Do you think CAGW is just a UN scheme to impose global governance...

[sokolow]

That's fuuny because you'd think cumulative historical and per capita emissions would play a part in that calculus.

[Cygnus X-1]

Hello. I'm new to these forums, and would like to ask some questions about this "climate debate".

 

I'm not a meterorologist, scientist, statatician, theorist, or even collage graduate...yep, just someone who is facinated by science, weather, history, astronomy, etc.

 

From what I understand, meteorlogical models (NAM, EURO,et al) depend on data from real time observations including balloons, aircraft, satellite, etc. and are all less than accurate, obviously...which leads me to wonder, and ask after reading this entire thread...

 

Where do "climate models" of the future get their DATA from to make predictions? This notion of climate models having ANY validity escapes me; seems entirely illogoical. 

 

So, I kindly ask for an explanation to this apparent simple discretion. Thanks!

[sokolow]

Hello. I'm new to these forums, and would like to ask some questions about this "climate debate".

I'm not a meterorologist, scientist, statatician, theorist, or even collage graduate...yep, just someone who is facinated by science, weather, history, astronomy, etc.

From what I understand, meteorlogical models (NAM, EURO,et al) depend on data from real time observations including balloons, aircraft, satellite, etc. and are all less than accurate, obviously...which leads me to wonder, and ask after reading this entire thread...

Where do "climate models" of the future get their DATA from to make predictions? This notion of climate models having ANY validity escapes me; seems entirely illogoical.

So, I kindly ask for an explanation to this apparent simple discretion. Thanks!

If your thesis project was building a climate model from scratch what kind of data would you imagine you'd need and where do you think you'd go looking for it?

Here is an intro

http://www.azimuthproject.org/azimuth/show/Climate+model

Here are some stripped down DIY models you can play with.

http://www.easterbrook.ca/steve/2013/01/simple-climate-models-to-play-with-in-the-classroom/

Here's access portals to the gubmint's climate data suites

http://www.ncdc.noaa.gov/

http://www.ncdc.noaa.gov/data-access/model-data

http://climatemodeling.science.energy.gov/

And of course fhe wikipedia entries on "climate modeling" are linkdense and detailed

[Cygnus X-1]

If your thesis project was building a climate model from scratch what kind of data would you imagine you'd need and where do you think you'd go looking for it?

Here is an intro

http://www.azimuthproject.org/azimuth/show/Climate+model

Here are some stripped down DIY models you can play with.

http://www.easterbrook.ca/steve/2013/01/simple-climate-models-to-play-with-in-the-classroom/

Here's access portals to the gubmint's climate data suites

http://www.ncdc.noaa.gov/

http://www.ncdc.noaa.gov/data-access/model-data

http://climatemodeling.science.energy.gov/

And of course fhe wikipedia entries on "climate modeling" are linkdense and detailed

LOL, I'm asking an honest question as a 45 y/o independant thinker.

[Cygnus X-1]

How can you model the future without data?

 

[sokolow]

LOL, I'm asking an honest question as a 45 y/o independant thinker.

that was a honest answer that made no assumptions about your age or qualities as a thinker? Imagine you'd like to model the performance of a howitzer or like, do a structural analysis & simulation of how a bridge would react to various (weather, seismic, traffic) stresses. How does a body approach a modeling problem of another type? We don't, of course, have data from the future.

[PhillipS]

How can you model the future without data?

 

 

Global Climate Models are physics-based representations of our understanding of the processes that drive Earth's climate.  The models are initialized with various scenarios and run to provide an idea of how climate may respond in the future.  Climate models aren't perfect, no model is, but they are powerful tools for studying climate change, and are our best tools for assessing the consequences of the changes we're making to the atmosphere.

 

If you don't have any confidence in climate models then what do you suggest as alternatives?  The Farmers Almanac?  Entrails of a chicken?  Blind faith that what happened in the past determines what will happen in the future?  People, businesses, and policymakers need some idea of what the future may hold in order to make informed decisions.  At this time Global Climate Models, limitations and all, are the best tool we have.  The good news is that scientists around the world are working to improve our understanding of climate, and to improve the models that embody that understanding.

 

If you really want to understand climate modeling better (assuming you're not just trolling) then you have some self-study ahead of you.  Sokolow provided links to good material, but neither he nor anyone else can spoon-feed you understanding - you have to build that for yourself.

[Isotherm]

One of the primary problems with global climate models is that their simplification of Earth's system is not representative of reality, and they do not accurately simulate a number of crucial variables in climate. An obvious factor that comes to mind is clouds, namely their behavior and ability to initiate feedback cycles. Another issue is that the inputs into these models are such that they largely overestimate the sensitivity of climate to Co2 radiative forcing, and thus their outputs are more extreme than reality dictates. The inputs need to be somewhat accurate and representative of Earth's climate for the outputs to be anywhere near something remotely believable.

[The_Global_Warmer]

One of the primary problems with global climate models is that their simplification of Earth's system is not representative of reality, and they do not accurately simulate a number of crucial variables in climate. An obvious factor that comes to mind is clouds, namely their behavior and ability to initiate feedback cycles. Another issue is that the inputs into these models are such that they largely overestimate the sensitivity of climate to Co2 radiative forcing, and thus their outputs are more extreme than reality dictates. The inputs need to be somewhat accurate and representative of Earth's climate for the outputs to be anywhere near something remotely believable.

 

Can you show some proof of this.

[sokolow]

@Isotherm On the one hand "yes" but on the other hand its not like its hasn't been discussed for decades that cloud effects are hard to model or parameterize. Its not like anyone is trying to hide that -- its surveyed in the IPCC report. In any case I don't gather that's what CX-1 is asking.

[nflwxman]

One of the primary problems with global climate models is that their simplification of Earth's system is not representative of reality, and they do not accurately simulate a number of crucial variables in climate. An obvious factor that comes to mind is clouds, namely their behavior and ability to initiate feedback cycles. Another issue is that the inputs into these models are such that they largely overestimate the sensitivity of climate to Co2 radiative forcing, and thus their outputs are more extreme than reality dictates. The inputs need to be somewhat accurate and representative of Earth's climate for the outputs to be anywhere near something remotely believable.

 

There is simply no strong evidence of this.  If you are talking about the last 10 years, which is marred by ENSO negative conditions and slumping solar radiance, that's not a good sample size.

 

A climate model can never be validated using small decade long records.  There is far too much uncertainty and variability in a given decade's weather do that.  

[Isotherm]

The idea that we haven't given the models enough time doesn't make much sense to me. Given the fact that the vast majority of models have overestimated the warming, it's clear that the parameters utilized in making the projections are not accurate in probably a number of ways. Variables were not weighted correctly, and some factors likely weighted more strongly than others, including other limitations.

 

[nflwxman]

The idea that we haven't given the models enough time doesn't make much sense to me. Given the fact that the vast majority of models have overestimated the warming, it's clear that the parameters utilized in making the projections are not accurate in probably a number of ways. Variables were not weighted correctly, and some factors likely weighted more strongly than others, including other limitations.

 

So that graph is scaled to 1979-1983 average?  And climate models have magically been incorrect with the earth warming at .2C/decade in the 80s?  That graph seems fishy as all heck man.  I think the small baseline increases the spread of the models making it look like they were worse than they actually are.  Here's the average of the AR4 models with a 1980-2000 baseline.

 

Edit: Lol that graph has us warming ~0.15-.2 degrees C between 2003-2008.  It's definitely off, because that would be nearly 0.35-0.4/decade.

 

[sokolow]

Is that 95 out of 100 physicians agree chart also Dr. Roy Spencer dot com?

[ORH_wxman]

So that graph is scaled to 1979-1983 average?  And climate models have magically been incorrect with the earth warming at .2C/decade in the 80s?  That graph seems fishy as all heck man.  I think the small baseline increases the spread of the models making it look like they were worse than they actually are.  Here's the average of the AR4 models with a 1980-2000 baseline.

 

Edit: Lol that graph has us warming ~0.2 degrees C between 2003-2008.

 

 

 

 

He posted CMIP5 models...your graph is AR4 models. They aren't the same. I've actually never seen that graph of AR4 models.

 

 

But the poor performance of the CMIP5 suite is documented in recent literature:

 

http://onlinelibrary.wiley.com/doi/10.1002/grl.50562/abstract

 

 

http://www.see.ed.ac.uk/~shs/Climate%20change/Climate%20model%20results/over%20estimate.pdf

 

 

 

We've debated the reasons for the discrepency, but there is no doubt that it exists.

[sokolow]

IIRC the AR4 less the Canadian is from a Tamino post a while back.

edit yah here we be

http://web.archive.org/web/20100322194954/http://tamino.wordpress.com/2010/01/13/models-2/

[StudentOfClimatology]

That looks an old, statistically-recalibrated AR-4 multi-ensemble spread, ending in 2009. Not exactly helpful, Re: near-term verification, considering that the forcings and data are over 6 years old.

[nflwxman]

That looks an old, statistically-recalibrated AR-4 multi-ensemble spread, ending in 2009. Not exactly helpful, Re: near-term verification, considering that the forcings and data are over 6 years old.

Please, by all means post a more updated calibrated graph of the CIMP5 models.  The one Isotherm posted above is clearly not correct.

 

BTW, there truly should be no such thing as "near term" verification for a model mean run in the past.  It's a smoothed out average of many individual deterministic runs.  It will not capture decadal natural variability well because it's a mean.  If the model is seeded with natural variability in hindcasts, and it performs, than it can be verified.  The study below in nature displays that concept well.

 

http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2357.html

[ORH_wxman]

Please, by all means post a more updated calibrated graph of the CIMP5 models.  The one Isotherm posted above is clearly not correct.

 

BTW, there truly should be no such thing as "near term" verification for a model mean run in the past.  It's a smoothed out average of many individual deterministic runs.  It will not capture decadal natural variability well because it's a mean.  If the model is seeded with natural variability in hindcasts, and it performs, than it can be verified.  The study below in nature displays that concept well.

 

http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2357.html

 

 

The problem is even if you do an analysis on the individual model runs, they still don't simulate it properly. They explain it in the papers I posted above. That's why we have 5-95% confidence intervals...so we can not just rely on the ensemble mean...but a range.

 

If the theories about this essentially being a 2 sigma natural variability event are true, then we will be well back within the CMIP5 confidence intervals by 2020-2025. So it won't take long for us to test them.

[JBG]

There is simply no strong evidence of this.  If you are talking about the last 10 years, which is marred by ENSO negative conditions and slumping solar radiance, that's not a good sample size.

When you use the term "marred" that implies that you are "wish-casting." In other words you want to find that there is AGW and thus you would like the data to fit your construct. It doesn't and it won't.

[The_Global_Warmer]

When you use the term "marred" that implies that you are "wish-casting." In other words you want to find that there is AGW and thus you would like the data to fit your construct. It doesn't and it won't.

 

 

What are you talking about? 

[StudentOfClimatology]

Please, by all means post a more updated calibrated graph of the CIMP5 models. The one Isotherm posted above is clearly not correct.

BTW, there truly should be no such thing as "near term" verification for a model mean run in the past. It's a smoothed out average of many individual deterministic runs. It will not capture decadal natural variability well because it's a mean. If the model is seeded with natural variability in hindcasts, and it performs, than it can be verified. The study below in nature displays that concept well.

http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2357.html

I agree with you 100% regarding the aspects of "near term" verification. Pretty much a hopeless exercise, as things stand now..

[nflwxman]

When you use the term "marred" that implies that you are "wish-casting." In other words you want to find that there is AGW and thus you would like the data to fit your construct. It doesn't and it won't.

I'm sorry if my choice of adjectives is distasteful to you.

 

"Marred: To impair the soundness, unsuitable"

[bluewave]

Please, by all means post a more updated calibrated graph of the CIMP5 models.  The one Isotherm posted above is clearly not correct.

 

BTW, there truly should be no such thing as "near term" verification for a model mean run in the past.  It's a smoothed out average of many individual deterministic runs.  It will not capture decadal natural variability well because it's a mean.  If the model is seeded with natural variability in hindcasts, and it performs, than it can be verified.  The study below in nature displays that concept well.

 

http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2357.html

 

This is it:

 

[nflwxman]

The problem is even if you do an analysis on the individual model runs, they still don't simulate it properly. They explain it in the papers I posted above. That's why we have 5-95% confidence intervals...so we can not just rely on the ensemble mean...but a range.

 

If the theories about this essentially being a 2 sigma natural variability event are true, then we will be well back within the CMIP5 confidence intervals by 2020-2025. So it won't take long for us to test them.

 

Agreed.  I have a feeling we will be below the mean until we get a few back to back nino events (when the PDO flips positive).  I wouldn't be surprised if we are above after that.  This is provided the solar/CO2 forcing assumptions are close to future realities.

[nflwxman]

This is it:

 

Screen shot 2014-09-15 at 4.28.38 PM.png

Thanks. That looks right.  I will assume the mean is about right in the middle of all those individual models.

 

Actually, Bluewave..what dataset is that?  It shows 1998 warmer than 2005 and 2010.  It also shows 2005 as warmer than 2010.  Is that a satellite dataset?

 

Clearly the choice of dataset makes a huge difference.  If you use the Hadcrut4 Cowtan and Way dataset, it looks like a much better story for the climate models.

[JBG]

 

When you use the term "marred" that implies that you are "wish-casting." In other words you want to find that there is AGW and thus you would like the data to fit your construct. It doesn't and it won't.

What are you talking about?

 

Definition of "marred" (link) :

 

Full Definition of MAR

transitive verb

1:  to detract from the perfection or wholeness of :  spoil

Examples of MAR

• A large scar marred his face.

• Her acting mars an otherwise great movie.

[Jonger]

Agreed.  I have a feeling we will be below the mean until we get a few back to back nino events (when the PDO flips positive).  I wouldn't be surprised if we are above after that.  This is provided the solar/CO2 forcing assumptions are close to future realities.

 

I would bet my house on the hunch that we will NEVER ride above the model mean.

[StudentOfClimatology]

This is it:

Screen shot 2014-09-15 at 4.28.38 PM.png

Man, the upcoming decade should be quite exciting for science junkies. Lots of theory testing.

[JBG]

In reference to climate models, that is the correct usage.  Climate models were marred by slumping solar activity.  

How is that usage consistent with good English?