My previous post, The Energy Balance Model, outlined the Physics of climate science. It explained how all climate models build on an equilibrium temperature that results from the balance between incoming absorbed shortwave radiation (ASR) and outgoing longwave radiation (OLR). Solar flux, albedo and greenhouse gases (GHG) all impact this balance. GHGs play a particularly important role in generating radiative forcing that increases temperature at ground level while at the same time lowering of temperature in the stratosphere, a key theoretical prediction that has been verified by lapse rate measurements. There is a palpable and growing sense of urgency and alarm about climate change impact among the 97% of climate scientists that believe it is driven by human activities. It’s not a debate any more requiring more evidence. As John Oliver memorably put it:
You don't need public opinion on a fact.
In this post we’ll look at the broad landscape for addressing the situation we find ourselves in. What is to be Done? There are two broad options - mitigation to address the causes and adaptation to address the impacts as outlined in the diagram below from CarbonBrief:
A good starting point for considering mitigation is Figure 4.4 from the IPCC’s 2023 Climate Change Synthesis Report (AR6) shown below:
A lot of important information is packed into this diagram. The left hand side of panel (a) outlines a range of selected climate responses and adaptation options across six broad dimensions, assessed in terms of their practical feasibility at scale in the near term up to 1.5C warming. Those six dimensions being: 1. energy system, 2. industry and transport, 3. cities, settlements and infrastructure, 4. land, ocean, food and water, 5. health and nutrition, 6. society, livelihoods and economies. The right hand side of panel (a) outlines selected mitigation options with their estimated cost and potential impact in 2023. Those costs are net lifetime discounted monetary equivalent of avoided GHG emissions relative to reference. The longest and bluest bars are the ones we need to focus on if our goal is to meaningfully scale up action to reduce temperature rise and the associated risk of hitting irreversible climate tipping points and to do so while minimising cost. The lower panel (b) provides an indication of demand-side mitigation options further out at 2050 based on a large number of bottom-up studies from around the world. The green arrows here represent demand-side emission reduction potentials across five sectors: Food, Transport, Buildings, Industry, Electricity. A few key themes emerge for action over the next five years running up to 2030 from examination of the diagram:
Continuing acceleration of energy transition to renewables focusing on expanding solar and wind.
Reducing conversion of natural ecosystems and investing in restoration through reforestation.
Reducing primary energy demand via electrification to be offset by greater demand for electricity.
Improving efficiency and reducing waste come up in many of the mitigations.
We can use generative AI to build document copilot for the entire AR6 Synthesis report then prompt it with a request to provide a summary of the key mitigations. The rationale for doing so would be to compare the AI response with my assessment. The method, called RAG (Retrieval Augmented Generation), involves breaking down the document text into indexed chunks which are stored in a local vector database. The most relevant chunks for a specific query are retrieved and fed as input to an LLM (Large Language Model) such as OpenAI’s GPT or Meta’s Llama to allow that LLM to compose a text-based response taking the supplied context into account. Here is a recipe that uses the popular langchain framework for building and using a RAG with the source AR6 pdf as input. An option is provided to generate the response entirely locally by using an Llama 3.2 LLM via Ollama, an open-source tool that allows you to run LLMs on you own laptop. Leveraging a local LLM for inference is noticeably slower than accessing it through an API but it also has a markedly lower environmental impact which is relevant given the topic. I also used napkin.ai to generate a non-RAG based report which is available here. The napkin response is noteworthy for calling out the importance of policies and governance, equity and justice, and human behavioural changes. These topics were largely missed from the RAG-based response. However, it’s unclear how much CO2 was generated using the napkin.ai. Here is an example Ollama interaction including the estimated number of tokens, which roughly maps to word count, used in the inference both in terms of input and output. The answer isn’t deterministic of course - if you try changing the prompt slightly or simply re-run it a few times, you get broadly similar yet different answers. The input token count is quite large because 30 separate RAG chunks are passed onto the LLM. Using an LLM from OpenAI or ChatGPT would be incur monetary and carbon cost. The cost using Ollama is effectively zero because the LLM it uses is running locally not via a paid API:
> What are the top 10 mitigations we need to focus on for maximum reduction in global warming?
Based on the provided text, here are the top 10 mitigations that can help reduce global warming:
1. Reduce cumulative CO2 emissions: Decrease greenhouse gas emissions from nearly all considered scenarios and pathways.
2. Transition to renewable energy sources: Shift from fossil fuels to renewable energy sources like solar, wind, and hydroelectric power.
3. Increase energy efficiency: Improve the energy efficiency of buildings, transportation, and industry to reduce energy consumption.
4. Electrify transportation: Promote the adoption of electric vehicles and public transportation to reduce emissions from transportation.
5. Carbon capture and storage: Implement carbon capture and storage technologies to reduce emissions from industrial sources.
6. Protect and restore natural carbon sinks: Preserve and expand forests, wetlands, and other ecosystems that absorb CO2 from the atmosphere.
7. Implement climate-resilient agriculture: Promote sustainable agricultural practices that enhance soil health, reduce synthetic fertilizer use, and promote agroforestry.
8. Reduce methane emissions: Implement policies to reduce methane emissions from agriculture, natural gas systems, and landfills.
9. Enhance energy storage and grid resilience: Develop and deploy advanced energy storage technologies and grid management systems to ensure a stable and reliable energy supply.
10. Promote sustainable land use: Encourage sustainable land use practices like reforestation, afforestation, and permaculture to sequester carbon dioxide from the atmosphere.
These mitigations can help reduce global warming by:
* Decreasing greenhouse gas emissions
* Increasing energy efficiency and renewable energy production
* Protecting and restoring natural carbon sinks
* Promoting climate-resilient agriculture and reducing methane emissions
* Enhancing energy storage and grid resilience
* Encouraging sustainable land use practices
By focusing on these top 10 mitigations, we can make significant progress in reducing global warming and mitigating its impacts.
---------- COST ----------
cost=$0.0, prompt_tokens=3880, completion_tokens=278, total_tokens=4158Central to all three assessments (human, Ollama and napkin.ai) is the switch over of as many human activities as possible from a reliance on fossil fuels as our main Primary Energy source to electricity powered by renewables. Primary Energy refers to the total energy content of natural resources before any conversion process such as coal, oil or renewable electricity. Realising this switch will require a Copernican revolution in mindset, a revival of the heliocentric view of life that was central to human existence for millennia before the industrial revolution. Where the sun is once again our primary source of energy via solar flux and wind on earth. Can Homo Sapiens channel the spirit of our antecedents who revered the sun gods? Is the next stage of our growth to evolve into Homo Helios? Installing mass scale renewable infrastructure backed up by grid scale battery arrays is one stage of the energy pivot. The second is decarbonising the global economy to leverage that renewable infrastructure and then ramping down fossil fuel usage hard. A regular objection raised is that it is too complicated to change our status quo and dangerous to ramp down too fast. But everything is impossible until it’s not and fossil fuels are hideously wasteful in terms of energy efficiency, far more so than renewables. So why do many people lobby for fossil fuels and insist that we are unable to replace them? Why haven’t we seen large-scale action to end fossil fuel usage? Jan Rosenow suggests that a Primary Energy fallacy is in operation. The conventional comparison made between different energy system inputs treats fossil fuels far more favourably than deserved:
Measuring energy systems purely on primary energy inflates the perceived contribution of fossil fuels while underestimating renewables’ efficiency and untapped efficiency potentials through electrification. …. In conventional fossil fuel systems, significant amounts of primary energy are lost as waste heat during combustion. For example, a coal fired power station only converts 40% or so of the coal burned into electricity. By contrast, renewable systems like wind and solar produce electricity directly.
Brett Christophers in his book The Price is Wrong provides a darker rationale linked to financialisation and the inability of advocates of capitalism to properly price carbon emissions:
[It] is not that transitioning to renewables is too expensive, but that saving the planet is not sufficiently profitable. Today's consensus is that the key to curbing climate change is to produce green electricity and electrify everything possible. The main economic barrier in that project has seemingly been removed. But while prices of solar and wind power have tumbled, the golden era of renewables has yet to materialize. The problem is that investment is driven by profit, not price, and operating solar and wind farms remains a marginal business, dependent everywhere on the state's financial support. We cannot expect markets and the private sector to solve the climate crisis while the profits that are their lifeblood remain unappetizing.
Nature-based solutions also figure prominently as mitigation options in both assessments. AR6 includes within nature-based solutions the planting of new forests, allowing degraded forests to regrow, improved forest management, improved cropland management and coastal wetland restoration including mangroves. These mitigations form part of a wider set of CDR technologies, or, as they are frequently termed in climate literature Negative Emissions Technologies or NETs, that have significant potential to remove CO2 from the atmosphere, reduce temperature rise, and help combat ocean acidification. NETs are core to mitigation in that they directly target the reduction of atmospheric CO2. The set of NETs includes:
Afforestation/Reforestation: methods of planting trees to increase the amount of forest cover. Afforestation involves planting trees on land that has never had forest cover before. Reforestation is the regrowth of previously forested land that has been degraded for other uses.
Soil Carbon Sequestration: this is the process of removing CO2 from the atmosphere and storing it in soil as organic carbon. Soil Carbon Sequestration methods include practicing no or low till agriculture and changing food crops.
Biochar: this is a charcoal-like substance made from organic materials that have undergone the process of pyrolysis where they are heated in low oxygen at high temperature.
BECCS: stands for bioenergy with carbon capture and storage, BECCS is the process of capturing and permanently storing CO2 from biomass energy generation.
DAC: stands for Direct Air Capture. DAC technologies extract CO2 directly from the atmosphere at any location unlike carbon capture which is typically conducted at the point of emissions. CO2 extracted through DAC can be permanently stored in deep geological formations or used in other applications.
Enhanced Weathering: this is a carbon removal approach where the natural process of weathering is speeded up though grinding silicate rocks into fine particles and spreading them on land to accelerate carbon capture. .
Ocean Fertilisation: a form of geoengineering that involves adding nutrients to the ocean to increase phytoplankton growth and remove CO2 from the atmosphere.
A recent BBC Panorama program provocatively entitled Can Scientists Save the World? surveyed frontier research work going on around the world in several of these fields including DAC, Biochar and Ocean Fertilisation. The overall sense was of CDR being a niche and early stage affair at present with insufficient scale potential to make much of a difference in the next decade or two. The various CDR methods listed above are not equal as this excellent CarbonBrief article points out in the diagram below. Only two of them, namely Afforestation/Reforestation and Soil Carbon Sequestration that have significant scale potential over the next decade or two. After 2050 DAC has the potential to scale up and start drawing down CO2 but much is unclear about the costs involved and the time it would take to use it to take tens of Gt CO2 out of the atmosphere. The sheer scale of removal these CDR technologies would have to support is unimaginable to most today. According to RMI it would require $1 trillion CapEx per Gt CO2 removed and we would need to be removing 40-50 Gt CO2 per year. As an industry, CDR at full scale would employ millions of workers:
Deploying ten gigatons of CDR per year by 2050 would mean that, within a few decades, this nascent industry would need to grow to a size larger than today’s global coal industry. This goal is achievable but will require significant and deliberate investment and growth in relevant supply chains.
The diagram above from the CarbonBrief article referenced earlier underlines the importance of backing entrepreneurs trying to save the world by ensuring investment in mitigations that have the biggest short term impact as well as moonshots. However, a huge imbalance exists between where early stage investment is going and where it needs to go. ClimateTech VC (CTVC) pointed this out in early 2022 with a graph illustrating a disproportionate level of venture capital going into EVs relative to ecosystem restoration. There’s a clear mismatch between Afforestation/Reforestation on the one hand and EVs/e-bikes on the other:
CTVC offered this sharp observation on the discrepancy:
Mismatch of climate venture dollars to relative emissions impact. The above charts are illustrative and directional, but it doesn’t take a precision ruler to notice the relative differences between emissions impact vs what venture capital funding has been prioritizing. In particular, over the past two years, $16B has been invested in electric mobility subsectors, with a combined ~1.7 GtCOe annual emissions reduction impact ($10B for 1 GtCO2e). Meanwhile, ecosystem restoration holds down 2.9 GtCOe yet received a meager $0.2B in comparative venture investment ($.07B for 1 GtCO2e).
Investors could help combat climate change by more proactively supporting early stage startups working on ambitious propositions to scale out nature-based solutions. HVAC (Heating, Ventilation, and Air Conditioning) and energy efficiency solutions also look underinvested relative to their potential. The imbalance on view here reveals a funding bias for technology-driven over nature-driven propositions. It aligns with a general tendency amongst investors to adopt a technology-first mindset when approaching mitigations. Propositions that involve AI, EVs and DAC are perceived to be cool and secure abundant funding while more mundane work on soil, tree planting and energy efficiency is underinvested. We could extend the climate oracle copilot to include climate intelligence around current investments. Then pose it this question - where should we be focussing our investment for maximum returns? The AI copilot built earlier again betrays a tech bias in its response:
> Where should we be focussing venture capital investment for maximum impact on mitigation areas?
Based on the information provided, I would recommend focusing venture capital investment in the following areas to maximize impact on mitigation:
1. Renewable Energy: Investing in companies that develop and deploy renewable energy technologies such as solar, wind, hydro, and geothermal power.
2. Energy Storage: Supporting startups that work on advanced battery technologies, hydrogen fuel cells, or other energy storage solutions to enable a smoother transition to renewable energy sources.
3. Carbon Capture and Utilization (CCU): Investing in companies that develop innovative CCU technologies to convert CO2 into valuable products such as fuels, chemicals, or building materials.
4. Electrification of Transportation: Focusing on startups that develop electric vehicle (EV) technologies, charging infrastructure, and battery recycling solutions to accelerate the adoption of EVs.
5. Sustainable Agriculture: Investing in companies that promote regenerative agriculture practices, vertical farming, and other sustainable agricultural methods to reduce greenhouse gas emissions from agriculture.
6. Climate-Resilient Infrastructure: Supporting startups that develop innovative climate-resilient infrastructure solutions such as sea walls, green roofs, or flood-resistant buildings.
7. Advanced Materials and Manufacturing**: Investing in companies that develop new materials and manufacturing processes that can help reduce carbon emissions and support a low-carbon economy.
When evaluating potential investments, consider the following criteria:
1. Scalability: Can the investment be scaled up to have a significant impact on mitigation?
2. Innovation: Does the company offer innovative solutions or technologies that can drive meaningful change?
3. Market Potential: Is there a large market opportunity for the company's products or services?
4. Financial Viability: Can the company achieve financial sustainability and return on investment?
5. Social Impact: Will the investment have a positive social impact, such as creating jobs or improving air quality?
By focusing venture capital investment in these areas, you can help drive innovation, scalability, and financial viability while maximizing impact on mitigation efforts.CDR technologies do not get much of a mention in AR6 Figure 4.4 or in the broader report. This is mostly a reflection of the fact that they have miniscule impact in the near term time frame up to 2050. The key reference to CDR in AR6 is in the Overshoot section and it is worth quoting in full:
Overshoot of a warming level results in more adverse impacts, some irreversible, and additional risks for human and natural systems compared to staying below that warming level, with risks growing with the magnitude and duration of overshoot (high confidence). Compared to pathways without overshoot, societies and ecosystems would be exposed to greater and more widespread changes in climatic impact-drivers, such as extreme heat and extreme precipitation, with increasing risks to infrastructure, low-lying coastal settlements, and associated livelihoods (high confidence). Overshooting 1.5°C will result in irreversible adverse impacts on certain ecosystems with low resilience, such as polar, mountain, and coastal ecosystems, impacted by ice-sheet melt, glacier melt, or by accelerating and higher committed sea level rise (high confidence). Overshoot increases the risks of severe impacts, such as increased wildfires, mass mortality of trees, drying of peatlands, thawing of permafrost and weakening natural land carbon sinks; such impacts could increase releases of GHGs making temperature reversal more challenging (medium confidence).
The larger the overshoot, the more net negative CO2 emissions needed to return to a given warming level (high confidence). Reducing global temperature by removing CO2 would require net negative emissions of 220 GtCO2 (best estimate, with a likely range of 160 to 370 GtCO2) for every tenth of a degree (medium confidence). Modelled pathways that limit warming to 1.5°C (>50%) with no or limited overshoot reach median values of cumulative net negative emissions of 220 GtCO2 by 2100, pathways that return warming to 1.5°C (>50%) after high overshoot reach median values of 360 GtCO2 (high confidence). More rapid reduction in CO2 and non-CO2 emissions, particularly methane, limits peak warming levels and reduces the requirement for net negative CO2 emissions and CDR, thereby reducing feasibility and sustainability concerns, and social and environmental risks (high confidence).
The ramifications of these two paragraphs could not be more profound. There’s a parallel with Covid-19 where we intervened to “flatten the curve” rather than “let the bodies pile up” is strong. In this case we don’t seem to be collectively acting to prevent the worst impacts from being realised. Andreas Malm’s recent book Overshoot explores how we have arrived at a point where those in authority appear to have given up on concerted effort to disinvest in fossil fuels. Overshoot was written before COP29. Events there merely confirmed its baleful vision of the future:
The overshoot conjecture arises out of the failure to fulfil the historical task of asset stranding, or the political destruction of fossil capital. But sooner or later, one way or another, some anthropogenic forces will have to face up to this task and execute it. The alternative is unlimited combustion, or the maximum realisation of the potentials for catastrophe.
Logic admits of three further alternatives to the political destruction of fossil capital: adaptation, carbon dioxide removal and geoengineering, alone or in combination. It follows that fossil capital will do everything in its might to make these stand in for its own downfall - in other words, to turn overshoot management into a substitute for mitigation. Unlimited combustion would then proceed under thin covers.
Note my highlights in bold. We focussed on CDR technologies (aka Negative Emission Technologies or NETs) in this post as part of surveying the mitigation landscape. The other two topics will be covered in more depth next time. Many of us who work in technology first roles are working on mitigations that involve energy decarbonisation and electrification but that is only part of the picture. As technologists we arc towards the techno-utopian, the belief that technology (which largely brought us here) can be relied upon to save the day. What Malm and tools such as En-ROADS make clear is that while technology is critical, we cannot combat climate change through renewable energy alone. We need to ramp up nature-based solutions like reforestation and we need policy instruments around carbon pricing to accelerate the move away from fossil fuels. CDR is going to be a massive growth industry in decades to come. However, current progress and scale remains a long way off where we need to be to avoid 1.5-2C temperature rise by 2100. As David Ho, Professor of Oceanography at Hawaii puts it:
All I seem to think about these days is CDR – carbon dioxide removal. Normally I say that it doesn't make sense to deploy [CDR] until we decarbonise drastically, because it is useless when we're still emitting more than 40bn tonnes of CO2 every year…But if we don't start now, we might not be able to scale [up] fast enough.
I'm thinking about that because it does have to go from something that most people have never heard of, to the biggest thing we've ever done, in a short time.
It's really hard to know with the new administration in the US … But the overarching story, of course, is that we're emitting more CO2 – things are getting worse, and we're not doing anything about it. And whether that remains the biggest story or not, I don't know, because it seems like everything that we do is small compared to the fact that we don't do anything about the continued use of fossil fuels.
We will dive deeper into adaptation and geoengineering in the next post. We’ll explore why Malm’s grim calculus around overshoot increases the likelihood of radical intervention with a specific form of geoengineering called solar radiation management (SRM) to buy time while we try to overcome objections in the energy system transition away from fossil fuels.
