Blind Adherence — Significant Variations in Food System Emissions Data (Part 1)
Written on
In discussions about reducing emissions to address climate change, it's essential to grasp some fundamental points that I began to outline in earlier sections.
Three key points for those who prefer concise information: - Emission figures related to food production are often unreliable. - As a result, these figures should not be taken at face value. - Encountering such numbers frequently indicates that the author lacks insight into their origins—similar to me attempting to write about musical notation without expertise.
Many readers are aware that one-third of global emissions derive from food systems. This statistic will serve as a foundation for evaluating its reliability.
Note: I have elaborated on the definition of food systems previously.
To identify the original source of this claim, we refer to a study by scientists at the Joint Research Center (JRC) of the European Union, published in the journal Nature, titled: "Food systems are responsible for a third of global anthropogenic GHG emissions" (Crippa et al., 2021). This research is commonly referred to as EDGAR FOOD. The study states:
> “It addresses the lack of detailed data for many nations by offering sector-specific contributions to food-system emissions, crucial for formulating effective mitigation strategies. In 2015, food-system emissions reached 18 Gt CO2 equivalent annually, accounting for 34% of total GHG emissions.”
This information rapidly circulated through social media and news outlets globally, becoming widely accepted as fact. Yet, few have scrutinized the entire study, including its methodology and the data behind these estimates. Yes, these are estimates. They rely on input data categorized as follows:
- Tier 1 — These emissions are rough estimates, often based on broad variables for areas larger than individual countries, such as the average energy emission factor for the EU. However, we know that such averages are misleading since each country has a distinct energy mix at any point in time.
- Tier 2 — More precisely derived from local estimates.
- Tier 3 — Accurate figures based on precise measurements (the smallest category in this hierarchy).
Considering this context, one must ask—how reliable is the data claiming that one-third of global emissions come from food systems? The answer is consistently the same. It is a blend of estimates with a small portion of accurate data. I encountered vague figures lacking clear calculations and amusing notes in the accompanying spreadsheets, such as:
> “To be completed by Francesco Tubiello”
That note referred to data from 2015, and it seems it remains incomplete even in 2021. This study was published in NATURE! If someone were to verify the data's consistency, format, and quality, such a note would not be acceptable. I must reiterate that this is a critical tool for determining GHG emissions reductions, not a scoreboard for local sports events.
And that was just an introduction. Now, let's delve into the specifics.
Total 18Gt of CO2e Distribution
Firstly, it’s important to note that the EDGAR FOOD analysis was based on the IPCC AR5 (2014) methodology, specifically GWP100 AR5, and focused primarily on data for the year 2015 (despite including information from 1990). The authors of the aforementioned study noted:
> “We have developed a new global food emissions database (EDGAR-FOOD) estimating greenhouse gas (GHG; CO2, CH4, N2O, fluorinated gases) emissions for the years 1990–2015.”
This distinction is crucial to avoid confusion between different datasets. Nonetheless, the data remains inconsistent.
EDGAR FOOD identified the precise figure of 17.95Gt CO2e in their publication in Nature (refer to Source data): 43016_2021_225_MOESM3_ESM.xls
The authors rounded this figure to 18Gt CO2e for their conclusions, which is acceptable.
To ensure accuracy, the JRC published the EDGAR DB with a time lag, meaning they released data for GHG emissions from 2015 in 2019. The updated EDGAR DB employs the more recent AR6 methodology, so be cautious when making comparisons! We should compare data for 2015 according to AR5. Any comparisons with other years would be misleading.
In 2015, total global emissions were recorded at 49.113Gt CO2e (JRC/EDGAR DB v5).
Returning to the calculations:
17.95Gt out of the total 49.1Gt CO2e equals approximately 36.6%, compared to the 34% cited in the study.
If 17.95Gt is indeed the correct figure, then the basis for the 34% calculation should have been 52.94Gt CO2e. However, EDGAR could not have recorded such emissions in its GHG Emissions Databases. The reason for this discrepancy is that such emissions were only seen in 2018, specifically at 51.2Gt CO2. The JRC/EDGAR only became aware of this in 2019, while the study utilized data from 2015. It is plausible that there was a mix-up of global GHG emissions in the study's calculations. This document was published in 2021, so this remains a speculation.
The Second Course — Ongoing Data Processing Mysteries
Examining the subsequent appendix of the mentioned Nature study reveals the following aggregated data:
I opened the second attachment from the same research. Surprising? Not at all. This chaotic data representation is quite typical in emissions data from this source. I have identified this data inconsistency and significant gaps in data taxonomy and handling in various documents from JRC/EDGAR and FAOSTAT.
If such a study has passed peer review in Nature, which has stringent submission criteria, it further validates my argument that—science requires a reboot. Sadly, I have encountered similar issues in Nature before. This serves as a reminder to critics who claim that only data published in official scientific journals is credible. Well, that’s not always the case.
This is evidently a failure of standard data controls at every conceivable level. Consequently, I reached out to the JRC leadership responsible for the EDGAR research. We initiated discussions on data science processing protocols and standards (a few months back) from top management downwards. Eventually, they suggested that if I wanted to work for the JRC as a data scientist, I should formally apply for a standard position. Remote positions, even for external consultants, are not permitted. For defining data science strategy principles? In 2022? An illustration of how officials can claim to combat climate change.
Let's revisit this image:
If 52% of CO2 originated from the total of 17.95Gt CO2e, it would mean that fossil fuels accounted for 52% x 17.95Gt = 9.3Gt CO2 from fossil fuels. This figure approximately aligns with the data from the EDGAR research, which indicated a total of 9.2Gt CO2 emissions for food systems in 2015. However, as usual, 9.3 is not equivalent to 9.2, even though they stem from the same source and lead to the same conclusion.
In 2015, JRC/EDGAR DB v5 estimated that 36.3 Gt of CO2 came from fossil fuels (again, these are estimates). This implies that food systems constituted 25.6% of global CO2 emissions from fossil fuels.
A Deeper Examination:
I attempted to categorize emissions from EDGAR FOOD into the same categories used by its author (JRC) in EDGAR Emissions. The outcome is as follows:
From this, we can conclude that global food systems in the power sector emitted fewer CO2 emissions from fossil fuels in 2015 (3,807 Mt) than the entire Chinese power sector (4,396 Mt) in the same year by 13.3%.
> Do you think there would be as much global criticism directed at China for its excessive power industry emissions as there has been towards global food system emissions? Or when considering that transport emissions from the entire global food system (1,548 Mt) were 12% lower in 2015 than US transport emissions (1,752 Mt), do you believe there would be similar criticism of US transport emissions?
I do not.
Do you feel that in the past five years, the movement advocating for the end of livestock farming in favor of plant-based diets has gained significant momentum? Do you think Bill Gates has commented on the need for China to reduce power sector emissions? Or his fellow investors in vegan food?
I do not.
You can explore my thoughts on the futility of industrial livestock farming here:
Population, livestock farming, and contentious emissions debates — a brief overview.
A Few More Perspectives
Countries responsible for 50% of global food system GHG emissions (according to the EDGAR Food DB): China, Indonesia, United States, Brazil, India, Russian Federation, Zambia.
These seven nations account for half of global food system GHG emissions. Here's a breakdown of their contributions:
For a reminder of the planet's largest polluters (the Pareto principle), you can refer to this source.
That should suffice for today—a small mathematical exercise for data scientists and those seeking more than superficial keywords from climate change data science.
Conclusion
Anyone embarking on research or straightforward interpretation of emissions data without a deep understanding of its context and creation will likely propagate misinformation. Unfortunately, this occurs frequently—not just here on Medium, but everywhere. Many self-proclaimed emissions experts emerge after merely reading a few keywords. They are not experts; they contribute to the dissemination of misinformation.
This discussion should serve as a reminder that if someone writes about emissions and generates arbitrary data, such an article or blog should be regarded as a shallow educational source.
To some, my critical remarks may resemble those of climate change skeptics. I must clarify—I am a staunch advocate for climate change education. However, I refuse to be a blind follower. It is crucial to identify and address mistakes made in climate change research from all angles. Misleading data has never benefited anyone more than those profiting from it.
This science is complex, not simple.
For a more comprehensive list of my published content on Medium, please refer to the following:
The content navigator I posted on Medium.
Here, you will find quick summaries and connections between various topics.
Enjoy!
References: Crippa, M., Solazzo, E., Guizzardi, D. et al. Food systems are responsible for a third of global anthropogenic GHG emissions. Nat Food 2, 198–209 (2021). https://doi.org/10.1038/s43016-021-00225-9
Or here: https://figshare.com/articles/dataset/EDGAR-FOOD_emission_data/13476666