Greenhouse Effect Experimental Designs

[BrianG]

To answer these questions, you need

 

a) to present a quantified model – some collection of number with units, even highly approximate, and at least simplistic mechanical models relating them –

 

;) to provide links or citations to descriptions of such work by others

 

that suggest that “air resistance on moving vehicles” or “the shape, height and quantity of buildings” has an appreciable effect on climate,

 

or

 

c) in good faith, ask others for help doing (a) or (:D.

 

If you’re not doing one of these, you’re

 

d) making unsupported claims in the form of questions.

 

In asking these two questions, Brian, are you doing © or (d)?

 

 

You miss the point, I am doing neither. I am answering freeztar's "fossil fuel experiment" argument. I am arguing that observation alone isn't sufficient to find cause.

 

I am not proposing or supporting the claims that more buildings or vehicles cause climate change, I am merely demonstrating that coincidence alone, without experimental test, can not eliminate unknown causes for climate change. I am looking for the most ridiculous coincidences possible, there are more buildings shaped like pyramids, now, than ever before. This correlates with increased atmospheric CO₂. Without experimental tests, there is no way to eliminate coincidence.

 

Experimental testings is the sine qua non of science.

[Zythryn]

Repeated random trials of releasing GHG and capturing GHG, looking for the smallest measurable temperature change..

 

Repeated random trials without accounting for other variables will give poor results (if any).

 

You need to measure what wavelengths CO2 (and other gases) absorb and which they are transparent to.

This is easy to do in a labratory setting.

You can also measure other physical properties of CO2, energy wavelengths coming ifrom the sun, energy wavelengths that are reflected from different parts of the earth's surface.

 

I don't see what results you expect to get from your random tests. Would you at least hold the tests in the same location during the same time of each year?

Would you even measure other variables in your tests, or just hope that random testing evened everything else out over time?

[BrianG]

Repeated random trials without accounting for other variables will give poor results (if any)...

I don't see what results you expect to get from your random tests. Would you at least hold the tests in the same location during the same time of each year?

Would you even measure other variables in your tests, or just hope that random testing evened everything else out over time?

 

The purpose of field tests on man made greenhouse gases is to find how much greenhouse gas causes how much warming. There may be climate feedbacks that will either amplify or reduce temperature changes. Only field tests can find those feedbacks, anything else is just coincidence, the mother of superstition: Coincidence - Wikipedia, the free encyclopedia

 

If there are no results from large scale greenhouse gas emission and capture experiments, that might mean that climate change mitigation isn't feasible. Due diligence would seem to require testing a climate mitigation strategy, before using it.

[Zythryn]

Brian, again, would you measure other variables that may affect temperature?

Your test is far less useful than the tests we have already done as it makes no attempt to isolate variables.

Can you at least give us recommended time frames of your suggested tests?

[BrianG]

Brian, again, would you measure other variables that may affect temperature?

Your test is far less useful than the tests we have already done as it makes no attempt to isolate variables.

Can you at least give us recommended time frames of your suggested tests?

 

The purpose of this thread is to propose and criticize experimental tests for climate change mitigation. My first post has three experiments on CO₂'s greenhouse effect on temperature. I've repeated, random emissions and captures of GHG would help cancel the effects of extraneous variables. The time frame would be, as long as it takes to show the smallest measurable climate change, then the experiments could be repeated as often as necessary.

 

What tests have you already done?

[Zythryn]

Brain, how do you decide that you have enough random data to eliminate other variables?

The tests I am aware of are those done in labs under controled circumstances.

For example, as I mentioned before, determining the wavelength absorbtion properties of CO2.

Measurement of the amount of CO2 in the atmosphere over the decades.

Backwards testing of climate models.

Icecore testing to determine the amount of CO2 in the air previous to measurements.

Measurements of the irradiation from the sun.

 

In your proposed test, you have no mechanism for even measuring other variable, therefore no method of accounting for them. You may have trends that last years or decades or even centuries, yet since you don't measure for them, you have no idea of how they may or may not affect your results.

You need a combination of real world experiments as well as tightly controled labratory tests.

[Zythryn]

...The time frame would be, as long as it takes to show the smallest measurable climate change, then the experiments could be repeated as often as necessary.

 

Ok, so let's say you have performed your burn of coal to release a huge amount of CO2 in the air and in 3 weeks you have an increase of global temperatures of .1 degrees celsius. How do you determine if that is a result of your experiment or something else?

[BrianG]

Continue to observe as the background levels of CO2 predominate, and if there's a low probability that we can attribute the warming to the CO2, try it again. If 90% of the times we burn a coal mine, and vent the CO2 then three weeks later we see warming, we've got a measurement for how much CO2 causes how much warming in how much time. We'll also see how other climate factors amplify or reduce the effect.

[TheBigDog]

The purpose of this thread is to propose and criticize experimental tests for climate change mitigation. My first post has three experiments on CO₂'s greenhouse effect on temperature. I've repeated, random emissions and captures of GHG would help cancel the effects of extraneous variables. The time frame would be, as long as it takes to show the smallest measurable climate change, then the experiments could be repeated as often as necessary.

 

What tests have you already done?

I had the good fortune a few years ago to go through some training on experimental design. In particular on multiple factorial experiments. In my training course I was on a team given a model airplane and a launching mechanism. We had to determine the best way to launch the plane and make it hit a target placed in the room. We were given a limited budget for test flights and had to record very carefully the conditions of each flight. At the end of the week we had a contest among all the teams where a target was placed randomly in the room, and we needed to use our understanding of the flight characteristics and launcher to hit the target. So the object was to do controlled experiments where there were many many factors in play.

 

Which factors were important? Where were noise? How could you neutralize some while controlling others? What were the settings (low end, high end, increment) critical to each factor to see in influence on the experiment? Cap it all off by having to work as a five man team and agree on how all of this happens.

 

In our experiment we had a rubber band launcher. It needed to be secured. it could be angled up and down, and turned left and right. You could add or remove rubber bands. you could adjust how far you pulled the rubber bands. You needed to know the elevation of the launcher relative to the target. Other factors just in the launch were obstructions, smoothness of the launch, tension on the left and right sides of the stretched bands, release of the plane from the launcher.

 

Once the plane was actually launched controlling (or at least understanding) the flight characteristics was the next challenge. Velocity, wing placement, wing trim, tail placement, tail trim, angle of attack, center of balance, total weight... it goes on and on.

 

We were given a budget to take 8 test flights. With those eight flights we had to tune both the plane and the launcher. We were later given another 8 test flights. So, how do you go about determining the main factors?

 

First we built the plane following the instructions and did some non-flight experiments to make sure it was balanced and such. We made a list of all these things and discussed which ones we thought we needed to vary, and which ones we would keep static to control.

 

The problem is that adjusting one factor will show a change in performance, but adjusting multiple factors shows interactions in factors, some which confound each other in ways that are counter intuitive; such as when a is increased c increases, when b is increased c increases, but when a and b are increased c decreases (this is the simplest example). You can have multiple factor reactions where when a is between 2 and 4 it neutralizes factors b and c, so no matter how much you adjust them they have no impact. But when a is above 4 or below 2 suddenly a and b are significant factors again.

 

With the airplane experiment the simplest solution is to remove as many factors as possible; crush the plane into a ball, wrap it in tape, and use the launcher like a catapult. It removes the majority of the aerodynamic factors of flying the plane giving you predictable flight characteristics, and makes the launcher itself the only set of variables to deal with.

 

Unfortunately you cannot simplify the atmospheric variables the same way.

 

Lets look at the large factors involved in Global Climate:

 

Heat of the sun

Mean distance from the sun

Seasonal angle of the sun to the poles

Earth's magnetic field

Polar ice cover

Ocean temperature

Ocean currents

Seasonal snow cover

Air temperature

Cloud cover

Inland surface water

Major volcanic activity

Forests and vegetation (seasonal and perennial)

Forest and brush fires

Atmospheric storm activity

Atmospheric content

Surface heat sinks

 

Every one of these has deeper elements that we can go into. Within and between these elements are multiple factors that interact with all the other factors, some complimenting, some negating, some confounding. What you propose is to take a single element of Atmospheric Content and through random releases and samplings of green house gasses determine with certainty what the impact of that one factor was on the greater picture; and you are hearing everyone say "It won't work, its too complicated." Because its too complicated.

 

And that "too complicated" is at the heart of my own problem with crediting CO2 for the observed temperature changes over the past century. Yes, it is warming. Yes, CO2 is a Greenhouse Gas with known properties. Yes, CO2 is at levels not seen in our recorded history. But that does not mean that the CO2 is the cause of the heating, because there are just too many other factors to consider. Especially when you look at how short term temperature swings are seen in our recorded history that would make our current trend unremarkable and unnoticed, except for the CO2 levels.

 

Is it because I don't want CO2 to be the culprit? No, it is because the evidence I have seen does not convince me. Because the climate is more complicated that just turning up or down the CO2. The climate is more complicated than what is measured by thermometer over the past 150 years. The climate is more complicated than our presumption of the steady state of glacial ice. The climate is more complicated than air temperature over a ten year period, or a hundred year period.

 

Climate models try and make logical sense out all of this complication and allow us to run simulations to try and predict the future based upon the past. Every year we learn more and refine the models. Then we run them again getting answers from the past that give us a closer approximation of the present, which gives us greater certainty in their forecast of the future. With each passing year we reduce the margin of error, and uncover more questions to be answered which bring us closer to a perfect model (an unattainable task).

 

Given all of that, what do you see at the purpose of a single variable testing?

 

Bill

[Zythryn]

OK Brian, so let's say you do this 20 times.

18 of 20 times you get an increase. The other 2 times you get decreases. However, the 2 times were after a major volcanic excplosion.

What conclusions would you draw?

[BrianG]

Repeated, random capture or release of CO₂ will [math]\text{\sout{cancel}}[/math] minimize extraneous variables. All things being equal over time, the only significant change will be the quantity of CO₂ emitted or captured.

 

If you can repeatedly and randomly emit a known quantity of CO₂ and subsequently measure a temperature increase, then you know how much carbon dioxide causes how much warming. This will help us test computer climate models, determine the climate benefits of CO₂ emission restrictions and teach good science.

[TheBigDog]

All things being equal over time, the only significant change will be the quantity of CO₂ emitted or captured.

You mean... balance?

I contend that this balance model is oversimplified, the amount of carbon dioxide in the air is constantly changing, there is no balance. It's not even as useful as the Bohr atomic model, can anyone name a practical application for the carbon dioxide balance model?

Interesting. There is no practical way to perform your proposed experiment and state with confidence that the results are due to the CO2 changes that you were making. There are too many other variable in play. Saying they will be equal over time will not make it so. I would suggest reading up on the execution of multiple factor controlled experiments and observational studies.

 

Bill

[BrianG]

Randomization may not be a well known element of experimental testing. I provide the following:

 

Randomized experiment - Wikipedia, the free encyclopedia

 

"Randomization reduces bias by equalising other factors that have not been explicitly accounted for in the experimental design (according to the law of large numbers). "

 

Randomized block design - Wikipedia, the free encyclopedia

 

"Blocking to "remove" the effect of nuisance factors

For randomized block designs, there is one factor or variable that is of primary interest. However, there are also several other nuisance factors."

[Essay]

...and in only several generations we can begin to see.... :shrug:

 

Blocking to "remove" the effect of nuisance factors

:lol:

 

Are you talking about "nuisance factors" like solar irradiance, clouds, ocean currents, ice caps, and hurricanes; or about somebody in particular?

 

~ :D

[BrianG]

Randomized block design

From Wikipedia, the free encyclopedia

In the statistics theory of the design of experiments, blocking is the arranging of statistical unit in groups (blocks) that are similar to one another. Typically, a blocking factor is a source of variability that is not of primary interest to the experimenter. An example of a blocking factor might be the sex of a patient; by blocking on sex, this source of variability is controlled for, thus leading to greater precision.

[Essay]

Blocking to "remove" the effect of nuisance factors

Thanks for that one.... :lol:

===

 

I also thought of you when I caught this program over the weekend.

 

You might be interested in this talk on the testing of medical devices (or the lack of testing... 80-85%!) for human use.

ResearchChannel - Prescription for Change at the FDA: A View from the Other Washington, Part 1

 

Professors at the University of Washington ask: Does the Food and Drug Administration need more rigorous reviews and trials before approving drugs and devices? Should the agency change the process for evaluating safety and effectiveness after products hit the market? What are the political and scientific forces that shape the context for FDA decision-making and how can the clinical and public health communities be included in the discussion?

 

Lots to learn.... on the limitations of "experiments."

 

~ :shrug:

[Zythryn]

Brian, the problem is 'all others things' Don't remain equal.

You need to measure the changes in the other variables and model their effects on the climate.

Your method is great for statistics where you have very large groups (I would guess 1000s to 'block out' extraneous variables).

However, by performing the test once, you change the results of all the tests to come in the next 50-100 years. The reason for this is that is how long CO2 affects the atmosphere. So your second test either has to wait 50-100 years, or you need a way to reset the amount of CO2 in the atmosphere back to your original levels.

Other variables follow different cycles varying in length from daily cycles, to cycles that last 100s of thousands of years.

The more variables you have the more times you would need to run your test and AGAIN you would need to hold the rest of the world's CO2 emissions constant and return CO2 in the air to the start point at the start of each test.

 

Now, your method of statistical tests are used to determine the accuracy of climate models.

A model is taken, run many times with the average taken. The results are compared to the actual data. But outside of purely statistical tests that can be run hundreds or thousands of times, I don't see the usefulness of it.