Geo Alpha Climate Research Strategy


Key Climate Strategy Goal: research technologies to achieve 1C cooling by 2100

 


Why we’re doing this

With roots in Earth Observation satellite system design, Geo Alpha exists to carry out applied machine learning research at the edge of the opportunity frontier to price and manage risk in noisy, complex systems. Global climate is the biggest such system there is.


Mainstream climate modelling (CMIP6) explores futures by pairing socioeconomic pathways (SSPs) with radiative forcing levels – e.g. SSP2-4.5 (“middle-of-the-road”) and SSP3-7.0 (“high emissions”). These scenarios, run by national labs and universities, feed into IPCC assessments and national risk planning.


Policy and emissions today still leave a very real chance of 2.5–3°C+ warming this century, with recent annual temperatures already brushing the upper edge of scenario ranges (SSP5-8.5). Even if that “worst edge” only has a few percent probability, it is the kind of planetary tail risk that deserves a hedge.


We are not a climate lab. But we are good at:

  • Building and stress-testing complex dynamic simulations
  • Working with noisy, high-dimensional data
  • Making disciplined decisions under uncertainty


The Geo Alpha Climate Research Strategy is about directing a share of our trading profits and technical skills toward planet-scale risk reduction, without confusing that with our trading book.

 


Impact of 1C cooling by 2100 on GDP and Mortality


PCC AR6 WGI (Table SPM.1) gives assessed global mean warming (vs 1850–1900) for these scenarios:


CMIP6 pathway    2050 warming (2041–2060) 2100 warming (2081–2100) If held 1°C cooler in that period

SSP1-2.6                              1.7°C                                              1.8°C                                              0.7°C / 0.8°C

SSP2-4.5                              2.0°C                                              2.7°C                                              1.0°C / 1.7°C

SSP3-7.0                              2.1°C                                              3.6°C                                              1.1°C / 2.6°C

SSP5-8.5                              2.4°C                                              4.4°C                                              1.4°C / 3.4°C


 


GDP SAVED WITH 1C COOLING  (Howard & Sterner-style level at 3°C)

 

CMIP6 Pathway Cumulative to 2050 (US$ Tn)         Cumulative to 2100 (US$ Tn)

SSP1-2.6                                                      110                                                             662

SSP2-4.5                                                      111                                                            1,195

SSP3-7.0                                                      191                                                            1,898

SSP5-8.5                                                      220                                                            3,364




2100 POPULATION UNDER CMIP6 Pathways


2100 population differences between SSP3 - 7.0 and SSP5 - 8.5 pathways (there is 1C difference between the two if we start now)


CMIP6 Pathway         2100 warming (°C)   Population (bn)

SSP3 - 7.0                                 3.6                                14.1

SSP5 - 8.5                                 4.4                                7.79


Population delta (1C Cooling): 6.31bn




Our focus: albedo and “light-touch” climate hedging


At the most basic level, almost all Earth’s energy comes from the sun. How much of that energy gets absorbed versus reflected is controlled by albedo – the reflectivity of surfaces and clouds.

  • Dark ocean absorbs most incoming sunlight.
  • Ice and snow reflect most of it.


This creates a powerful feedback loop:

Less polar ice → lower albedo → more absorbed heat → even less ice
More polar ice → higher albedo → more reflected heat → more ice


Crucially, this feedback works both ways. In principle, carefully increasing reflectivity in key regions – especially the Arctic – could nudge the system toward:

  • Cooler polar summers
  • More sea-ice formation
  • Knock-on moderation of some global warming impacts


But it could also:

  • Disrupt monsoons
  • Change storm tracks
  • Create winners and losers between regions


Our Part 1 strategy is to treat albedo-based interventions as a candidate hedge and ask, with real data and real models:

Under what conditions, if any, could albedo-based interventions be safe, reversible, and governable – and under what conditions should they not be on the table at all?

We are particularly interested in whether modest, temporary albedo changes could help push temperatures back toward late-20th-century levels over the next 30 years, when combined with aggressive emissions cuts and carbon removal.

 


Part 1 – Simulation first

We are starting where the risk lives: in the models and the physics, not in hardware. Over the first 12 months we will fund a tight, focused simulation program with three components:


1. Quantifying “how much” cooling would even be needed

Using simple, IPCC-consistent climate models, we will:

  • Take several emissions futures:
  • A strong mitigation path
  • A “middle-of-the-road” path
  • A high-emissions SSP3-7.0-style path
  • Ask: How much additional negative radiative forcing (cooling) would be needed over the next ~30 years to bring global average temperatures back toward late-20th-century levels?
  • Impose safety constraints:
  • Avoid violent year-to-year swings in temperature
  • Avoid forcing levels that are obviously beyond what any realistic intervention could achieve

This gives us a numerical target: if the required cooling is massive or would cause extreme “termination shock” on exit, that alone is evidence against further pursuit.


2. Re-analysing what the world already knows

We then mine existing multi-model SRM and albedo experiments (GeoMIP and related work) to build a risk atlas:

  • What happens to:
  • Indian and West African monsoons
  • Sahel rainfall
  • ENSO, storm tracks and extremes
  • Under different patterns and strengths of solar/radiative modification?

We’re not reinventing decades of community work; we’re standing on it, analysing it with our own tools, and looking specifically at “safe, reversible, governable” criteria.


3. New targeted Arctic albedo experiments with leading climate centres

Finally, we commission 1–2 leading Earth System Modelling groups (e.g. UKESM, NCAR, Max Planck) to run a small, clearly-defined set of new experiments:

  • Use standard future scenarios as baselines:
  • SSP2-4.5 (mid-century)
  • SSP3-7.0 (mid-century)
  • Plus a late-century high-forcing state (SSP3-7.0 or SSP5-8.5)
  • Add idealised Arctic albedo increases, tuned to represent:
  • A moderate cooling (~−0.5 W/m² globally)
  • A stronger cooling (~−1.0 W/m² globally)
  • Then measure:
  • Arctic sea-ice and temperature changes
  • Global mean temperature
  • Regional rainfall, especially monsoon regions and the Sahel
  • Large-scale circulation (jets, storm tracks, ocean overturning)


Rather than modelling specific technologies we are asking the question:


If the Arctic were made more reflective by a known amount, what does the Earth system do in response under SSP2-4.5, SSP3-7.0 and even more extreme warming?


The full technical specification for participating modelling centres – experiment setup, variables, file formats – is being published as open documentation in our ALMA WP3 specification on GitLab .

https://github.com/GeoAlpha-ClimRes/Project-ALMA

 


Open science, not private edge

This is not about building a private “house climate model” to trade off.

We commit to:

  • Open methods and code
  • Our prototype Arctic albedo simulator is already public (see the “Climatrix” notebook on our GitHub: https://github.com/GeoAlpha-ClimRes/Climatrix).
  • The ALMA WP3 technical specification is open on GitLab; future analysis code will follow.
  • Open data, with reasonable delays
  • Model outputs and derived diagnostics will be released, subject to partner agreements, after initial analysis and publication.
  • No trading on non-public climate research
  • Geo Alpha will not use non-public outputs from this program as trading signals.
  • We are doing this as philanthropic risk research, not as an alpha stream.
  • Mitigation-first
  • Any discussion of albedo interventions will explicitly state:

They are not substitutes for rapid emissions cuts and carbon removal, at most a conditional back-up if society fails to decarbonise fast enough.

 


Part 2 – Laboratory work and engagement (strictly conditional)

Only if the simulation work suggests there might be a narrow, potentially beneficial and tolerable regime will we consider:

  • Small-scale laboratory and materials studies (e.g. how reflective coatings behave, what monitoring might look like).
  • Governance and ethics work:
  • Legal frameworks
  • Liability and justice issues
  • Roles for climate-vulnerable countries and indigenous Arctic communities
  • Stakeholder engagement:
  • Independent scientists
  • Civil society and environmental NGOs
  • Policy communities


We will not fund or advocate outdoor climate intervention trials without:

  • Broad, legitimate international governance
  • Independent scientific leadership
  • Transparent monitoring and public consent


Our default expectation is that Part 1 (simulation) may well conclude that albedo-based approaches are either:

  • Too risky and inequitable to pursue, or
  • Only conceivable in extremely narrow, conditional ways that require years of public governance work before any field activity.


Either way, that is knowledge worth having.

 


Where this goes

At the end of the first 12-month phase we expect to have:

  • A quantified answer to:
  • “How much cooling would we be asking albedo interventions to deliver under different futures?”
  • A risk map that highlights where regional harms are likely to show up first.
  • A clear recommendation on whether albedo-based climate engineering should:
  • Be ruled out,
  • Be kept under watching brief only, or
  • Justify a larger, openly-governed, multi-year research program led by the public science community.


If the science and governance signals say “no” – that albedo is not a safe or governable hedge – we will treat that as a successful outcome and redirect future climate funding toward mitigation, adaptation and other forms of resilience.



If they say “maybe, but only under strict conditions”, we will make any next steps public, slow, and governed – not private, fast, and unilateral.

 



Vasily Koledov
CEO, Geo Alpha


For scientific and technical enquiries, please see the ALMA WP3 specification on GitLab or contact us at clim@geo-alpha.com