|
Post by acidohm on Feb 27, 2019 19:25:29 GMT
www.quantamagazine.org/cloud-loss-could-add-8-degrees-to-global-warming-20190225/Yes the latest from climate 'scientists' is that if you heat a water planet by addition of sufficient quantities of the magic molecule CO 2 the water will stop evaporating and forming clouds! And everything will fry due to the lowered albedo just like it did previous times when CO 2 was ten times what it is n....... oh wait... I must admit I have seen some extreme CumuloNimbus I have yet to see one broken up by heat. But if its been broken up how could you see it?? š
|
|
|
Post by nautonnier on Feb 27, 2019 20:35:30 GMT
The real problem for all their clever modeling is the weather in the ITCZ. That is ALWAYS hairy to fly through even if you cheat and cut across the Gulf of Mexico. So the Cb in the Hadley cells there can reach to 70,000 ft or more with eye watering convection - yet none of these 'suck dry air in' issues. There's a bit too much water about for that - especially when the SST's get hot evaporation and all that...
|
|
|
Post by sigurdur on Feb 28, 2019 3:36:51 GMT
|
|
|
Post by nautonnier on Feb 28, 2019 11:05:08 GMT
There was an interesting post in WUWT by Willis where he pointed out that the agreement - the probability of the same result by chance - of a study should be modified by Bonferroni Correction. It would seem to apply in this case and Santer's claims of significance - have I misconstrued its application? wattsupwiththat.com/2019/02/25/labitzke-meets-bonferroni/
|
|
|
Post by blustnmtn on Feb 28, 2019 11:07:44 GMT
There was an interesting post in WUWT by Willis where he pointed out that the agreement - the probability of the same result by chance - of a study should be modified by Bonferroni Correction. It would seem to apply in this case and Santer's claims of significance - have I misconstrued its application? wattsupwiththat.com/2019/02/25/labitzke-meets-bonferroni/I donāt think you have.
|
|
|
Post by nautonnier on Feb 28, 2019 11:12:22 GMT
One of the replies led to this interesting paper. "It follows that this synthesis of empirical data conclusively reveals that CO2 has not caused temperature change over the past 38 years but that the rate of change in CO2 concentration has have been influenced to a statistically significant degree by the temperature level. Note that it is not likely for a rise in CO2 concentration to cause the temperature to increase and for the temperature level to control the rate of change of CO2 concentration as this would mean that there was a positive feedback loop causing both CO2 concentration and temperature to rise continuously and the oceans could have evaporated long ago."climateauditor.com/
|
|
|
Post by Ratty on Feb 28, 2019 11:24:33 GMT
|
|
|
Post by nautonnier on Feb 28, 2019 14:14:41 GMT
I saw that as it affects 'the day job'. I think Machine Learning like BlockChain is oversold. Not only is the repeatability suspect but the training is suspect too. For example - whoever would train the system to avoid a pedestrian pushing a bicycle across the fast lane of a freeway? THUD Oh... there's a case we should have taught - note to self add pedestrians pushing bikes. Also how do you verification test a system based on machine learning? For safety critical items that is a requirement - but you can't as you don't know how it learned. If I have to have triplexed systems for safety I have to ensure that they are different systems but come to the same result - how are 3 machine learning systems going to do that as the do this then that type of specification does not apply?
|
|
|
Post by nautonnier on Feb 28, 2019 15:24:28 GMT
"Lying with science: a guide to myth debunking Pseudoscience is on the rise ā and the media is completely hooked Matt Ridley
Increasingly, in a crowded market for alarm, it becomes necessary to make the scares up. More and more headlines about medical or environmental panics are based on published scientific papers, but ones that are little more than lies laundered into respectability with a little statistical legerdemain. Sometimes, even the exposure of the laundered lies fails to stop the scare."www.spectator.co.uk/2019/03/lying-with-science-a-guide-to-myth-debunking/A pity that Matt Ridley uses Wakefield as an example as there are far better ones. If you do an internet search on: Gut Brain Connection You will find many many papers now that support Wakefield's primary hypothesis. He was struck off primarily for incorrect procedures and failure to follow human trial protocols not for his claimed results (although that gave the vested interests the incentive to chase him). However, as Ridley says, to add to the Machine Learning debacle we now have clickbait myths.
|
|
|
Post by blustnmtn on Mar 1, 2019 23:08:51 GMT
J. Curry has an excellent critique of the Santer paper: judithcurry.com/2019/03/01/critique-of-the-new-santer-et-al-2019-paper/#more-24778āsince I have reason to doubt the validity of the Santer et al. model I donāt accept their conclusions. They havenāt shown what they say they showed. In particular they have not identified a unique anthropogenic fingerprint, or provided a credible control for natural variability over the sample period. Nor have they justified the use of Gaussian p-values. Their claim to have attained a āgold standardā of proof are unwarranted, in part because statistical modeling can never do that, and in part because of the specific problems in their model.ā
|
|
|
Post by Ratty on Mar 1, 2019 23:20:21 GMT
J. Curry has an excellent critique of the Santer paper: judithcurry.com/2019/03/01/critique-of-the-new-santer-et-al-2019-paper/#more-24778āsince I have reason to doubt the validity of the Santer et al. model I donāt accept their conclusions. They havenāt shown what they say they showed. In particular they have not identified a unique anthropogenic fingerprint, or provided a credible control for natural variability over the sample period. Nor have they justified the use of Gaussian p-values. Their claim to have attained a āgold standardā of proof are unwarranted, in part because statistical modeling can never do that, and in part because of the specific problems in their model.ā She sounds uncertain.
|
|
|
Post by missouriboy on Mar 2, 2019 1:45:17 GMT
J. Curry has an excellent critique of the Santer paper: judithcurry.com/2019/03/01/critique-of-the-new-santer-et-al-2019-paper/#more-24778āsince I have reason to doubt the validity of the Santer et al. model I donāt accept their conclusions. They havenāt shown what they say they showed. In particular they have not identified a unique anthropogenic fingerprint, or provided a credible control for natural variability over the sample period. Nor have they justified the use of Gaussian p-values. Their claim to have attained a āgold standardā of proof are unwarranted, in part because statistical modeling can never do that, and in part because of the specific problems in their model.ā She sounds uncertain. As uncertain as a heart attack.
|
|
|
Post by nautonnier on Mar 2, 2019 10:27:18 GMT
|
|
|
Post by nautonnier on Mar 2, 2019 10:31:56 GMT
For anyone interested these should show detail similar to the 'ClimateGate' CRU emails.
|
|
|
Post by Ratty on Mar 2, 2019 12:15:57 GMT
For anyone interested these should show detail similar to the 'ClimateGate' CRU emails. I have the first and second tranche of the leaks but not enough time in the remainder of my life to do a comparison. Random sample: ... with high confidence?
|
|