Is "Consensus" shifting?

[waitforufo]

 

just my opinion...

[iNow]

just my opinion...

 

So, in sum, the data I shared is valid, and clearly NOT wrong, despite your claim to the contrary. Thanks for conceding that.

[waitforufo]

So, in sum, the data I shared is valid, and clearly NOT wrong, despite your claim to the contrary. Thanks for conceding that.

 

When did I say that the data you shared in your post 64 was wrong? This is my first post in this topic.

 

I just think that you and those that create such plots have a rather alarmist view.

[iNow]

disregard

Edited July 16, 2009 by iNow
I messed up... =)

[insane_alien]

iNow, waitforufo != jryan

[bascule]

I just think that you and those that create such plots have a rather alarmist view.

 

How is plotting model output "alarmist"?

[iNow]

iNow, waitforufo != jryan

 

BAH! They all look the same to me!

 

Thanks, IA.

Sorry, waitforufo. I totally screwed the pooch on that one.

[waitforufo]

 

The image in my last post is no longer displayed so I offer this new one with additions. Bascule asks "How is plotting model output "alarmist"?" In my new image I include comments on "alarmist" projections based on acceptance of a dominant causal relationship of global temperature to man made greenhouse gases like CO2 and Methane. I also provide one of many acceptable global temperature projections based on the historical global temperature record alone. This projection is not "alarmist."

 

Since a layman like myself can easily see that the current global temperature is still within normal statistical variance, why should I be concerned with projections based on a relationship that is not yet proven to be causal or dominant?

[bascule]

The image in my last post is no longer displayed so I offer this new one with additions. Bascule asks "How is plotting model output "alarmist"?" In my new image I include comments

 

Your comments are extremely misinformed. Do you see the size of the error bars on the historical reconstructions? You've cherry picked a particular point in one reconstruction, and used that to try to argue that current warming is unprecedented.

 

"alarmist" projections based on acceptance of a dominant causal relationship of global temperature to man made greenhouse gases like CO2 and Methane.

 

The projections are based on model output, which is based on a rigorous understanding of the climate system. It's not like climate scientists were sitting around one day and decided "let's blame it all on CO₂!" then framed their argument to match that assumption. This is a conclusion based on decades of research and modeling.

 

Again: you've failed to demonstrate how plotting model output is "alarmist". There's nothing alarmist about it. These are scientists putting forward their best estimates with a rigorous and methodological approach.

 

Since a layman like myself can easily see that the current global temperature is still within normal statistical variance

 

And you base this on what... opening up a graph in paintbrush and drawing some lines on it? Did you download their data and do a statistical analysis on it? Are you even remotely qualified to do that?

[Psycho]

I'd say we've entered "Do Not Feed the Trolls" territory. I don't think you're going to get any worthwhile information out of jryan.

Did that not happen many pages ago when they were exactly the same as the page before with jryan just saying the graph was wrong and not providing any evidence and with pretty much everything him saying about the graphs being wrong, or am I the only one that noticed this.

[swansont]

Did that not happen many pages ago when they were exactly the same as the page before with jryan just saying the graph was wrong and not providing any evidence and with pretty much everything him saying about the graphs being wrong, or am I the only one that noticed this.

 

Regardless of the accuracy of this assessment, the discussion should focus on the topic at hand and the arguments presented.

[waitforufo]

Your comments are extremely misinformed. Do you see the size of the error bars on the historical reconstructions? You've cherry picked a particular point in one reconstruction, and used that to try to argue that current warming is unprecedented.

 

When I look at the plot I see a red line. That red line, if I understand it correctly, is an estimate of global temperature based on the historical record. It is an estimate because there are uncertainties in that historical record. Again, if I understand the graph, the gray fuzz, if you will, on either side of the red line represents the uncertainty in the historical record.

 

Am I wrong so far?

 

I accept the red line as the mean historical record. The person presenting the data simply averaged out the uncertainty.

 

How am I doing?

 

But how does this uncertainty work? I don't believe it works like band limited white noise added to a coherent signal. If it were like band limited white noise added to a coherent signal, averaging out the noise would be the best approach.

 

Instead, all we know is that the temperature did in fact lie somewhere within the bounds of the gray fuzz. Also, since sudden global temperature shifts are unlikely a smooth functional plot is reasonable.

 

Have I gone off the rails yet?

 

The plot I present fits with the gray fuzz of uncertainty and is a smooth function. I would argue that any plot that met that criteria would be an acceptable estimate of the historical record (wait for it) based on the global temperature record presented in the graph.

 

Did I miss something?

 

The green dashed line shows the maximum high temperature deviation of the historical temperature record including uncertainty.

 

Is this wrong?

 

If this is true, then as long as we stay below this line everything is okay.

 

Right?

 

So now we have to look into the future. If one were to make projection based on the historical temperature record including uncertainty, without considering other data sources, one would expect that temperatures have now peaks and would begin to cool.

 

Without considering other data sources, could one predict something else based on the historical temperature record including uncertainty? I think not.

 

The projections made in the graph are based on an acceptance of a strong causal relationship of global temperature to man made greenhouse gases like CO2.

 

Am I missing something here?

 

The projections are based on model output, which is based on a rigorous understanding of the climate system. It's not like climate scientists were sitting around one day and decided "let's blame it all on CO₂!" then framed their argument to match that assumption. This is a conclusion based on decades of research and modeling.

 

I agree, their models are based on accepting the concept of a strong causal relationship of global temperature to man made greenhouse gases.

 

Again: you've failed to demonstrate how plotting model output is "alarmist". There's nothing alarmist about it. These are scientists putting forward their best estimates with a rigorous and methodological approach.

 

I find them "alarmist" because we have not crossed the boundary of the historical record. At the moment global temperatures are still within the "normal" range. The graph shows this clearly.

 

And you base this on what... opening up a graph in paintbrush and drawing some lines on it?

 

That is exactly what I did. Engineers, like me, extract valid data from plots like this every day. Why do you think plots are published in engineering handbooks? There is nothing unscientific about such an analysis.

 

Did you download their data and do a statistical analysis on it?

 

No I looked at the plot and did the statistical analysis with brain.

 

Are you even remotely qualified to do that?

 

Well I have been a communications engineer for better than two decades. Most of my carrier has been dealing with extracting week signals from noise. I have also taken several graduate level courses in statistics. So yea, I would say I'm at least not a novice.

[bascule]

When I look at the plot I see a red line.

 

I see two red lines, FWIW. I assume you're talking about the one that's a historical reconstruction, not the one denoting the highest temperatures in the past 400,000 years.

 

That red line, if I understand it correctly, is an estimate of global temperature based on the historical record. It is an estimate because there are uncertainties in that historical record. Again, if I understand the graph, the gray fuzz, if you will, on either side of the red line represents the uncertainty in the historical record.

 

Am I wrong so far?

 

I accept the red line as the mean historical record. The person presenting the data simply averaged out the uncertainty.

 

How am I doing?

 

It's not just the "mean". The red and yellow lines represent two independent historical reconstructions. The plots are the data that came out of the model.

 

But how does this uncertainty work? I don't believe it works like band limited white noise added to a coherent signal. If it were like band limited white noise added to a coherent signal, averaging out the noise would be the best approach.

 

Instead, all we know is that the temperature did in fact lie somewhere within the bounds of the gray fuzz. Also, since sudden global temperature shifts are unlikely a smooth functional plot is reasonable.

 

Have I gone off the rails yet?

 

The plot I present fits with the gray fuzz of uncertainty and is a smooth function.

 

And instead of getting it out of a model, you pulled it out of your ass.

 

I would argue that any plot that met that criteria would be an acceptable estimate of the historical record (wait for it) based on the global temperature record presented in the graph.

 

Did I miss something?

 

Yes. There's no methodology behind your line. You just drew a line on the chart, and claim it's good because it's within the uncertainties of a real model.

 

I'm done. I can't take anymore.

[waitforufo]

It's not just the "mean".

 

Come on. Look at that red line again. Can you really tell me that red line within the gray uncertainty is not simply the mean of the uncertainty? Are we not looking at the same plots?

 

And instead of getting it out of a model, you pulled it out of your ass.

 

Yes, this is exactly what I did, but how is it wrong?

 

There's no methodology behind your line. You just drew a line on the chart, and claim it's good because it's within the uncertainties of a real model.

 

First you say I don't have a methodology and then you quite correctly explain my methodology. Yes, my methodology is very simplistic. Again, how is it wrong?

 

I'm done. I can't take anymore.

 

I appreciate your frustration, but please have empathy for mine. The graph was presented as proof. Its intent is to show how all outcomes point to a dire future. When I look at the plot and read how it is described I can see a simple alternate interpretation. The plot shows variance in the historical global temperature. We are not currently outside that variance. So what's to worry about?

 

When I look at this graph, it seems to me that its author is simply using improved accuracy over time to support projections of future warming. I intentionally pulled a plot line "out of my ass" that showed past warming during periods where uncertainty was large. I believe the line I created is equally likely to the line presented by the author of the graph. If this is wrong please explain how? Perhaps I do not understand how the uncertainty presented should be used. I did however clearly explain how I used it. Can you provide an alternate explanation? If not at least admit that you do not know.

 

The point of my post is that we know far too little about the mechanisms that drive global temperature to place any credibility in dire projections shown in the graph.

[bascule]

Come on. Look at that red line again. Can you really tell me that red line within the gray uncertainty is not simply the mean of the uncertainty? Are we not looking at the same plots?

 

No, it is NOT the mean of the uncertainty. The uncertainty extends out from the red line, which is generated from model output. You're reasoning is completely backwards. They begin with the model output (the red line) and then plot uncertainty from there, not vice versa.

 

First you say I don't have a methodology and then you quite correctly explain my methodology. Yes, my methodology is very simplistic. Again, how is it wrong?

 

Your methodology is you pulled it out of your ass?

 

That's not a very sound methodology, sorry.

[waitforufo]

They begin with the model output (the red line) and then plot uncertainty from there, not vice versa.

 

I'm fine with that interpretation. My plot is still within the uncertainty so it is probable. I still contend that it is equally likely.

 

 

Your methodology is you pulled it out of your ass?

 

That's not a very sound methodology, sorry.

 

Yes, I freely admit that. But that is my point. The graph presented by the author does not pass the sniff test. If a guy with a few statistics classes under his belt can use the data presented to reach an opposite conclusion what good is the analysis presented?

[bascule]

I'm fine with that interpretation. My plot is still within the uncertainty so it is probable. I still contend that it is equally likely.

 

Yes, I freely admit that. But that is my point. The graph presented by the author does not pass the sniff test. If a guy with a few statistics classes under his belt can use the data presented to reach an opposite conclusion what good is the analysis presented?

 

I give up. You're hopeless.

[waitforufo]

I give up. You're hopeless.

 

Actually, I'm full of hope. I see a bright future with few problems, particularly few climate problems.

 

Perhaps this image will provide you with some hope.

 

 

If not that one perhaps this will.

 

 

Wow, a whole 0.1C above normal in 100 years. But that small of an average deviation is obviously just noise.

 

The average temperature of 53.0°F (11.7°C) was 0.2°F (0.1°C) above the 20th century (1901-2000) mean. -National Climatic Data Center

20 January 2009

 

The people at NOAA seem hopeful as well. Perhaps you should read this.

 

http://www.ncdc.noaa.gov/oa/climate/research/2008/ann/us-summary.html

[bascule]

Actually, I'm full of hope.

 

You're full of something... it isn't hope.

 

I see a bright future with few problems, particularly few climate problems.

 

Let's pretend our problems don't exist and they will disappear! It's like The Secret!

 

Perhaps this image will provide you with some hope.

 

[cherry picked data explained by an unusually strong La Nina]

 

If not that one perhaps this will.

 

[graph showing warming average surface temperatures for the continental US]

 

Wow, a whole 0.1C above normal in 100 years. But that small of an average deviation is obviously just noise.

 

Yes, that is correct. 2007-08 were unusually cold. If you look at the values coming out of the smoothing filter the difference is 1.1C. Isn't cherrypicking fun?

[waitforufo]

You're full of something... it isn't hope.

 

Let's pretend our problems don't exist and they will disappear! It's like The Secret!

 

Yes, that is correct. 2007-08 were unusually cold. If you look at the values coming out of the smoothing filter the difference is 1.1C. Isn't cherrypicking fun?

 

 

I did not create this data. I did not create this plot. I did not draw that mean line.

 

I also did not write or provide all of the other information at this post.

 

http://www.ncdc.noaa.gov/oa/climate/research/2008/ann/us-summary.html

 

The fine people at NOAA did. Do you have a problem with my source?

 

With regard to cherry picking, the plot shows the last 100 years, a period of time that includes all of the significant CO2 contributions of man to the atmosphere. This seems to be a significant period of time to me. Not to you?

[bascule]

 

I did not create this data. I did not create this plot. I did not draw that mean line.

 

I also did not write or provide all of the other information at this post.

 

http://www.ncdc.noaa.gov/oa/climate/research/2008/ann/us-summary.html

 

The fine people at NOAA did. Do you have a problem with my source?

 

No, but you're using their graphs to make arguments they are not.

 

The purple line is not the trend line. One of its endpoints is the 20th century mean, the other is the mean for 2008:

 

Based on data through the end of 2008, the contiguous U.S. experienced a nationally averaged temperature that was the coolest in more than ten years. The average temperature of 53.0°F (11.7°C) was 0.2°F (0.1°C) above the 20th century (1901-2000) mean.

 

I'm not sure why they plotted it like that, however you seem to be using it to argue that the overall trend is 0.1 C of warming (for the contiguous 48 states).

 

And then the world, right? That seems to be what you're implying.

 

With regard to cherry picking, the plot shows the last 100 years, a period of time that includes all of the significant CO2 contributions of man to the atmosphere. This seems to be a significant period of time to me. Not to you?

 

You cherry picked the year 2008, which was unusually cold. All your arguments were based on 2008, not the overall trends.

 

Can you do me a favor and... read harder before you try to use something like NOAA's data to present arguments that are in direct contradiction to NOAA's conclusions?

 

http://www.ncdc.noaa.gov/oa/climate/globalwarming.html#q3

 

Global surface temperatures have increased about 0.74°C (plus or minus 0.18°C) since the late-19th century' date=' and the linear trend for the past 50 years of 0.13°C (plus or minus 0.03°C) per decade is nearly twice that for the past 100 years. The warming has not been globally uniform. Some areas (including parts of the southeastern U.S. and parts of the North Atlantic) have, in fact, cooled slightly over the last century. The recent warmth has been greatest over North America and Eurasia between 40 and 70°N. Lastly, seven of the eight warmest years on record have occurred since 2001 and the 10 warmest years have all occurred since 1995.

 

Recent analyses of temperature trends in the lower and mid- troposphere (between about 2,500 and 26,000 ft.) using both satellite and radiosonde (weather balloon) data show warming rates that are similar to those observed for surface air temperatures. These warming rates are consistent with their uncertainties and these analyses reconcile a discrepancy between warming rates noted on the IPCC Third Assessment Report (U.S. Climate Change Science Plan Synthesis and Assessment Report 1.1).[/quote']

[waitforufo]

No, but you're using their graphs to make arguments they are not. The purple line is not the trend line. One of its endpoints is the 20th century mean, the other is the mean for 2008:

 

This statement is simply not true. The NOAA link [http://www.ncdc.noaa.gov/oa/climate/research/2008/ann/us-summary.html] provides the following surrounding the 100 year data graph.

 

Originally Posted by NOAA

Based on data through the end of 2008, the contiguous U.S. experienced a nationally averaged temperature that was the coolest in more than ten years. The average temperature of 53.0°F (11.7°C) was 0.2°F (0.1°C) above the 20th century (1901-2000) mean.

 

The average temperature for the U.S. has not been this close to the 20th century mean since 1997. The January - December statewide temperature ranking map for 1997 display a close resemblance to the 2008 map. The anomalous warmth that the contiguous U.S. has seen the past 10 years was generally isolated within the West, Southwest, and Northeast Regions during 2008. In contrast, the Central and East North Central regions of the U.S. were below the 20th century mean. This resulted in a near normal nationally averaged temperature of 53.0°F (11.7°C).

 

For clarity I again post the graph.

 

 

You don’t even need the above NOAA text to see that the purple line is the trend line. The blue line is the data. The red line is the filtered data. The filtered red line appears to be a simple linear regression of the blue line. For simplicity sake, look at the red line. Captured between the red line and purple line is an area. If the area captured below the purple line is equal to the area above the purple line, then the purple line is indeed the trend line. This is obvious by simple observation.

 

I'm not sure why they plotted it like that, however you seem to be using it to argue that the overall trend is 0.1 C of warming (for the contiguous 48 states).

 

I am not arguing this, the NOAA data is clearly demonstrating this.

 

And then the world, right? That seems to be what you're implying.

 

The continental contiguous 48 states are indeed part of the world. I have a particular interest in this area. It is where I keep my stuff.

 

You cherry picked the year 2008, which was unusually cold. All your arguments were based on 2008, not the overall trends.

 

I simply displayed that year in particular. The plot above shows 100 years. I accept all 100 of those years and the trend line provided by NOAA. Data NOAA is so proud of they attached there seal to it.

 

Can you do me a favor and... read harder before you try to use something like NOAA's data to present arguments that are in direct contradiction to NOAA's conclusions?

 

http://www.ncdc.noaa.gov/oa/climate/globalwarming.html#q3

 

From this link you provide the attached quote.

 

Originally Posted by NOAA

Global surface temperatures have increased about 0.74°C (plus or minus 0.18°C) since the late-19th century, and the linear trend for the past 50 years of 0.13°C (plus or minus 0.03°C) per decade is nearly twice that for the past 100 years. The warming has not been globally uniform. Some areas (including parts of the southeastern U.S. and parts of the North Atlantic) have, in fact, cooled slightly over the last century. The recent warmth has been greatest over North America and Eurasia between 40 and 70°N. Lastly, seven of the eight warmest years on record have occurred since 2001 and the 10 warmest years have all occurred since 1995.

 

Recent analyses of temperature trends in the lower and mid- troposphere (between about 2,500 and 26,000 ft.) using both satellite and radiosonde (weather balloon) data show warming rates that are similar to those observed for surface air temperatures. These warming rates are consistent with their uncertainties and these analyses reconcile a discrepancy between warming rates noted on the IPCC Third Assessment Report (U.S. Climate Change Science Plan Synthesis and Assessment Report 1.1).

 

The bold text, in my opinion, are examples of cherry picking.

 

Finally in a previous post you stated the following.

 

cherry picked data explained by an unusually strong La Nina

 

Imagine if I were to point to a part of the 100 year graph and state "ignore that bit because it occurred during an unusually strong El Nino. So who is cherry picking? Again I accept all 100 years.

[iNow]

This statement is simply not true.

 

<...>

 

For clarity I again post the graph.

 

You don’t even need the above NOAA text to see that the purple line is the trend line.

Sorry. Try again.

 

Long term MEAN [math]\ne[/math] Trend Line.

[bascule]

Sorry. Try again.

 

Long term MEAN [math]\ne[/math] Trend Line.

 

Yes, here's a graph (courtesy of Weather Underground) showing both the trend line and the long-term mean:

 

Average December temperatures for the U.S. from 1895 to 2006. Image credit: National Climatic Data Center.

 

http://www.wunderground.com/blog/JeffMasters/comment.html?entrynum=609

[Mokele]

You don’t even need the above NOAA text to see that the purple line is the trend line. The blue line is the data. The red line is the filtered data. The filtered red line appears to be a simple linear regression of the blue line. For simplicity sake, look at the red line. Captured between the red line and purple line is an area. If the area captured below the purple line is equal to the area above the purple line, then the purple line is indeed the trend line. This is obvious by simple observation.

 

There are several fairly serious mathematical errors in the above section that need correcting.

 

1) The purple line is explicitly labeled as the mean. A mean is just a single value, thus appears as a horizontal line. It conveys no information about the relationship over time.

 

2) The filtered data is not a linear regression. Otherwise it would be a straight line with a given slope. You're thinking of a general linear model (GLM), which can include squared, cubed, etc terms. However, that is also wrong. Filtered data =/= any sort of regression. A filter is a mathematical construct (originally from electrical use) which removes frequencies from a plot in a particular way.

 

For instance, I sometimes do electrophysiology on muscle tissue. I get unwanted signals due to, for instance, the wire moving as the animal swings its leg (which will appear as a shifting baseline). I remove unwanted signals by using a 10 hz high-pass filter, which means any fluctuations of less than 10 Hz (movement artefacts due to the animal's motion, etc) are removed, leaving only the high-frequency components. I can do this safely because I know the firing rate of muscle in vivo is always over 10 Hz. The baseline shift is gone, and now I can do proper EMG analysis.

 

Similarly, somewhere in the related text for this graph (or at least the technical documentation), it should say what the filter parameters were. Perhaps they used a 3-year high-pass, removing some inter-annual variation.

 

It's also worth noting that a filter is substantially more complex than a running average, too. I'm sure Wikipedia has something on the math behind it.

 

3) The area thing is a common misconception about linear regressions. What a least-squares linear regression does is try to minimize and equalize the squares of the distances from the points to the lines. This is not the same thing as the area, and in linear regressions across data with a few exceptional outliers, the area relationship you suggested will be violated, as will the equality of simple distance.

 

Furthermore, and most simply, were it a regression line ("trend line" is pretty much never used by anyone in science), it would be labeled as such.

 

Mokele

Edited July 24, 2009 by Mokele