2011 Global Temperatures

[skierinvermont]

No 1961-1991

[BethesdaWX]

Much of the difference is that the GFS initialization maps on Maue's website use a 1979-2010 baseline. Then there is the fact that there is a lot of variability within a month. Even though we might dip to -.6C or something, it jumps around a lot and goes through wild swings so no single month finishes anywhere near -.6C.

GFS and ECMWF model initializations show more of a warming trend than HadCRUT or GISS over the last 30 years. I suppose now you will retract your statement about them being accurate.

There is no such thing as 1979-2010. This isn't satellite LT data, this is surface data. You mean 1971-2000? Thats what it was when I last checked.

Even with different baselines, regardless, there is still a huge difference. If the GFS avg initializations come in with FEB at -0.1C, GISS should come in at +0.1C tops.

FYI, I'm not talking about the ECMWF "private" LR model, I'm speaking of the Global Model used for 10 day forecasting.

That has to have accurate data visible, so I'll go with that.

[skierinvermont]

There is no such thing as 1979-2010. This isn't satellite LT data, this is surface data. You mean 1971-2000? Thats what it was when I last checked.

Even with different baselines, regardless, there is still a huge difference. If the GFS avg initializations come in with FEB at -0.1C, GISS should come in at +0.1C tops.

FYI, I'm not talking about the ECMWF "private" LR model, I'm speaking of the Global Model used for 10 day forecasting.

That has to have accurate data visible, so I'll go with that.

It says right on the chart that it is a 1979-2009 baseline.

[BethesdaWX]

It says right on the chart that it is a 1979-2009 baseline.

Oh you mean Ryan Maue's site. The old UAH baseline was flying in my head for some reason...hehe who knows

However, you are missing the main point. Regardless of the different baselines, GISS still has tended to run warmer.

GFS has always run colder than GISS in initializations since I started paying attention.

[BethesdaWX]

FWIW, the latest GFS trended colder yet again.

Initialization at -0.17C Drop has begin.

Ends up at -0.64C

[nzucker]

FWIW, the latest GFS trended colder yet again.

Looks as if the Southern Hemisphere is almost all colder than normal, with large pockets of cold anomalies in Chile and Argentina, the tropical Pacific, and of course Antarctica. The arctic outbreak over western Asia and the US Northern Plains looks to drive the cold anomalies on this side of the world. Very impressive extent of cold.

[skierinvermont]

Oh you mean Ryan Maue's site. The old UAH baseline was flying in my head for some reason...hehe who knows

However, you are missing the main point. Regardless of the different baselines, GISS still has tended to run warmer.

GFS has always run colder than GISS in initializations since I started paying attention.

No it hasn't ... I can tell you that reanalyses of GFS and ECMWF initialization show MORE warming than GISS does.

You said Maue was "plotting" these GFS initializations himself. Care to post that chart so we can compare for ourselves?

Of course now you will probably change your mind and decide it is inaccurate.

[nzucker]

No it hasn't ... I can tell you that reanalyses of GFS and ECMWF initialization show MORE warming than GISS does.

He said "since I started paying attention" so he's talking about the recent past, not the entire extent of the record.

[skierinvermont]

He said "since I started paying attention" so he's talking about the recent past, not the entire extent of the record.

I've been following it too and it hasn't. It's frequently quite warm and is just subject to volatility since it is daily data.

Makes sense considering the long term trend is even warmer than GISS. If you've been following the thread of the conversation he implied that the GISS record is biased warm based on this GFS-init data... so he clearly is not talking about just the last month or two.

[nzucker]

I've been following it too and it hasn't. It's frequently quite warm and is just subject to volatility since it is daily data.

Makes sense considering the long term trend is even warmer than GISS.

Isn't the trend on GISS .18C/decade?

That would imply we should have warmed near 2C since 1900, but we haven't. What do you mean that the long-term trend is warmer?

[skierinvermont]

Isn't the trend on GISS .18C/decade?

That would imply we should have warmed near 2C since 1900, but we haven't. What do you mean that the long-term trend is warmer?

Huh? That is the trend on GISS since 1980...

I'm saying the long term trend of weather model initialization data (GFS, ECMWF etc.) reanalyses are higher than that of GISS.

[nzucker]

Huh? That is the trend on GISS since 1980...

I'm saying the long term trend of weather model initialization data (GFS, ECMWF etc.) reanalyses are higher than that of GISS.

OK...I thought you meant it made sense that the GFS/ECM initialization data was warmer given that GISS trends are low compared to the long-term global trend, which clearly isn't the case.

In any case, February and March are going to be cold months globally, then we see what happens with the Niña as we move towards the warm season

[Clifford]

Looks as if the Southern Hemisphere is almost all colder than normal, with large pockets of cold anomalies in Chile and Argentina, the tropical Pacific, and of course Antarctica. The arctic outbreak over western Asia and the US Northern Plains looks to drive the cold anomalies on this side of the world. Very impressive extent of cold.

Keep in mind that the Ryan Maue website uses a very odd spectrum. Look at the key at the bottom.

http://www.coaps.fsu.edu/~maue/weather/

White: -0.5 to 0.5 (ok... pretty neutral)

Blue: -0.5 to -4

Grey: + 0.5 to 3 (what's with that? Who uses GREY for HOT?)

Yellow/Orange: +3 to +8... etc.

So, looking at the cart.

The oceans look about half grey/light yellow (+0.5 to +4)

and about half blue (-0.5 to -4)

Anyway, I think the use of Grey (which is usually considered neutral) on the chart causes a visual cold-bias.

[BethesdaWX]

No it hasn't ... I can tell you that reanalyses of GFS and ECMWF initialization show MORE warming than GISS does.

You said Maue was "plotting" these GFS initializations himself. Care to post that chart so we can compare for ourselves?

Of course now you will probably change your mind and decide it is inaccurate.

1) Care to post GFS/ECMWF re-analysis? I know exactly what you're thinking of, and its not the OP ECMWF 10 day.

2) No thats not what I said, read my friggin post. "Will be" and "Was" are two different things. And this matters......why?

Debate like a man.

[BethesdaWX]

Keep in mind that the Ryan Maue website uses a very odd spectrum. Look at the key at the bottom.

http://www.coaps.fsu.../~maue/weather/

Anyway, I think the use of Grey (which is usually considered neutral) on the chart causes a visual cold-bias.

Agree....and its definitely a pain in the ass, I'd prefer a more objective color table.

[nzucker]

Agree....and its definitely a pain in the ass, I'd prefer a more objective color table.

Yeah, I don't like his color scheme at all but it's still easy to tell where the warm and cold anomalies are...generally warmer than average in the High Arctic and Southern US...colder than average in the Tropical Pacific, Western Asia, Central Canada, Northern Tier of the US, Antarctica, Australia, and the temperate latitudes of South America.

[skierinvermont]

1) Care to post GFS/ECMWF re-analysis? I know exactly what you're thinking of, and its not the OP ECMWF 10 day.

2) No thats not what I said, read my friggin post. "Will be" and "Was" are two different things. And this matters......why?

Debate like a man.

As you can see from the following chart, the one you have been referencing from Ryan Maue's website, it is entitled "NCEP CFSR 2m Raw Temp Anomalies"

NCEP CFSR stands for the "NCEP Climate Forecast System Reanalysis."

The most recent version of this reanalysis was published by the American Meteorological Society.

http://journals.amet.../2010BAMS3001.1

It is of course constantly updating with every run of the GFS.

As you can see on page 1040 of the above link, the trend of the NCEP CFSR from 1979-2009 was 1.02K/31yrs or, .33C/decade. This is significantly faster warming than GISS or HadCRUT.

As I said originally, the exact same product that you are looking at and saying "look it shows -.6C anomalies next week" shows a warming trend of .33C/decade over the last 31 years (based on a linear regression calculation. The NCEP CFSR is actually notoriously warm among all temperature indexes. This is the exact same product that you have been quoting from Maue's website.

This is why I find your sudden fascination with the NCEP CFSR product from Maue's website so ironic. It is literally the warmest data source available for the last 30 years.

[tacoman25]

As you can see from the following chart, the one you have been referencing from Ryan Maue's website, it is entitled "NCEP CFSR 2m Raw Temp Anomalies"

NCEP CFSR stands for the "NCEP Climate Forecast System Reanalysis."

The most recent version of this reanalysis was published by the American Meteorological Society.

http://journals.amet.../2010BAMS3001.1

It is of course constantly updating with every run of the GFS.

As you can see on page 1040 of the above link, the trend of the NCEP CFSR from 1979-2009 was 1.02K/31yrs or, .33C/decade. This is significantly faster warming than GISS or HadCRUT.

As I said originally, the exact same product that you are looking at and saying "look it shows -.6C anomalies next week" shows a warming trend of .33C/decade over the last 31 years (based on a linear regression calculation. The NCEP CFSR is actually notoriously warm among all temperature indexes. This is the exact same product that you have been quoting from Maue's website.

This is why I find your sudden fascination with the NCEP CFSR product from Maue's website so ironic. It is literally the warmest data source available for the last 30 years.

I really don't think you can directly compare a source based on linear regression calculation to ones based on actual observed data.

[nzucker]

As I said originally, the exact same product that you are looking at and saying "look it shows -.6C anomalies next week" shows a warming trend of .33C/decade over the last 31 years (based on a linear regression calculation. The NCEP CFSR is actually notoriously warm among all temperature indexes. This is the exact same product that you have been quoting from Maue's website.

This is why I find your sudden fascination with the NCEP CFSR product from Maue's website so ironic. It is literally the warmest data source available for the last 30 years.

I just think it's a cool product to look at for short-term trends. It's not meant to be used as a measure of long-term climate change, but it's fascinating to see the movement of the various airmasses that influence our weather, the impact of ENSO on Australia and South America, and the major cold wave in the Western US. I'm definitely very interested in the fact that nearly all of the Southern Hemisphere appears to be experiencing below average temperatures this February, looks as if the impact from the Strong Niña has been much more powerful down there. Also, the Northern Hemisphere which was responsible for much of the warming in the past few years, appears to be cooling down quickly, especially in the higher latitudes like the Canadian Prairies and Russia which were some of the fastest warming regions in the last 30 years. I wonder if the high-latitude cold wave is just a fluke or part of a permanent trend towards more availability of arctic air in those regions. I've certainly noticed that the cold air seems "fresher" this winter, might be part of the trend or just coincidence.

I would be wary of the ECM analysis for global temperature...it has a documented warm bias of nearly 1C at 850mb in North America, and close to 1.5C at the surface. I'm not sure how this affects the global initialization/reanalysis data as compared to forecasting, but it makes the observations a bit suspect. When I forecast in the longer range off of the Euro, I always adjust for these well-known biases which have been shown in studies.

I also wonder if the GFS is showing more realistic temperature trends now since it just had an upgrade during Spring 2010...the .33C/decade trend, which seems excessively warm compared to all other data, may be a result of how bad the model was before the NOAA put in the latest improvements last April. I've definitely noticed the GFS has been much more useful at showing pattern changes in the long-range this winter, and it's also had some coups on major storms such as the adjustment west with the Boxing Day Blizzard.

[skierinvermont]

I really don't think you can directly compare a source based on linear regression calculation to ones based on actual observed data.

???

I'm not saying one should. Also the linear regression was used to calculate the trend line... not to calculate the CFS Reanalysis (CFSR). That's based on complex model physics including the GFS but it also models the oceans etc...

Bethesda has basically been running around siting how frigid the CFSR/GFS initialization is currently and how this proves GISS is biased too cold. My point is that the CFSR shows a 30 year trend of .33C/decade, while GISS shows only .18C/decade. So if anything CFSR would show that GISS is biased cold.

However I PERSONALLY am not saying one should use the CFSR as a global surface temperature index, since it is not based on surface temperature observations.

It is just ironic that Bethesda has been saying CFSR demonstrates that GISS is biased cold, when CFSR shows more warming than GISS does.

[skierinvermont]

I just think it's a cool product to look at for short-term trends. It's not meant to be used as a measure of long-term climate change, but it's fascinating to see the movement of the various airmasses that influence our weather, the impact of ENSO on Australia and South America, and the major cold wave in the Western US. I'm definitely very interested in the fact that nearly all of the Southern Hemisphere appears to be experiencing below average temperatures this February, looks as if the impact from the Strong Niña has been much more powerful down there. Also, the Northern Hemisphere which was responsible for much of the warming in the past few years, appears to be cooling down quickly, especially in the higher latitudes like the Canadian Prairies and Russia which were some of the fastest warming regions in the last 30 years. I wonder if the high-latitude cold wave is just a fluke or part of a permanent trend towards more availability of arctic air in those regions. I've certainly noticed that the cold air seems "fresher" this winter, might be part of the trend or just coincidence.

I would be wary of the ECM analysis for global temperature...it has a documented warm bias of nearly 1C at 850mb in North America, and close to 1.5C at the surface. I'm not sure how this affects the global initialization/reanalysis data as compared to forecasting, but it makes the observations a bit suspect. When I forecast in the longer range off of the Euro, I always adjust for these well-known biases which have been shown in studies.

I also wonder if the GFS is showing more realistic temperature trends now since it just had an upgrade during Spring 2010...the .33C/decade trend, which seems excessively warm compared to all other data, may be a result of how bad the model was before the NOAA put in the latest improvements last April. I've definitely noticed the GFS has been much more useful at showing pattern changes in the long-range this winter, and it's also had some coups on major storms such as the adjustment west with the Boxing Day Blizzard.

Off topic.. but I would love to see the stats showing the ECM warm bias.

Keep in mind that if the bias is consistent, it wouldn't affect trends.

The advantage of the ERA and CFS reanalyses is that they use a consistent data assimilation method throughout their entire period and we know the model physics used are state of the art and predict weather better than any time in history. The disadvantage is that it doesn't actually use surface observations. In addition, the number and type of observations fed into the models changes over time even though the model itself remains consistent (using todays ECMWF/GFS to reanalyze the past). It's just one more piece of confirmational evidence. I believe the ECMWF analysis agrees very closely with GISS.

[nzucker]

Off topic.. but I would love to see the stats showing the ECM warm bias.

Keep in mind that if the bias is consistent, it wouldn't affect trends.

The advantage of the ERA and CFS reanalyses is that they use a consistent data assimilation method throughout their entire period and we know the model physics used are state of the art and predict weather better than any time in history. The disadvantage is that it doesn't actually use surface observations. In addition, the number and type of observations fed into the models changes over time even though the model itself remains consistent (using todays ECMWF/GFS to reanalyze the past). It's just one more piece of confirmational evidence. I believe the ECMWF analysis agrees very closely with GISS.

I'm not sure where to find the charts of the ECM warm bias; there's a graphic that looks like a sounding which shows bias at different levels of the atmosphere. I believe it was posted in the NYC thread during the lead-up to the 1/28 snowstorm when I claimed that the European track of the system would be snow for NYC metro even though the model was showing rain verbatim. It was a successful argument as Central Park received 19"..in any case, the charts are cool to see but not really necessary. You and I both know the ECM has a warm bias since we've been doing weather forecasting for ages...it's one reason I predicted NYC to hit 90F during the September heat wave and failed, as the ECM was showing +20C 850s which really verified around +17C and didn't allow Central Park to get to 90F with the lower sun angle. After that debacle, I always take 1C off whatever the ECM shows at 850mb, and more at the surface, and it's worked well for me.

How do the ERA and CFS analyses work if they don't use surface observations? I thought a network of surface obs was integrated into the models along with satellite data and buoys. Is this network of surface data not available for temperature reanalysis?

I wouldn't really say the models remain consistent...there's been so many changes to the programming over the years, which have generally improved the GFS and ECM significantly but may be a limiting factor in using the data for reanalysis of climate trends. The ECM resolution was made finer in the 2010 upgrade, the GFS had various changes made in the April upgrade including better pressure predictions for tropical storms, reduction of QPF bombs, etc...so the GFS we see this winter isn't really the GFS we saw last winter even if the maps look the same.

[skierinvermont]

I'm not sure where to find the charts of the ECM warm bias; there's a graphic that looks like a sounding which shows bias at different levels of the atmosphere. I believe it was posted in the NYC thread during the lead-up to the 1/28 snowstorm when I claimed that the European track of the system would be snow for NYC metro even though the model was showing rain verbatim. It was a successful argument as Central Park received 19"..in any case, the charts are cool to see but not really necessary. You and I both know the ECM has a warm bias since we've been doing weather forecasting for ages...it's one reason I predicted NYC to hit 90F during the September heat wave and failed, as the ECM was showing +20C 850s which really verified around +17C and didn't allow Central Park to get to 90F with the lower sun angle. After that debacle, I always take 1C off whatever the ECM shows at 850mb, and more at the surface, and it's worked well for me.

How do the ERA and CFS analyses work if they don't use surface observations? I thought a network of surface obs was integrated into the models along with satellite data and buoys. Is this network of surface data not available for temperature reanalysis?

I wouldn't really say the models remain consistent...there's been so many changes to the programming over the years, which have generally improved the GFS and ECM significantly but may be a limiting factor in using the data for reanalysis of climate trends. The ECM resolution was made finer in the 2010 upgrade, the GFS had various changes made in the April upgrade including better pressure predictions for tropical storms, reduction of QPF bombs, etc...so the GFS we see this winter isn't really the GFS we saw last winter even if the maps look the same.

Actually I'm not so sure that surface data is put into the models.. at least the GFS and Euro. Sort of makes sense given even the initial surface temps on the GFS are usually wrong.

The CFS reanalysis doesn't use the 1985 GFS to reanalyze the temps in 1985... it takes the current day GFS an reanalyzes 1985 temps with that using the same data that was fed into the GFS that was operational in 1985. That is why I say the model remains consistent throughout the reanalysis. However, the data being fed into the model is different. As you can imagine, more data is available today than in the 1980s. A good demonstration of what I am saying is the chart on page 1041:

http://journals.amet.../2010BAMS3001.1

Basically what the chart shows is the 5-day forecast verification scores for the CFSR (which basically uses present day GFS model physics and data assimilation although not exactly, I'm still not totally clear on the difference) and compares it to several other models. One of the comparisons is the GFS that was used at that time. Another (labeled R1) is the old NCEP-NCAR reanalysis that was based in the 1994 version of the GFS.

Basically the chart is a test of the model physics and data assimilation. If it can accurately predict the weather 5 days out, it must be fairly accurate in simulating the initial conditions. *** Note I am not saying it's a substitute for actual observations. The actual data of the CFSR is the initial conditions, however, the ability to take these initial conditions and predict the weather 5 days into the future (say use the initialization for 00z Nov 21 1984 to predict 00z Nov 26 1984 using the CFSR which uses present day GFS model physics) is a good test of the accuracy of the CFSR. The CFSR must have a fairly accurate representation of initial conditions on Nov 21 1984 if it can use that to predict the weather on Nov 26 1984. Likewise, it must have an accurate representation of initial conditions on Nov 26, 1984, if it can use that to predict the weather 5 days after that. Etc.

As we can see, the operational GFS (red line) steadily improves. Today's GFS is vastly better than the GFS that was in use in the 1980s.

The GFS line intersects the R1 line for the NCEP-NCAR old reanalysis (blue line) in the early to mid 1990s. That makes sense too because the old NCEP-NCAR reanalysis (which the CFSR replaces) used a version of the GFS from 1994.

The GFS line intersects the CFS reanalysis (black line) in the last 2000s, because the CFSR is basically running an up to date version of the GFS (I guess there are some differences).

We can also see that the CFSR remains consistently high in terms of its ability to predict the weather 5 days into the future, but does show some slow increase. The increase is attributable to the fact that more data is available to assimilate into the initial conditions than there was in the 1980s. But even in the 1980s, the CFSR has a high 5-day verification score. This indicates that there is enough data going into the CFSR initial conditions, even for the 1980s, to accurately forecast the weather 5 days later.

I anticipate that the CFSR will be replacing the old NCEP-NCAR reanalysis. Of course, it's still not as good as ERA-40 or ERA-Interim. ERA-40 uses the ECMWF physics so naturally will be better than NCEP. It goes all the way back to the 50s, however it suffers the same problem the CFSR does - less data the farther back you go. ERA-Interim is the highest quality and most appropriate for climate monitoring.. it uses the same model physics and essentially the same input data throughout the entire reanalysis. In other words, it maintains a completely consistent methodology throughout which is essential for climate monitoring (unless you know that changes to your method won't affect your results). However, it doesn't begin until 1989, when there is widespread satellite data.

[nzucker]

I anticipate that the CFSR will be replacing the old NCEP-NCAR reanalysis. Of course, it's still not as good as ERA-40 or ERA-Interim. ERA-40 uses the ECMWF physics so naturally will be better than NCEP. It goes all the way back to the 50s, however it suffers the same problem the CFSR does - less data the farther back you go. ERA-Interim is the highest quality and most appropriate for climate monitoring.. it uses the same model physics and essentially the same input data throughout the entire reanalysis. In other words, it maintains a completely consistent methodology throughout which is essential for climate monitoring (unless you know that changes to your method won't affect your results). However, it doesn't begin until 1989, when there is widespread satellite data.

I don't think the objective of these reanalysis projects is to measure changes in global temperatures; from the report cited, it seems their main use is as input for seasonal forecasts, such as the CFS and the national outlooks published by NOAA every winter. Thus, we're probably being a bit picky about the details. Given the lack of observational data included in ERA and CFSR, I think GISS and Hadley are the most appropriate for surface trends in global temperature, not the ERA and CFSR analyses. I just wish we had a more comprehensive monitoring network over the Arctic and Africa so we didn't have to make large-scale extrapolations or leave certain areas blank, considering the many questions about the satellite data and whether the lack of simultaneous nadir overpass is causing a lower trend in troposphere temperatures than the reality. Despite our vast meteorological knowledge and the advancements in computer modeling, we still don't have a way to track the global temperature with great accuracy.

I don't know if you're the one to ask, but does the ERA and CFSR try to use 2011 quality data to determine 1980s temperatures? For example, if the 1985 GFS had a resolution of 100km compared to 40km today, would extrapolations be made in order to allow 1985 data to be plugged into the current model with its higher resolution and greater network of satellites and surface stations observing weather conditions? And what is done about missing data in sparse areas of the world? Is it a GISS method of extrapolation or a Hadley method of just leaving unknown areas out?

Bethesda's approach of using the GFS for short-term variation in global climate seems appropriate to me regardless of the long-term trend being dubious on CFSR. I think it's a good tool to show monthly trends in which areas are seeing below/above average temperatures and what impact ENSO is having on temperatures in the tropical latitudes. So I don't find it inconsistent that he posts this data yet criticizes the long-term trend it shows, given that these models are much better suited for immediate temperature anomalies than 30-year trends.

[skierinvermont]

I don't think the objective of these reanalysis projects is to measure changes in global temperatures; from the report cited, it seems their main use is as input for seasonal forecasts, such as the CFS and the national outlooks published by NOAA every winter. Thus, we're probably being a bit picky about the details. Given the lack of observational data included in ERA and CFSR, I think GISS and Hadley are the most appropriate for surface trends in global temperature, not the ERA and CFSR analyses. I just wish we had a more comprehensive monitoring network over the Arctic and Africa so we didn't have to make large-scale extrapolations or leave certain areas blank, considering the many questions about the satellite data and whether the lack of simultaneous nadir overpass is causing a lower trend in troposphere temperatures than the reality. Despite our vast meteorological knowledge and the advancements in computer modeling, we still don't have a way to track the global temperature with great accuracy.

I don't know if you're the one to ask, but does the ERA and CFSR try to use 2011 quality data to determine 1980s temperatures? For example, if the 1985 GFS had a resolution of 100km compared to 40km today, would extrapolations be made in order to allow 1985 data to be plugged into the current model with its higher resolution and greater network of satellites and surface stations observing weather conditions? And what is done about missing data in sparse areas of the world? Is it a GISS method of extrapolation or a Hadley method of just leaving unknown areas out?

Bethesda's approach of using the GFS for short-term variation in global climate seems appropriate to me regardless of the long-term trend being dubious on CFSR. I think it's a good tool to show monthly trends in which areas are seeing below/above average temperatures and what impact ENSO is having on temperatures in the tropical latitudes. So I don't find it inconsistent that he posts this data yet criticizes the long-term trend it shows, given that these models are much better suited for immediate temperature anomalies than 30-year trends.

Yes, this is why, as you'll note, I have said in every post that I would not substitute ERA or CFSR for Had or GISS. I just think it's ironic (and logically impossible) that Bethesda is arguing GISS is biased cold based on CFSR data even though CFSR shows more warming than GISS does. He also has argued that this model data must be accurate since it is used to predict the weather.. which both you and I are saying it's not (I mean it must have some decent accuracy but still not substitute for actual observations). ERA-Interim is probably pretty good for climate monitoring since it uses consistent modeling and consistent input data. Bethesda hasn't criticized the long-term trend of CFSR (to the contrary in fact he argued that the GFS inits must be accurate since they are used to predict the weather), he doesn't even seem to understand that CFSR (which is the data source he has been quoting from Maue's website) shows more warming than GISS or HadCRUT in the long term. No doubt if he did realize that CFSR/GFS init show more warming in the long run than GISS or Had he would change his mind about its accuracy (and agree with what both you and I are saying regarding its accuracy or lack thereof).

Regarding your question, you understand from my last post that CFSR and ERA use current GFS and ECMWF model physics, right? I'm pretty sure that the raw data that was used in the 1985 GFS can be just as easily plugged into the modern GFS/CFSR. Post-1980 there's satellite data was/is used. There's probably also satellite data prior to the 1980s. So that doesn't require extrapolation.

I don't really see how the changes in grid spacing mean you will have to extrapolate. Satellite data has higher resolution than either the 1985 GFS or the modern GFS. So in both cases you are sort of doing the oppostie of extrapolation - you are taking near-infinite resolution and making it more coarse by breaking it into 40km or 100km grids.

Radiosonde data on the other hand is a single point of data that has to be extrapolated whether you are are using 40km or 100km resolution for your weather model.

[BethesdaWX]

As you can see from the following chart, the one you have been referencing from Ryan Maue's website, it is entitled "NCEP CFSR 2m Raw Temp Anomalies"

NCEP CFSR stands for the "NCEP Climate Forecast System Reanalysis."

The most recent version of this reanalysis was published by the American Meteorological Society.

http://journals.amet.../2010BAMS3001.1

It is of course constantly updating with every run of the GFS.

As you can see on page 1040 of the above link, the trend of the NCEP CFSR from 1979-2009 was 1.02K/31yrs or, .33C/decade. This is significantly faster warming than GISS or HadCRUT.

As I said originally, the exact same product that you are looking at and saying "look it shows -.6C anomalies next week" shows a warming trend of .33C/decade over the last 31 years (based on a linear regression calculation. The NCEP CFSR is actually notoriously warm among all temperature indexes. This is the exact same product that you have been quoting from Maue's website.

This is why I find your sudden fascination with the NCEP CFSR product from Maue's website so ironic. It is literally the warmest data source available for the last 30 years.

Why do you think my post has anything to do with debunking/disproving a warming trend? This has nothing to do with the long term trend. This is the 2011 temps thread, not a warming trend AGW/CO2/Solar/pigsh*t jackass thread.

Either way, you completely missed the point of my post, in comparing GISS to GFS..(CFS/CFSR is a renalayisis product FYI)

[nzucker]

Yes, this is why, as you'll note, I have said in every post that I would not substitute ERA or CFSR for Had or GISS. I just think it's ironic (and logically impossible) that Bethesda is arguing GISS is biased cold based on CFSR data even though CFSR shows more warming than GISS does.

Do you mean GISS is biased warm?

If that's the case, I can understand what he is saying for very short-term trends, that GISS has been showing an anomaly of like .4C during a Nina where the models are showing it to be much colder. (Of course Feb is the most important month to judge given what we've seen, and we're not there yet...) If the models that have been generally mild-biased during the past 30 years are showing extreme cold, that could be significant. Also, the latest outbreaks are probably hitting into places with more data like Western Russia and the NW US.

[skierinvermont]

Do you mean GISS is biased warm?

If that's the case, I can understand what he is saying for very short-term trends, that GISS has been showing an anomaly of like .4C during a Nina where the models are showing it to be much colder. (Of course Feb is the most important month to judge given what we've seen, and we're not there yet...) If the models that have been generally mild-biased during the past 30 years are showing extreme cold, that could be significant. Also, the latest outbreaks are probably hitting into places with more data like Western Russia and the NW US.

Yes - that's what I meant. Even in the short term it doesn't make sense because there is still correspondence. You are again forgetting to adjust the baselines. Once you adjust the baseline (subtract .32C from GISS) and average across the whole month, CFSR agrees with the GISS monthly anomalies. So far the month has only averaged around zero on the CFSR/GFS and JRA. That would put us at .32C on GISS. GIven the drop for the last 10 days, I guessed GISS would come in at .2-.3C. Even if it doesn't agree exactly, it doesn't mean anything "significant" since both sources are somewhat imprecise for a single month, but have agreed month to month for 31 years.

I would like to make one clarification to an earlier post.. the .33C/decade for CFSR was land only. This is still higher than the GHCN land-only trend of .30/decade but not by much.

[Mencken_Fan]

6 foot 4, but I'm actually the shortest dude in my family.

My bro who helped me train is 6 foot 7, my other bro is 6 foot 5, my dad is 6 foot 10. I'm actually heading out to McHenry Maryland to visit the family, been years since I've seen them

Wow...sounds like a family of giants!

BTW.... I went to Walt Whitman High School; are you familiar with that are? I grew up in Sumner which is on the left side of outbound Mass. Ave. about a mile and a half from the D.C. state line. Do you know the Mass. Ave. area? How about Goldsboro Road or River Road?

When I was a kid, lower Montgomery County was mostly heavily-wooded rural with just a few neighborhoods (I was 13 when the Beltway opened - with 2 lanes each side - and very little traffic, lol.) Extreme lower MoCo (Mass. Ave. extended area) had D.C. addresses even though we were physically in Maryland. We were served by the Friendship Heights Post Office aka "Washington 16, D.C." - before zip codes came into being. With the advent of zip codes we became "Washington, D.C. 20016".....which finally became "Bethesda, Md. 20816."

People used to ask me: "If you live in Maryland, how come your address is Washington, D.C.?" I had to explain, over and over and over....lol.

Have a great week.

[Mencken_Fan]

Was told Via email thru ECMWF pay site my brother has. I don't see why he would lie to me, he never has before.

"Around -1C" was the quote.

I have a hunch the models are overdoing things, but regardless, the globe is about to cool to levels not seen for decades.

We have an moderate la nina, weak -PDO raging +AMO, and yet we see this ridiculous drop??? Solar definitely having its impact.

To think, this is only Intracycle cooling

During the last cold PDO phase global temps didn't plunge; they only stopped rising (in what a chart technician would call a "period of consolidation of the previous gain." If I were an odds-maker, I'd make the same scenario the odds on favorite for the next couple of decades. (Every aberration along the way will of course result in hoops and hollers from one side or the other.)

I'll probably go to my grave never finding the ultimate answer, lol.