5 ms·

That is a high level of consensus for a field where the predictive models (which have been getting published since the 1980s) are extremely bad at predicting th

by jrd79 5y ago

That is a high level of consensus for a field where the predictive models (which have been getting published since the 1980s) are extremely bad at predicting the future. Last time I looked into it was about 10 years ago, but at that time, the models were garbage at actually predicting anything. The climate is a complex system that is hard to model and hard to predict. Scientists in the field should admit this rather than overstating their certainty about any particular conclusion. And catch-all statements that "we all agree on X!" seem particularly suspect. The merit of a scientific assertion is established by the clear exposition of reproducible facts, not survey results. [EDIT] To be clear, the predictive power of the models is relevant to the question of causality because absent a controlled experiment (obviously impossible) or an exhaustive list of relevant statistical controls (also obviously impossible), establishing causality can only be done by developing an accurate (and thus predictive) model and using that model to measure causality.

machinerychorus 5y ago
It's not a prediction about the future, it's based on measuring what has happened. "Humans caused climate change". This statement implies nothing about what will happen in the future
- jrd79 5y ago
  The same approaches and models that are used for predictions are used to establish the attribution of cause. But while the latter cannot be tested, the predictive power of the models can be tested and is thus a suitable proxy for the power of the models to attribute cause accurately.
  - tzs 5y ago
    There are other ways to find causes. For instance CO₂ from burning fossil fuels does not contain ¹⁴C. CO₂ from other sources other than volcanoes does contain ¹⁴C. If you see the fraction of atmospheric CO₂ that has ¹⁴C going up more than can be accounted for from volcanoes you've got a smoking gun pointing to humans.
    - jrd79 5y ago
      Respectfully, you are missing the point. The claim being tested is whether humans caused climate change, not if they increased the CO2 concentration in the atmosphere. Explaining the relationship between the two is point of the model and cannot be tested for accuracy as you suggest.
      - tzs 5y ago
        We know that increasing the CO₂ concentration in the atmosphere increases temperature not because models say it does. We know because we can observe it. We can measure the intensity and wavelength distribution of incoming solar radiation. We can measure the effects various gases have on that radiation. We can measure what reaches the ground. We can measure the intensity and characterization of what is radiated back out from the ground. We can see that a significant fraction incoming radiation at wavelengths too short to be bothered by CO₂ gets radiated out from the surface at longer wavelengths that CO₂ absorbs, causing the CO₂ to act as an insulator slowing the transfer of radiated energy away from Earth which raises the equilibrium temperature on our side of it.
        jrd79 5y ago
        You present a plausible conceptual model of the core principle of a greenhouse gas. There are numerous questions between your conceptual model and explaining what fraction of the observed warming can be attributed to that effect. What other side effects do the elevated levels of CO2 have on the climate that might also affect the temperature? What are the second-order effects from the additional heat introduced to the climate by the greenhouse effect? Does it cause more clouds, and if so, what is the effect of those increased clouds on the radiation balance? Changes in precipitation? Vegetation growth? What other atmospheric constituents have had changes to their levels that might have also changed the climate? What other non-greenhouse-gas-related inputs effect the global mean temperature? Solar output? Deforestation? Can we accurately model those inputs so that we can sum up all contributing factors and see that it agrees with the observed record? What other natural sources of CO2 exist and how much is produced by those now and in the past? And then there are measurement accuracy questions: Do we have an accurate measurement of the global mean temperature? How far back is that measurement reasonably accurate? What sampling biases exist due to the siting of airports near large urban areas? Of the 5000 odd airport weather stations for which we have reasonably long historical records, more than 2000 are in the United States and large swaths of Africa and Asia have few to none. How does this affect our ability to reliably measure the global mean temperature? We have very limited measurements of the air temperature over the vast area of the world' oceans. How do we deal with the obvious uncertainties that exist due to this? Speaking of the global mean temperature, what exactly is the definition? You can't actual measure that directly, so you need to carefully define what it is, create a model to allow its computation, and then apply that model to the raw data. Here's an example: Global mean temperature at a given time is the instantaneous integral of the air temperature at 2 meters above the surface of the Earth. But we just have hourly sampled temperatures (that are not even at a consistent time within each hour) at a few thousand locations worldwide, so our model will be to correct the time offsets by linear interpolation with the temporally adjacent samples. To be able to sample the Earth uniformly for the numerical integral, we'll need to interpolate between triangulated stations and correct for elevation changes and proximity to large bodies of water (both of which are extremely complex models themselves). All this doesn't even deal with missing data (extremely common) from the station network. And what do we do over the ocean? How far back do we have accurate measurements of CO2 concentration in the atmosphere? Do we have a good measurement of the integrated reflective spectrum across the Earth's surface? All of the above combined requires a complex physical and numerical model that is very far from your toy example of pure measurement. Such a model is necessarily an approximation with unknown inaccuracies, and the only way I can think to test any proposed model is to wait and see how well it predicts the future. I'm not asserting that global warming is not happening or that it is not caused by humans. I've just not read or heard anything that answers that question in a convincing manner and I can think of no way to get to a firm answer than producing a model of the global climate that can accurately predict the future and to use that model to measure the fraction of observed warming that has been caused by human-produced greenhouse gasses (and other human-caused climate inputs). And some statistics on a keyword search of climate science papers by social science researchers is not going to convince me or anyone else capable of critical thinking.

by jrd79 5y ago

machinerychorus 5y ago

It's not a prediction about the future, it's based on measuring what has happened. "Humans caused climate change". This statement implies nothing about what will happen in the future

jrd79 5y ago
The same approaches and models that are used for predictions are used to establish the attribution of cause. But while the latter cannot be tested, the predictive power of the models can be tested and is thus a suitable proxy for the power of the models to attribute cause accurately.
- tzs 5y ago
  There are other ways to find causes. For instance CO₂ from burning fossil fuels does not contain ¹⁴C. CO₂ from other sources other than volcanoes does contain ¹⁴C. If you see the fraction of atmospheric CO₂ that has ¹⁴C going up more than can be accounted for from volcanoes you've got a smoking gun pointing to humans.
  - jrd79 5y ago
    Respectfully, you are missing the point. The claim being tested is whether humans caused climate change, not if they increased the CO2 concentration in the atmosphere. Explaining the relationship between the two is point of the model and cannot be tested for accuracy as you suggest.
    - tzs 5y ago
      We know that increasing the CO₂ concentration in the atmosphere increases temperature not because models say it does. We know because we can observe it. We can measure the intensity and wavelength distribution of incoming solar radiation. We can measure the effects various gases have on that radiation. We can measure what reaches the ground. We can measure the intensity and characterization of what is radiated back out from the ground. We can see that a significant fraction incoming radiation at wavelengths too short to be bothered by CO₂ gets radiated out from the surface at longer wavelengths that CO₂ absorbs, causing the CO₂ to act as an insulator slowing the transfer of radiated energy away from Earth which raises the equilibrium temperature on our side of it.
      - jrd79 5y ago
        You present a plausible conceptual model of the core principle of a greenhouse gas. There are numerous questions between your conceptual model and explaining what fraction of the observed warming can be attributed to that effect. What other side effects do the elevated levels of CO2 have on the climate that might also affect the temperature? What are the second-order effects from the additional heat introduced to the climate by the greenhouse effect? Does it cause more clouds, and if so, what is the effect of those increased clouds on the radiation balance? Changes in precipitation? Vegetation growth? What other atmospheric constituents have had changes to their levels that might have also changed the climate? What other non-greenhouse-gas-related inputs effect the global mean temperature? Solar output? Deforestation? Can we accurately model those inputs so that we can sum up all contributing factors and see that it agrees with the observed record? What other natural sources of CO2 exist and how much is produced by those now and in the past? And then there are measurement accuracy questions: Do we have an accurate measurement of the global mean temperature? How far back is that measurement reasonably accurate? What sampling biases exist due to the siting of airports near large urban areas? Of the 5000 odd airport weather stations for which we have reasonably long historical records, more than 2000 are in the United States and large swaths of Africa and Asia have few to none. How does this affect our ability to reliably measure the global mean temperature? We have very limited measurements of the air temperature over the vast area of the world' oceans. How do we deal with the obvious uncertainties that exist due to this? Speaking of the global mean temperature, what exactly is the definition? You can't actual measure that directly, so you need to carefully define what it is, create a model to allow its computation, and then apply that model to the raw data. Here's an example: Global mean temperature at a given time is the instantaneous integral of the air temperature at 2 meters above the surface of the Earth. But we just have hourly sampled temperatures (that are not even at a consistent time within each hour) at a few thousand locations worldwide, so our model will be to correct the time offsets by linear interpolation with the temporally adjacent samples. To be able to sample the Earth uniformly for the numerical integral, we'll need to interpolate between triangulated stations and correct for elevation changes and proximity to large bodies of water (both of which are extremely complex models themselves). All this doesn't even deal with missing data (extremely common) from the station network. And what do we do over the ocean? How far back do we have accurate measurements of CO2 concentration in the atmosphere? Do we have a good measurement of the integrated reflective spectrum across the Earth's surface? All of the above combined requires a complex physical and numerical model that is very far from your toy example of pure measurement. Such a model is necessarily an approximation with unknown inaccuracies, and the only way I can think to test any proposed model is to wait and see how well it predicts the future. I'm not asserting that global warming is not happening or that it is not caused by humans. I've just not read or heard anything that answers that question in a convincing manner and I can think of no way to get to a firm answer than producing a model of the global climate that can accurately predict the future and to use that model to measure the fraction of observed warming that has been caused by human-produced greenhouse gasses (and other human-caused climate inputs). And some statistics on a keyword search of climate science papers by social science researchers is not going to convince me or anyone else capable of critical thinking.