PODCAST with Dr. Roy Spencer – Climate models and coronavirus models are being used to set public policy. Both have proven to be failures. It’s that old “uncertainty monster” again.
Climate models and coronavirus models have both been used to predict the future, and both have been used to form public policy. In this podcast between Anthony Watts and UAH climatologist Dr. Roy Spencer, we examine the common denominator of failure they both share. We also look at why Carbon Dioxide hasn’t been visibly reduced due to the economic shutdowns.
“It’s feb 17th
The US death toll stands at 0
The US case count stands at 68.
The CEO of your New York city hospital wants to know
How many cases should he plan for?
How many beds will he need
simple question.
It’s feb 17th
The US death toll stands at 0
The US case count stands at 68.
The CEO of your New York city hospital wants to know
How many cases should he plan for?
How many beds will he needHow many ventilators?
These are real questions.
How do you answer a question about the future?
Hint? you have to model unless you have a time machine.
So What is your answer?”
If one truly understand the limitations of models(which they should if people’s lives are depending on their answer) then they should provide the model output, along with other real world realities that are unknown that will likely make the model wrong, so the user anticipates it being wrong and is prepared for that likely outcome.
Models could never predict the outcome of an unprecedented shut down(based on human behavior that has never been modeled before) along with an unprecedented virus with scant data about its transmission rate and other key elements needed to accurately predict the outcome.
Do use the models but use common sense in interpreting the output/projections based on applying good human judgement to make adjustments from the get go, along with not overselling their skill to yourself or others.
If people are shut in and the human interpreting the model that is giving their expert opinion knows that affect is not properly dialed into the model………should they not dial down the model projection?
We know that they did a poor job communicating that and spent to much time focusing on exactly what the latest models said. This is what misled people and why the models were widely criticized for being so wrong. The models actually were 100% correct. They predicted exactly what the humans programmed them to say. What the humans did not program them with made the humans using them wrong. Those humans should have known better………….should have been very aware of the not modeled, key information that was going to cause them to predict the right answer to the wrong problem.
As an operational meteorologist for 38 years, part of my brain wants to believe a weather model that forecasts an extreme weather event on day 14. Especially if its an individual operational model. But the part of my brain that remembers the great number of times of a similar scenario, that never happened tells my entire brain to adjust my thinking so that its skeptical. Still entirely aware of that extreme solution and that it MIGHT happen but using experienced judgment about model skill when applying its practical use 14 days out.
If somebody wanted to know how many hospital beds and ventilators would be needed very early in the pandemic, taking actions based on a model that will probably be wrong was counterproductive………….just like me, telling something that in 14 days, my weather model just told me we are going to have a blizzard over the state of Michigan and they should get prepared for it now.
Anthony
You expressed the opinion that weather forecasting models are better than they were 30 years ago. Would you attribute that improvement to better models, or to better and more comprehensive data resulting from Doppler radar and geosynchronous weather satellites? Also, I have read that there is a consensus that European computer models perform better than US models. If you agree with that, to what would you attribute the better performance?
Very cool
they did power analysis
https://youtu.be/Z2hfGcTokiY?t=191
Open source model
https://www.post-gazette.com/news/health/2020/03/27/COVID-19-coronavirus-University-Pittsburgh-hospitalization-FRED-pandemic-model-case-study/stories/202003230067
There is near-zero possibility that any virus-caused reduction of emssions of CO2 will be detected by analysis of CO2 mole fractions in the atmosphere.
The accuracy of such measurements is no better than +/- 1 ppm CO2 2 sigma, when theorists are looking for changes of 0.2 ppm, according to some public estimates. Geoff S
The most serious problem with the Climate Models is the political clampdown on the discussion of their methodologies and validity. I don’t think that has yet happened with the Covid Models.