Prioritizing the Forces of Nature
Gerald Ratzer Professor Emeritus, McGill, Montreal
Introduction
This article demonstrates howAI can be a useful tool in stimulating a discussion on a topic where not all the facts are known or knowable. The ideas supplied by the author in the form of questions to AI show how this can trigger a more in-depth analysis of complex topics. The concept here is to examine a range of possible items and haveAI expand the list, comment on the relevance of the data mined, sort the items in rank order and then do a form of factor analysis to see how the most important items interact with each other and form clusters. As the title points out, the article will concentrate on the Forces of Nature, but the methodology can be applied to other areas where exact data is not available. Using metrics based on physical characteristics, further discussion is possible that goes beyond just unprovable opinions. This is a step to a more data-driven approach where the Scientific Method cannot be applied.
Artificial Intelligence (AI) Systems
There are many differentAI systems which are freely available. Copilot (Microsoft) was used in this exercise, butAI is also available from all major technology companies, like Google (Gemini),Apple Intelligence and companies that specialize in justAI – Grok, Claude, ChatGPT (OpenAI), etc. Most of these systems have a free starter version and a more advanced, more capable version available on a subscription basis.
From your favorite browser, type in one of theAI names above, and you will most likely get connected to the free edition, which is sufficient for initial learning, if you have not already tried this out. Acouple of guidelines are to provide some context of what area you would like to dig into. Also, you should realize that an AI system will search the Internet for many sites that contribute to the response. The real power ofAI is the analysis it does on the relevant websites’information. This analysis is guided by the training data and process used to construct the complete AI system and chat interface.
I find that if the context and data are well-defined, the responses are accurate and useful. However, mostAI systems have some small print with each conversation saying thatAI systems do make mistakes and the information provided may be inaccurate. These systems are very complex and rely on huge databases, so the warning of inaccuracy is quite valid. Some users have called these inaccuracies “lies”, while the output is a reflection on the designers, programmers and the accuracy of the databases used [2]
Methodology
This study is aimed at generating a list of the most important sources, sinks and processes that are responsible for our daily weather, and when averaged over 30 years, for our climate. This process led to a classification of the sources, sinks and processes, followed by the creation of a metric based on physical characteristics. This was followed by a ranking in the first instance, and then an analysis of the interaction between the most important items. So, this is a scientific way of removing opinion and political bias and relying on the best modern technology to categorize the Forces of Nature.
Generating the line items
To start the conversation withAI, the context was briefly described, and a short list of items was proposed. The first query would look like this –
“I am interested in exploring all the different sources, sinks and processes that contribute to our weather and climate. This would include the Sun and variations in solar radiation; interaction with the atmosphere, the land and ocean surfaces; the dynamics and energy flows in the air and oceans; plate tectonics and geothermal energy. Please add to this list to form a more complete set of factors for our weather and climate”.
So, this is a polite request toAI – and not just a simple one-site query.
The response took about 20 seconds for Copilot to generate, and this is shown at the beginning of reference [1]. Here is a shortened version of the 26-line items that were suggested.
External Energy Sources
1. Solar Radiation – Primary driver of weather and climate.
2. Milankovitch Cycles – Long-term changes in Earth's orbit and tilt affecting climate.
3. Cosmic Rays – May influence cloud formation (still under study). …
Internal Earth Energy Sources
5. Geothermal Energy – Heat from Earth’s interior, influencing localized weather (e.g., geysers). …
Oceanic Processes
8. ENSO (El Niño–Southern Oscillation) – Major driver of global weather variability.
9. Ocean Currents – Redistribute heat across the globe (e.g., Gulf Stream). …
Atmospheric Dynamics
12.Hadley, Ferrel, and Polar Cells – Large-scale circulation patterns.
13.Jet Streams – Fast-moving air currents that steer weather systems.
14.Conduction and Convection – Transfer of heat within the atmosphere and between surfaces. …
Surface and Land Influences
20.Albedo Effect – Reflectivity of surfaces (ice vs. forest vs. urban areas).
Anthropogenic Influences
24.Greenhouse Gas Emissions – Trap heat and alter the climate. … 26-line items in total.
This is now a longer list and includes many new items beyond the few that were suggested. The next important step is to do some Critical Thinking and review the list to see that they all make sense. I did add a few more items to achieve a total of 31 sources, sinks and processes. One useful addition fromAI, that I did not ask for, is the classification and grouping of the different variables that are obviously related.
Physical-based Metric
This, in turn, led to a request for a Metric or measure that can be applied to a wide range of processes, which are not all collected data in the same units. The most obvious measure might be energy and energy flows, but this cannot be applied to all 31 variables. Yes, incoming solar radiation, atmospheric and oceanic flows all involve units like watts/metre2 , but albedo (0 to 1), and topography are in completely different units. Some would argue that we are comparing apples and oranges, but the proposed metric is based on four reasonable factors. Again, the metric is generated by AI.
1. Energy flux
• Measure in W/m² (or kJ/kg for phase-change processes)
• Normalize: factor_value / max_value_across_all → 0–1 scale.
2. Spatial scale
•Assign: global = 1.0, hemispheric = 0.75, regional = 0.50, local = 0.25
3. Temporal scale
•Assign: multi-decadal+ = 1.0, annual = 0.75, seasonal = 0.50, daily/hourly = 0.25
4. Direct impact
•Assign: direct forcing = 1.0, partially indirect = 0.50, secondary = 0.25
Once each of the 31 factors is scored (0–1) byAI on these four axes (all in the same units now), multiply by the normalized weights (0.35, 0.25, 0.20, 0.20) and sum to get a composite importance index.
I went over the four factors in the Metric and do think they are reasonable. What is open to question, in my mind, is the relative weighing of each factor. My own personal choice would be for the first two (.35 and .25) to be higher (.4 and .3), but I did not change them, so my bias is excluded from the further calculations and the ensuing discussion.
Ranking the Forces of Nature
Given the set of 31 weather variables and the proposed physics-based metric, the AI system was asked to rank the variables. For theAI system, this is now a simple process of sorting the 31 items based on the relative importance metric for each variable. Again, this is decided byAI, based on essentially three queries to
generate a list of variables, classify their data units and sort them based on a welldefined physical metric. Here is the beginning and end of the full list, which is in reference [1].
Here are the first 5 energy sources and physical processes that impact Earth's weather and climate, along with a refined importance based on physical properties such as energy flux, spatial and temporal scale, and direct impact on atmospheric dynamics:
Table: Weather and Climate Influencing Factors
# Factor Importance Score (0–100) Justification
1 Solar Radiation 100 Primary energy source driving Earth's weather and climate systems.
2 Water Cycle 90 Fundamental for precipitation, humidity, and energy exchange.
3 Cloud Formation and Dynamics 85 Major influence on radiation balance and precipitation.
4 Ocean Currents 80 Redistribute heat globally, affecting regional climates.
5 ENSO (El Niño–Southern Oscillation) 75 Causes significant interannual climate variability.
. . 27 Cosmic Rays 22 May influence cloud formation (still under study). 28 Lunar Gravitational Effects 20 Tides and subtle atmospheric influences. 29 Radiative Cooling 18 Earth emits infrared radiation to space.
30 Evaporative Cooling 17 Occurs when water evaporates from surfaces.
31 Cloud Radiative Effects
15 Clouds can cool the surface by reflecting sunlight or warming it by trapping heat.
I have no problem with the first 5 items, but I was surprised to see the tail end of the list. I would have said that Evaporative Cooling is a vital part of the Water Cycle, which in turn is part of the Cooling Cycle, that moves heat energy from the Tropics to the Poles. The Water Cycle is number 2 on the list – so a discussion could take place as to whether item 30 should really be in the list at all, as it is a subset of item 2. I would be quite happy if item 30 were dropped.
Analysis of Interactions
The output ranking is interesting, but can we do a further analysis of the information generated? In statistics, several different techniques can be used when you have a matrix of different variables and collected data, usually in the form of a matrix. These work when all the data is in the same units. The classic techniques go by names such asANOVA(Analysis of Variance), PCA(Principal Component Analysis), and FactorAnalysis. Since the data was not in the form to allow these classic methods to be used, theAI was asked for the closest approximation to PCA by using 2- and 3-way interactions among the variables. This methodology generates many combinations, so theAI system limited the printed output to the first 20, highest-scoring interactions. The start of this table is shown below.
Top 5 Two-Factor Interactions
For the 3-way interactions, the table starts with Top 4 Three-Factor Interactions
4
These tables show the ideas and methodology, but are rather repetitive. This led to the nextAI request.
Clustering
Here, the idea is to summarize the data and look for clusters that would rate the main research areas, based on thisAI analysis.
Cluster 1: Solar–Water–Atmosphere Interactions
Shared Themes: Solar Radiation, Water Cycle, Cloud Formation and Dynamics
Insight: These interactions reflect how solar energy powers the water cycle, which in turn drives cloud formation and large-scale atmospheric circulation.
Cluster 2: Ocean–ENSO–Solar Interactions
Shared Themes: Solar Radiation, Ocean Currents, ENSO
This cluster emphasizes ocean-atmosphere coupling and its role in climate variability.
Insight: These combinations highlight how solar energy interacts with oceanic processes like ENSO and thermohaline circulation to modulate global climate patterns.
After all this, theAI system volunteered to produce a visual chart to show the clusters and their main interactions.
Conclusions and Discussion
The process used and the ease of generating this output are good examples of how AI can assist a better understanding of the different Forces of Nature and how they interact. You may not agree with the rank order of the line items or the details of the physical metric, but it does explore all the main areas and adds details that may not have occurred to you. I think the cluster outcome is valid and can help determine the main areas of future climate research and funding.
For instance, it should be noted that CO2 and radiative LWIR (Long Wave Infrared Radiation) are minor topics in the less than 1% category, as commented by Will Happer [3]. Reference [4] concurs to say CO2 warming is negligible, and close to the end of the rank list for good reasons. Taking this a step further, we could say
that references [3] & [4], along with this study, suggest that the ECS (Equilibrium Climate Sensitivity) should be on the low side and close to zero!
Also, this analysis points out that the IPCC and the Mainstream Media have got their facts and message wrong and are barking up the wrong tree by attacking CO2 warming and supporting Net Zero. Net Zero ignores the benefits of CO2 (more is better for bio-growth; a contributing factor to record world crop yields) Net Zero’s dependence on unreliable energy generation, like wind and solar, has no basis for the huge implementation cost, the grid stability impact (e.g. the Iberian blackout) [5], and produces immeasurably small benefits, based on thisAI study.
References
[1] Chat with Copilot.
[2] Podcast with Jonathan Cohler https://jpands.org/vol29no3/cohler.pdf
[3] Recent paper by W. Happer and W. van Wijngaarden
“For example, “instantaneously doubling” CO2 concentrations, a 100% increase, only decreases radiation to space by about 1%.”
https://scienceofclimatechange.org/wp-content/uploads/SCC-2025-vWijngaardenHapper.pdf
[4] H. D. Lightfoot, G. Ratzer Earth’s Temperature: The Effect of the Sun, Water Vapor, and CO2
2022-11-23 https://set-publisher.com/index.php/jbas/article/view/2425
“The effect of CO2 on warming the Earth has a negligible increase in Earth’s temperature.”
[5] Net Zero in the UK and the US.