15 Jul 2008

Underwater in Wisconsin

A great post by a farmer describing rains (10 inches in 36 hours!), floods and snows and how to cope:
It really is not a question of "organic" vs. "conventional"; it is a question of being a part of the world around us, rather than trying to conquer the world. We need to fit in, just as farmers have fit in for thousands of years. They made mistakes and took their knocks like we did last week. We lost some soil in a few fields, but no more than the "no-till" farmers that rely on chemicals rather than cultivation to control their weeds. I know that none of the toxic chemicals headed for the Gulf of Mexico in the flood waters washed out of our fields, and there is some comfort in that.
Bottom Line: Farming is a tough job. If global warming makes storms worse (yes, it will), then it's going to get tougher.

10 comments:

Kevin Dick said...

I don't think the science is even close to settled on how higher temperatures will affect storm frequency and severity. Here are a couple of other Grist links from this year pointing to model-based research showing the opposite:

http://www.grist.org/news/2008/01/23/hurricane/

http://www.grist.org/news/2008/04/14/hurricanes/

Then there's Pielke Jr.'s research showing that the actual effects of hurricanes are remaining constant (normalized for wealth levels and population distribution):

http://sciencepolicy.colorado.edu/admin/publication_files/resource-2476-2008.02.pdf

Once again illustrating that we're dealing with a lot of uncertainty with respect to AGW's impact.

David Zetland said...

Kevin -- first grist link: "the research is based on flawed data that was rejected by scientists". Second link: "much remains to be figured out when it comes to warming and hurricanes".

So there is (at best) some doubt on hurricanes, but there is NO doubt on higher variability in rain/snow/runoff, etc., which is where Wisconsin fits in (no hurricanes there!). The climate FACTS as well as models for California are all about higher variation.

I'm sticking with GW = "yes we will" have more variation.

Kevin Dick said...

I think your objections to the Grist links don't make Bayesian sense. Scientists commonly claim that other the scientists screwed up when contradictory results emerge. So such claims don't carry any Bayesian weight with me until they actually demonstrate the alleged flaws in a peer reviewed paper. The telling point is the second article is co-authored by one of the first scientists to propose a link between hurricanes and AGW. When scientists change their minds, that carries quite a bit of Bayesian weight with me.

So your "at best" assessment is off IMHO. Looks like a 50-50 proposition to me. I could see someone who'd reviewed the research holding a 60-40 or 70-30 prior. But 80-20 or 90-10 looks unjustified.

I think there's a lesson to be learned here with regards to your assertion that there is "NO doubt" about higher variance precipitation. It does seem very likely given the current model results, but I thought that about hurricanes too until a few months ago. I'd give it an 80-20 at this point. I could see you believing 90-10 or 95-5. But 99-1 or 100-0? Seems unjustified.

For example, there's no evidence of greater effects on people from extreme weather:

http://www.csccc.info/reports/report_23.pdf

Flood risk seems to go either up or down depending on regional factors:

http://www.agu.org/pubs/crossref/2007/2006WR005099.shtml

Given this, are you really absolutely free of doubt? This is important because it affects what one thinks is good policy.

David Zetland said...

kevin -- good analogy:

Here are my ratios (prob of bad/prob of good) from GW:

Hurricanes -- 80/20
Icemelt/sealevel -- 99/1
Floods -- 95/5
Storm variation -- 90/10
Snowpack - 95/5

Affects on environment -- 100/0
Affects on flora/fauna -- 95/5
Affects on humans -- 80/20

Kevin Dick said...

Thanks. I assume those are all conditioned upon GW occurring? What would you put the chance of AGW as a significant source of warming? For me, it's somewhere around 50-50.

For the record, here are my ratios (conditioned on significant warming from any source):

Hurricanes -- 50/50
Icemelt/sealevel -- 90/10
Floods -- 67/33
Storm variation -- 67/33
Snowpack - 90/10

On effects, it's almost by definition 100-0 if you define effects as change. For significantly increased human mortality rates due to weather, I put it at 50-50.

So it looks like my priors are significantly lower than yours. Robin Hanson would say that we should work out the source of those differences, but I'm skeptical his procedure can be made to work in practice. But this is nevertheless useful info for me.

David Zetland said...

AGW as significant source? 100/0

Robin's method is good for opinions, but not a draw from a probability distribution (or, even worse, uncertainty, which has no PDF).

Kevin Dick said...

Looks like you come from a different school of Bayesianism than I do ;-)

The first lesson was never say you're 100% sure. We could actually get an infinitely negative grade for doing so and being wrong. This happened to two people in one of my graduate decision analysis courses. But I think the prof took pity on them.

Would you really bet at infinite odds on AGW being true? Your entire net worth versus $1? How about 1000:1?

Oh, the final lesson was that there's no difference between an opinion and a probability distribution. Uncertainty is always in the mind. I know that frequentist schools that differ on this. But I've never been convinced by their arguments.

David Zetland said...

Kevin -- ok, I'll see you and raise you -- 99/1 odds on AGW :)

I'm not going to agree with your statement on uncertainty. The Knightian version is, by definition, without a PDF, which means that it may be in the mind but not on the mind, i.e., Rumsfeld's unknown unknowns.

Kevin Dick said...

99:1 seems fair enough for someone who is really sure. If there were a low transaction cost mechanism for doing so, I'd put together a bet. One of the hidden lessons from graduate school was that "certainty arbitrage" would be a good business ;-)

We were taught that once you acknowledge the existence of Knightian uncertainty and still act, your behavior reveals an implicit probability assessment. But I can see how that's hard to swallow.

David Zetland said...

Kevin -- (un)fortunately, I don't plan to be around for the bad stuff, so collecting on 20-50 year bets is difficult -- unless you're a vampire. :)

Post a Comment

Note: only a member of this blog may post a comment.