In comments, JK asked:
Has anyone checked the trends in UAH/RSS vs the models? Also, I believe Dr. Pielke Sr, Jr. et al published a paper showing that the trends for NCDC et al are likely overstated due to boundary condition measurement issues.
Has anyone plotted their revised views vs. the models? I believe they don’t plot out a trend, just show an overestimation bias.
Sounds like an interesting project, but one for which I lack the technical skills.
The answer is: Chad has undertaken this challenge.
The task is time consuming because he has to first create the monthly temperature series, check that he’s done it all correctly, and perform comparisons afterwards.
But it appears he has replicated Santers computations for the time series, and is now cranking through various models. Full hypothesis testing must wait until he’s finished and he is wisely mostly posting all his verifications first.
Still, Chad did post some numerical values for Jan 2000- July 2009 TLT trends the CCCMA CGCM 3.1 T47 global and tropical TLT trends. These are:
| Run | Global C/decade |
Tropical Trend C/decade |
| 1 | 0.269 | 0.406 |
| 2 | 0.157 | 0.075 |
| 3 | 0.370 | 0.432 |
| 4 | 0.379 | 0.339 |
| 5 | 0.395 | 0.433 |
| 6 | 0.314 | 0.337 |
| Average | 0.314 | 0.337 |
| St. Dev. | 0.090 | 0.135 |
So, this model shows brisk tropospheric warming after 2000. However, Chad then turned to GFDL 2.0 CM 2 Run 1 and found it has a fairly flat trend after 2000. (I’d report the value for the trend… but I didn’t see it in the post. So, the omission is not intended to de-emphasize that result.)
The trends reported by RSS and UAH are near 0C/century during the similar period and are shown below:
Before making conclusions, it’s best to wait until Chad feels confident with everything and has also computed trends for all cases. He is carefully going through the model data and comparing his results for temperature trends to the values Santer published for periods before 2000. It’s time consuming. Each model will be different.
In the meantime, you can probably see the results for individual models as Chad churns them out! Visit Trees for the Forest.
Sorry to be confused, but UAH and RSS show zero warming (over that time period), but Chad’s runs show .3c/decade? Are they measuring the same things?
Originally, I was basically thinking of a graph along the lines of the one you posted, showing UAH/RSS trends (but preferrably back to the start of the satellite record).
What is Chad trying to show? I looked at his website and was rather confused.
Sorry for missing the point …
Chad’s runs aren’t measuring anything. They are processing the model data and extracting the trend associated with a simulation.
Chad is
a) Writing a program to extract the tropospheric temperatures from the gridded model data.
b) Comparing the temperatures series for the models to the ones Santer published. (This is a wise thing to do. It helps him verify he gets the same trends that Santer would have.)
c) Posting some temperature series and the comparison to show people his program is working
The part you are interested in is step… oh… f or g where Chad
g) Takes the temperature series he found, compares to measurements and tells us whether the measurements are consistent or inconsistent with the model projections.
I got to skip steps like a-c with surface temperature montly data because already processed temperature time series were available at the climate explorer. But Chad has to do all the other things. If you don’t know why he’s doing it, it can seem inexplicable.
If you read, over time, you can see how the temperature series look, and get a “preview” of the final comparison. For example: Every single run for Model CCCMA CGCM 3.1 T47 over predicts the observed tropospheric warming from 2000-july 2009. However, the next model he ran doesn’t seem to over predict. There are dozens of more runs. We can’t guess how they will compare to observations. So… we have to wait before making conclusions.
If watching individual time series as they come out doesn’t interest you, and all the careful checking confuses you, then you can wait until Chad gets to step…. “g”.
Ah, I understand. He is repoducing the models in the above run analysis to get their trends. I thought he was reproducing the temperature records (since I know sat measurements require some jiggering to get from raw numbers to final numbers).
Thanks for the clarification, I appreciate it.
I suspect he will also write a script to get the observations for the tropical troposphere. But, right now, he’s doing the model bits.
I know you don’t believe in using model data from before the projections were “locked” lucia, but UAH has had at least 144 months, or twelve years, of no warming now. Just making an observation.
Andrew_FL–
Sure. But the thing is, Chad is processing the data, and until he’s comfortable with his time series, we really can’t compare. What if as he processes more cases, he finds a bug? Etc.
That said: he’s providing quite a comforting level of detail. Also, it’s difficult to believe the models, on the whole, will predict no warming over 12 years. But, still, Santers conclusion stopping in 1990 was that the low-warming fell within the uncertainty bands of the models. Evidently, the runs are noisy over that time period. (Actually, they should be owing to the volcano eruptions.)
We’ll see what happens with more data particularly data that includes a nice lower variability stretch without a Pinatubo eruption.
Those time series are for the troposphere. I’m generating series for channels 2, 4 and 6. I’ve validated my program by comparing it to Santer’s data. Just a matter of sitting patiently while my slow computer processes 1-2 timesteps/second (each model has at least 3,000 timesteps).
Re: Chad (Comment#20278)
Hey Lucia and Chad,
interesting topic, please keep it going!
However it somehow reminds me of the SETI screensavers,
not only because it takes a while to calculate 🙂
Cheers,
LoN
Thank LoN.