governed by a government, which is the final authority over those people. There is a political setup that governs everyone in the country.”
There is, however, a possibility that some entries may be territories.
In addition, as the dataset uses “United Kingdom of Great Britain and Northern Ireland” and “United States of America” to represent the UK and US respectively and for readability it may be best to amended these to UK and US.
It should be noted that names of countries and territories are subject to change over time. The most clear representation of this is the inclusion of “USSR” which was dissolved in 1991 to current day Russia and other nations. (Office of the Historian, 2021) This is likely to impact the dataset as countries which ceased to exist or came into existence during the years of 1961 and 2020 are likely to have less data points than established nations.
Duplicated data
There are some columns in this dataset where the information has been duplicated. For example, “Year Code” and “Year” share the same information. As such, one column can be dropped without affecting the reading of the data.
Likewise, “Flag” and “Flag Description” hold the same information. However as Flag presents the information in code form it could could pose issues for the reader, thus “Flag” is dropped. This is mirrored in the “Item Code” and “Item”. Once again “Item Code” is dropped to ensure readability is not hampered by the changes to the dataset.
Year
The dataset covers a 49 year time period, starting from 1970.
Item
The dataset covers 9 energy industries which include the following:
It appears that the items are not evenly distributed in the dataset. It favours some industries such as Gas-Diesel oil and Motor Gasoline over Fuel oil used in fisheries. This could be the lack of popularity of some energy types over others but comparisons between the industries may be skewed due to this imbalance.
Furthermore, comparing the items against the year, suggest that there are missing values for “Fuel oil
used in fisheries” and “Gas-diesel oils used in fisheries”. This is likely to cause issues in the project.
Further investigations indicate that the missing values are in 1970 and 2019. As such, it may be best to use only data from 1971 and 2018.
Flag description
The “Flag Description” variable contains three variables, 'FAO estimate', 'International reliable sources', and 'Aggregate, may include official, semi-official, estimated or calculated data’. However they too are not distributed in the dataset evenly. 'Aggregate, may include official, semi-official, estimated or calculated data’ entries are significantly limited in comparison to the other data types.
In fact the “Aggregate, may include official, semi-official, estimated or calculated data” account for less than 1% of the data. As such, this variable is dropped particularly as the variable is an aggregate of several calculations. For reference, FAO estimates and international reliable sources account for 64% and 35% of the data. Therefore these variables are left in the data source.
The distribution of the flag description against items is such that there are more records of Motor Gasoline and LNG under the FAO estimates. However international reliable sources provide more data for gas-diesel oil and electric.
Distribution of data
The data is extremely broadly distributed with many values in the range of 0 to 25000 range with a strong skew to the right.
This distribution remains relatively unchanged when each energy industry is examined individually.
In other words, the data contains a high number of mathematical outliers as further emphasised by the following boxplots. Further research is required to understand the cause of these fluctuations, although causes could include poor weather leading to increased demand for heating.
Regardless this poses an issue when using mean averages as it is sensitive to extreme outliers.
This appears to be the same when reviewing the distribution of the data in context of the Flag description type, i.e. data contains outliers.
Works Cited
Below are some of the particular key resources used in this project.
Jee, Ken, and Andrara Olteanu. “Challenge 1 Tutorial - Line Chart (Seaborn).” Kaggle, 2022, Challenge 1 Tutorial - Line Chart (Seaborn). Accessed 7 March 2022. Knaflic, Cole Nussbaumer. Storytelling with Data: A Data Visualization Guide for Business Professionals. Wiley, 2015. Plotly. “Plotly express with Python.” Plotly, 2022, https://plotly.com/python/plotly-express/. Accessed 15 March 2022. Shaikh, Reshama. “Enriching Data Visualizations with Annotations in Plotly using Python.” Medium, 2021, https://medium.com/nerd-for-tech/enriching-data-visualizations-with-annotations-in-plotly-using-python-6127ff6e0f80. Accessed 14 March 2022.
"Wind and other clean, renewable energy will help end our reliance on fossil fuels and combat the severe threat that climate change poses to humans and wildlife alike." - Frances Beinecke
Commenti