Like last month, we are betting on UAH late this month. I’ve set the cut-off date to Nov. 30– but remember, ties go to the the first to bet a particular value. Here is the form. You know the drill. (Remember: it has now been established historically, bettors have bet on the ‘cool’ side. I have no idea if you can exploit this information to improve your odds of winning; I especially have no idea if you can use it this month.)
[sockulator(../musings/wp-content/uploads/2011/UAHBets5.php?Metric=UAH TTL?Units=C?cutOffMonth=11?cutOffDay=30?cutOffYear=2013?DateMetric=November 2013)sockulator]
Might as well talk about the weather.
Hmm… I’m seeing light snow out the window. Not unusual for November. ( Returning before hitting publish, I see the snow stopped. No accumulation yet. We’ll see.)
Speaking of November, I need to go buy turkey. Turkeys in the 16lb -whatever lb range are $0.39/lb at Butera ( $0.49/lb at Jewell. ) I always hunt for the smallest possible qualifiying turkey, and buy two: one to brine now and one to cook later in the year (usually around New Years Day). I know you are all wondering, why so small? After all, at $0.39 lb turkey is practically free. Answer: The large turkeys don’t fit in the brining bucket. Plus, Jim cooks this on the grill, and the smaller bird means less risk of over exposure to freezing rain.
I know your next question: given the small size, why buy only two turkeys? After all, surely a married American couple with no kids and whose relatives mostly eat turkey with their inlaws needs much more than 16 lbs of Turkey for Thanksgiving? Well, the ‘not enough turkey’ problem is solved by the brothers-in-law who usually also buy a a turkey, and nearly always chose larger one. This results in a “family” thanksgiving turkey count of two turkeys (which will be shared by 4 this year) and a frozen one used later in the year. With two turkeys, we’ll likely tally 36+lbs and plenty of left overs. Which we need because we host “Turkey Alfredo Day”, scheduled every Black Friday, relieving us of any temptation to visit crowded malls on that day. (Turkey Alfredo involves breaking out the pasta roller, lots of cheese, cream, etc.)
By Saturday or Sunday, the carcasses will be used to make turkey vegetable soup. (Tip: buy paper ‘to go’ cups, and freeze individual soups portion and add these to your brown-bag lunches. Unfreeze and heat the soup in a microwave oven. Yumm! )
Besides betting, I suggest everyone share Thanksgiving day recipes. Even thought I tend to make the same thing over and over– and to some extent, the brothers-in-law have now decided they are going to cook stuff too, I’m always game for suggestions. Especially suggestions for left-overs. 🙂
I placed my guess, and went all in. Given that my quatloo debt is approaching that of the treasury, I suppose I’m going to have to learn to cope with the sequestration of some of my quatloos. Come on UAH, baby needs new shoes!
I don’t do recipes so much as dishes. Recipes are too confining. One of my Thanksgiving favorites is creamed spinach with onion rings (not crisp fried onions) on top. Cream cheese and/or heavy cream are a must.
ugh. So far my prediction amounts to ‘up’ sometime between now and the end of the year or January.
I’ve got to see if I can’t refine that a little bit this week.
lucia,
I’m sure you are aware of this, but a lot of turkeys now come pre-brined (Butterball® frozen, e.g.). If the label states “contains x% of a solution”, you don’t want to brine it further, unless you like a lot of salt. A turkey labelled ‘fresh’ probably hasn’t been treated. Another tidbit: Frozen means the turkey has been stored below 0F. The turkey can be hard as a rock, as the meat freezes at about 26F, and still not need to be called ‘frozen’.
According to the information on butterball.com, all Butterball® turkeys, fresh or frozen ‘contain up to x% of a solution of water, salt, spices and natural flavor.’ For ‘fresh’ x=4 and frozen x=8.
DeWitt,
Jim uses a brining recipe by Steven Reichland’s cookbooks. It involves maple syrup, salt and so on. BBQing involves hickory chips. (He pointed out the hickory log he got to be sure I don’t toss it in the “burn” pile with sticks.) Commercial pre-brining is no substitute for this! In some sense I guess we are “re-brining”, but it’s still better.
I’m aware of the fresh/frozen issue. In iowa, the grocery store carried a brand of chicken that really had never been frozen, and those really, truly were better than merely “claimed not frozen and sold defrosted” chickens which taste no better than frozen turkey you defrost. (The defrosted for your convenience chickens can be more convenient if you didn’t plan far ahead and just pick up chicken near the last minute, but they don’t taste better.)
I could look for really-truly not frozen turkey, but I know lots of the claimed unfrozen are no better than “conveniently defrosted for the consumer”. So I buy the cheapest frozen turkeys I can get. Those are usually Jennie-O (which are also pre-brined). I have no idea why Butterball costs more. I’ve had both. I like both! I currently have 1 solid as a rock in my freezer. I’ll be hunting for a second one to use the $13 off coupon on. (That’s what gets it down to the $0.39 lbs price– if you score a small one that is just large enough to qualify.)
Re: lucia (Nov 22 16:44),
I use an Alton Brown Good Eats recipe, but I won’t use it on a turkey that has already been treated. My impression is that it’s getting harder to find untreated turkeys.
If you like smoked turkey, feast your eyes on this one from the Kreuz Market in Lockhart, TX, lots of other good stuff as well including smoked boneless prime rib. Unfortunately, I wasn’t able to get there when I was in Austin for the Formula 1 race last week.
Basting Turkeys is quite good at this time of year. However I see Steve McIntyre has already cooked Mr Way.
Prefer to talk about UAH, going down all the way to 0.17.
Reason poles are colder. less arctic amplification means it will lead to the rest of the world freezing.
UAH will be negative by February 2014.
Do you have ALDI in USA ? An Aldi Turkey would be 0.37 cents per lb, guaranteed.
Here’s an Australian recipe for boiled galah. I’m sure it will brighten up Thanksgiving Day.
Shoot, pluck and clean a galah. Drop it into a pot of boiling water, along with a stone. When the stone is tender, throw out the galah and eat the stone.
angech,
We have an Aldi here. It’s across the street from the Jewell. But I’ve never seen whole turkeys in the Aldi.
Hector,
I’ll stick with BBQd turkey!
This year, we are conidering a curry rub on the turkey, and roasting it with apples, peeled roange slices, herbs and potatoes in the cavity.
Hector,
Hilarious. Thanks for the chuckle.
Lucia,
I had never heard of brining a turkey before and have just looked it up. It seems to be a habit peculiar to the US and Canada. I looked for kitchen advice on the subject in French, Spanish and German. My statistically meaningless survey suggests its description can be found only in translations from American English or in French description from Quebecois sites or in Spanish descriptions from American Spanish sites. I could be wrong. Does the US have a different turkey species from Europe?
In the UK, it is typical to make a rich stuffing – with “instant infarction” levels of fats – and to leave the stuffed turkey overnight to benefit from some cold marinading effects. I cover the turkey completely with streaky bacon and cover with aluminium foil before cooking. Slow cooking of the turkey and frequent basting ensure that all of those disgustingly rich fats invade the turkey meat. The result is a delicious, moist turkey inculcated with the flavour of the stuffing.
Is the brining done so that you can get a moist turkey and yet keep your cholesterol levels out of the red?
Answering my own question – European domesticated turkeys were all bred from a (wild) Mexican sub-species, starting in the early 1500s. The domesticated birds were then taken back to the US by English colonists. So I guess we should be eating the same sub-species. Amazing what you can learn in climate science.
From wikipedia:
But you guys probably already knew that.
Re: Paul_K (Nov 23 09:05),
Cooking a turkey with cold stuffing in it is a recipe for an overcooked turkey or food poisoning. You simply can’t get the stuffing hot enough during baking to be sure that all salmonella are dead until the meat temperature is far beyond well done. The way around this is to cook the stuffing first and put it in the bird hot.
Basting doesn’t actually work either. All you’re doing is lengthening the cooking time by cooling off the oven every time you open the door.
Brining, OTOH, does work. We know this because a lot, if not most, turkeys in the US have been brined before packaging. You can also weigh the meat before brining and after cooking. A brined turkey gains weight during brining and weighs more after cooking than an untreated turkey. In other words, it’s juicier.
DeWitt,
Good point. I should have mentioned that the meat components are all pre-cooked before being added to the stuffing. Any danger if it exists comes from bacteria left on the interior of the turkey not being killed off in the cooking.
You seem to be saying that you can only cook a turkey with stuffing if either (a) you put the stuffing in hot or (b) you brine the bird to allow you a longer cooking time without drying out the bird. I assure you that I have not killed anyone or had a dried out bird yet, but I will certainly contemplate trying this process.
Paul_K
Looks like you found your answer. But I would emphasize that we Americans call turkey is a bird that is indigenous to the Americas. See http://en.wikipedia.org/wiki/Domesticated_turkey . Many traditional Thanksgiving dishes highlight foods that were indigenous to the Americans and not Europe. Examples: mashed potatoes, sweet potatoes, squash, cranberries. Oddly, while many people do eat chocolate on Thanksgiving and it is indigenous to the Americas, I wouldn’t call it “traditional Thanksgiving” so much. That said, when I was a kid, I insisted on Chocolate pie. 🙂 )
http://en.wikipedia.org/wiki/List_of_food_origins#Vegetables
Stuffing is also traditional here. Note: Dewitt is correct about the dangers of stuffing. Lots of people do it anyway. I follow Jacques Pepins recipe and cook the stuffing outside the turkey. But my choice isn’t entirely due to health considerations. It’s difficult to deal with the stuffing in the BBQd turkey. You also don’t accumulate much in the way of drippings when you BBQ which is a good reason to cook two turkeys (or get an extra turkey leg to make turkey broth. When making turkey gravy, chicken broth is a very poor substitute for turkey.)
Hmmm… that sounds good. Turkey is a very low fat bird. Brining helps keep it juicy but slathering with bacon would likely work too!
Brining definitely keeps it moist. But the cholesterol is not an issue in this household. My cholesterol readings are very, very low. ( very. The highest reading ever was 153, the lowest 127. I never remember the units. but as you can see…. low.) Jim’s cholesterol is also lowish.
Note that brining also adds sugar. And with maple syrup and various seasonings, rest assured that the brining affects the flavor. It’s sort of like smoking and seasoning affects ham! So, you brine for flavor.
I can attest to the fact that brining keeps meat moist. It also works for chicken. You can try brining a chicken in the fridge over night and see the result for yourself. However, if you are salt sensitive and have high blood pressure your doctor may advise against brining.
Paul_K
I should add that my mom always cooked stuffing inside the turkey and many, many other people also do so. I’ve always read all the health announcements but have never heard of salmonella outbreaks during or after Thanksgiving. (On the other hand, I’ve never checked statistics.)
But, as I said: I cook mine outside because putting it inside the BBQd turkey is dicey. Our grill doesn’t have the precise temperature control of an oven.
Re: lucia (Nov 23 11:15),
I don’t pay a lot of attention to some of the warnings about food preparation. I eat steaks rare to medium rare and eggs fried over easy (liquid yolks). It’s probably more important to prevent cross-contamination from uncooked food to cooked food. If you’re immune system isn’t compromised, though, it’s unlikely a little salmonella will cause problems. I actually prefer the taste of stuffing cooked separately.
It’s possible, I guess, that the extra thermal mass of stuffing inside the breast cavity keeps the breast meat a little colder and allows one to completely cook the legs and thighs while not overcooking the breast. I’ve never instrumented a turkey that thoroughly to see.
We baste our turkey (and chickens, too) internally without opening the oven by placing the bird breast down. The fatback melts and flows via gravity into the breast. Even an overcooked bird will have a moist, juicy breast. The downside is non-crispy breast skin.
Brining, in my experience, ruins the drippings for making gravy. I’m sure there is a fix for that, but I don’t like that much salt in the meat in any event.
Never stuff a turkey.
Another trick to save time on prep and cleanup is to roast sweet and regular potatoes with the turkey.
The point of placing fruit in the caity is to allow the high moisture content of the fruit to steam out, and to add the subtle flavors into the turkey meat. The potaots are a good addition, not only to save time, but to also increase the flaovr spectrum of the entire meal.
The bacon idea seems very interesting.
Has anyone heard of a curry rub on the turkey itself?
Humm… The Blackboard, where people talk turkey! :-0
Dewitt,
I’m not necessarily recommending my Mom’s turkey. I could tell stories…. Ok. I will tell a story. . .
When Jim and I were dating, he came to Mom’s for T-Day. Just as I opened the front door, the oven erupted in flames. I kid you not.
Why you ask? Well, a fellow teacher had told my mother that the “trick” to moist turkey was to place the roasting platter in (or tented with) … a …. paperbag. Or at least paperbag was mom’s understanding of what the fellow teacher told her. ( Those plastic/polymer roasting bags did exist at the time.)
Anyway, Mo decided she was going to make the moistest turkey ever and somehow or another rigged up some paper bag like contraption. (I am not making this up.)
Of course, 5/6 people in the family told her this was unwise. That the other teacher must have meant those new polumer roasting bags. But our views did not prevail. In the fullness of time, and just as Jim (who had never met Mom) arrived, the bag caught on fire. (No big surprise to 5/6 people in the family.)
Bizarrely enough, nothing else caught fire. Also, the turkey was fine. It even tasted good with no “burned paper bag flavor”.
Needless to say, Mom did not try that ‘trick’ ever again. I wish I could say she stopped listening to all the ‘tricks’ her school teacher friends ‘revealed’, the list of either totally useless or positively ridiculous ‘tricks’ she would implement is rather large.
BTW: Mom is actually a mostly competent person. And mostly she cooks pretty well, (though as it happens, I am allergic to her ‘best’ dish which is paella containing shrimp. I can’t eat crustaceans.) But still. This was the first time Jim met Mom. And we both agreed that any novel Turkey preparations tips from Mom were to be viewed with suspicion.
Also: Mom never brined her turkey. Alas, the teacher friends didn’t suggest that. Or if they did, she never tried it. Not sure why not. It really works.
Howard
This fix is to buy two turkey drumsticks, simmer them separately and use that broth for gravy. 🙂
Growing up, stuffing always went inside the turkey (or chicken). For large gatherings, there was an extra casserole to augment what could fit inside the bird. Due to the preferences of newer family members, the “outside” stuffing is now left to form a crunchy top, so we have “soft” and “hard” stuffing. I prefer the soft variety.
well…as defending Oct winner….I have placed my bet….only took me several years to win once!!
As a professional chef, I take a different approach to my turkey (both the 25+ pounders at work and the little 10-12 pounders at home).
The first thing I do is dismantle it – I take off the legs/thighs, remove the first two segments of each wing, and then cut the back off. What I’m left with is the breast (still on the bone) with the meaty wing segment on each side and the leg/thigh pieces. The back and outer wings go into a pot for stock along with the neck and giblets.
I then take the leg/thigh pieces and make what’s called a ballotine – I remove the bones and tendons (again into the stock pot), stuff with dressing and tie up into a roll.[There are some decent youtube videos on making this.]
Some advantages of all this are: I don’t need to brine the turkey (the white and dark meat are cooked separately eliminating the overcooking worry); the actual roasting takes about half the time (smaller pieces); I can cook at least part of the dressing inside the ballotine (it’s small enough to reduce salmonella worries); I can make the stock while the rest is cooking (allowing both basting with stock and making gravy from the stock); I can do all of this prep the day before; the final presentation is really good looking; and carving is much easier.
Cooking stuff wrapped in paper is a standard technique. In French, it’s called en papilotte Normally one uses parchment paper, which can withstand higher temperatures and is safe for food contact. To me, that’s more steaming than baking, sort of like putting a potato wrapped in aluminum foil in an oven and calling it ‘baked’.
I often use a sheet of parchment paper rather than corn meal to keep french bread from sticking to the peel. I can pull it out after a few minutes when the surface of the bread sets and oven spring is complete.
Lucia: I remember the paper-bag perfect turkey rumors in the days before brine and deep-frying became de rigueur with the smart set.
kch: Thanks! Does that work for chicken as well?
Lucia, “Also, the turkey was fine. It even tasted good with no “burned paper bag flavorâ€.
Turkey Flamebag? I recommend a beer butt turkey myself for on the grill. Not that I like Fosters, but it fits.
I really, really don’t like ‘roasted’ in a plastic bag turkey. It may be moist, but it’s always overdone. As far as I’m concerned that’s braising, not roasting. That’s fine for pot roast, but not for turkey. I suspect paper bag turkey isn’t all that different.
DeWitt,
I also don’t like bird cooked in the plastic bags. But at least they don’t burn. I think the “papillotte” idea could be fine, but (a) not with brown grocery store bags and (b) possibly not for something the size of a turkey. Sometimes, some applications don’t scale up. That said, I suspect there was some detail about the tip that wasn’t communicated to Mom. Because “Flamebag Turkey” is unexpected and scary.
dallas,
We make beer can chicken on the grill in the summer. It’s very good. I’ve never tried beer can turkey. Turkeys are so large. I know I buy two at Thanksgiving, but really we can’t be making turkeys regularly. Way too much food!
kch,
Your turkey sounds great!
Howard –
You could, but chickens are enough smaller that there’s generally no compelling reason to do all that if what you’re looking for is just a roast chicken. If you want to break down a whole chicken, the more common thing would be to serve the ballotines as one meal and the boneless supremes (breasts) as another.
Lucia –
Paper bag cooking actually has some tradition to it, including for turkeys. Kind of a poor man’s parchment paper, I guess. The key is to thoroughly grease the bag (lard is best) before inserting the turkey (or whatever), as well as to avoid too high a temperature for too long. It does help seal in flavour and moistness when done right.
kch (Comment #121479)
November 23rd, 2013 at 6:19 pm
Ina Garten made something like you describe today on her show. I sometimes think we try too hard with the traditional American Thanksgiving turkey when some of these other recipes would take less effort – to get it right – and be more tasty. This gutless cook might well attempt one on a date other than Thanksgiving.
http://www.barefootcontessa.com/recipes.aspx?RecipeID=481&S=0
kch,
Ahhh! I suspect this detail was not communicated.
This is where the problem might have arisen. After all, if you are serving a family of 6 plus guests, you often buy a large turkey. I don’t know the size of the Turkey Mom bought. But while I hunt for the smallest cheap turkeys, and absolutely never buy one larger than 20 lbs, she may have bought a larger one. Turkeys can easily be 25 lbs, and I think if you hunt you may be able to get larger than 25 lbs.
That said: while big Turkeys cooked whole do look impressive in photos, they take a long time to cook, need large pans, breast dry out and so on. (But then, you know that. That’s why you use your technique!)
Speaking of which
DeWitt Payne, ” It may be moist, but it’s always overdone.”
I think that is era dependent. During Turkeydryas, the 180F recommended minimum internal temperatures produced exceptionally dry Thanksgiving Day main courses leading to increased alcohol consumption and Turkey flamebag incidences. Now that the USDA has adjusted the internal Turkey temperature record downward to 165F, a mere 5 F above optimum moistness, your Poultry in a Bag experience is likely, (90% confidence interval), to be more rewarding. Luckily, the USDA over-estimate was an outlier, no need to lose confidence in the highly skilled and intelligent linear no threshold modelers associated with safeguarding our national food supply.
People might want to be aware of this paper, which has important claims. I’ve seen no discussion of it yet for some reason, and its math is beyond me.
http://www.pnas.org/content/early/2013/10/28/1313476110.full.pdf?with-ds=yes
lucia (Comment #121490)
November 24th, 2013 at 7:35 am
That was a funny bit about the turkey drop. You’d be too young to have heard the original telecast. It reminded of my turkey story (fair warning for readers here to go onto other posts).
My sons and I used to hear turkeys which were claimed to be wild in a nearby forest preserve but we would seldom see them as would be expected from wild turkeys. Much to my surprise a few years ago a flock of “wild” turkeys appeared in our backyard – and around Thanksgiving time. Out comes the camera to save this scene for posterity and I tell the family and neighbors about these “wild” turkeys. A few days later these turkeys reappear and this time they walk unafraid onto our patio and poop on it. My view of these birds changes on the spot from a wild creature to a not very wild nuisance that just made my patio a mess.
I heard later and had suspected this all along that these turkeys were probably feral turkeys that escaped domestication many years ago. For the the last 2 years these birds have been making almost a daily appearances in the yards like ours that border on a long line of trees running between the lots. My wife has “trained” the birds not to linger in our backyard and now they quickly cut through to the neighbors yards. I have seen these birds in a single line of 12 running down a sidewalk, for no good reason I can tell, like miniature ostriches. They are also decent fliers as I learned after applauding the yellow lab two houses down for chasing the birds from his yard.
Anyway, anyone in the Chicago area looking for a range fed and probably certifiably organic turkey for Thanksgiving that you can personally select from the flock let me know. If the authorities become involved in this transaction we can tell them that we are permanently relocating these “wild” birds.
Hmmm… I believe you live about 10 minutes drive from me. That said: the thought of defeathering a turkey and so on is not enticing to me.
Kenneth,
Wherever firearm based hunting is restricted, usually due to (human) population density, turkeys seem to flourish. On Cape Cod, I see lots and lots of them. If your turkeys actually can fly (gain altitude for considerable distance on their own power, not just run-hop-and-glide), then they probably really are wild turkeys, or maybe a hybrid of escaped domestic and wild turkeys. Wild turkeys fly quite well, and roost in trees to avoid predators…. like the many foxes and coyotes we have on Cape Cod.
.
Lucia,
A cousin of mine shoots several turkeys each year (he lives in upstate New York) and says that cleaning them is not bad…. of course, he thinks cleaning a 150 pound deer is not bad either. 😉
Jim Bouldin, thanks for that link. I’ve only had a chance to skim the paper, but it’s interesting. I’m not a fan of its comments on reproducibility though. It seems to leave out a number of reasons that can be impossible, such as a lack of detail in descriptions.
Anyway, it’s definitely good to see attempts to map Bayesian analysis to Frequentist. The conflict between the two schools of thought has caused so much unnecessary trouble. Mapping one to the other would help reduce that, and it would also help improve the general rigor of analyses.
Which, not to take sides or anything, the Bayesian school really needs.
Lucia, we cook three 22lb turkeys for Thanksgiving. 2 are deep fried after brining (juniper berries, apple, dark brown sugar, sage, thyme, rosemary, sea salt, cumin, orange peel and turkey broth). Each bird takes about 85 minutes in 360 degree peanut oil. The third bird is brined for 24 hours and roasted for 2.5 hrs, breast side down at 375 degrees. At the end of 2.5 hrs the bird is turned breast side up and finished at 450 degrees for ~ another 1.5hrs (breast temp=160 degrees). You get the crispiest skin if you first rub it with equal amounts of salt and baking powder. We make it with an oyster/spinach stuffing, half of which is baked in the oven and the other half stuffed into the birds main cavity in cheesecloth. The cheesecloth is easily removed and that stuffing is mixed with the oven roasted one and finished at 375 degrees. Mmmmm.
No problem Brandon, thanks for your comments.
My mind is decidedly not made up on the overall usefulness of Bayesianism, and I know a lot of others in that position too. However, a big part of that might stem from not understanding it fully, so I keep my mouth closed. On the bigger issue, I think the lion’s share of non-reproducibility in science stems from other causes, messy ones not neatly addressed by simple equations. And I like to think I can make some decent arguments that p = .05 is actually too strict, not too lenient.
I’ve arged before that in an abstract enough sense, Bayesian and Frequentist analysis can be viewed as the same thing – the difference is just in the implementation. I don’t think that’s a popular view though.
Anyway, a problem with Bayesian analysis is there are a lot more choices you can make with them. A lot of time it seems arbitrary which ones get made. That’s bad as every choice directly impacts the meaning of the results. That makes it difficult to ensure one interprets results properly. Given Frequentist analysis can do anything Bayesian can, it’s best just to pick whatever you’re most comfortable with.
I agree on the issue of non-reproducibility. I think the authors exaggerated the significance of their work in that regard.
As for p, .05 is too strict for some things and too lenient for others. It just depends on what you’re examining. If you’re okay with giving Frequentist heart attacks, you should argue against using a static p value at all. Instead, calculate the p value at which significance stops and give it as a % chance of truth.
“Wild turkeys fly quite well, and roost in trees to avoid predators…. like the many foxes and coyotes we have on Cape Cod.”
SteveF, you make good point about wild turkeys flying and I was thinking along your lines when I saw the birds in flight. Regardless I do not consider these birds or the Canadian geese that stay for the winters in our area and poop all over the side walks “wild”. We have coyotes in the area and I was thinking if those turkeys were not good flyers perhaps the coyotes could take of them.
Two years ago we had a coyote pup use a compost heap for keeping warm a couple of late fall days ago in the back of our yard. When I would simply turn the door knob on our back door his ears would perk up and he was ready to run. Now I consider that wild.
kenneth
By that measure (i.e. running away), many feral cats would be wild. Yet, they can be easy to tame particularly if they were once handled by humans if you are patient feed them and sit still while they eat. It’s often difficult to find a bright line between “wild” and “not wild”.
That said: coyotes are on the wild side of the line. Feral turkeys? Not sure.
SteveF,
I’ve discussed the possibility of Kenneth shooting a turkey for use with Jim. We both agree that at least with respect to our dinner, Kenneth can sit back and just leave his Turkeys alone. We don’t want to defeather the things. Maybe if the apocalypse should ever arrive, we’ll hone our turkey cleaning skills. But in the meant time the $0.49/lb solid as a rock frozen turkeys will do for us.
Brandon, yes, I can see where frequentist and Bayesian approaches are not entirely distinguishable, conceptually, though I’d have a hard time explaining it to somebody.
I agree with your point on the context-specific validity of different p values. What I have in mind is something more general, and independent of the subject matter and risk analysis considerations. It’s just based on the simple idea that in many/most traditional hypothesis testing situations you’re making a dichotomous decision (accept the null hypothesis or don’t). If that’s the case then it’s not unreasonable to use a simple majority (i.e. p = 0.50) to determine which of the two has the greater weight of evidence behind it. Granted, you pay for this in an elevated Type I error rate, but I think there are philosophical reasons for being willing to accept that, especially in complex systems having many variables and with a poor understanding of the cause and effect relationships among them…but that’s for another place I suppose.
Does anyone know why WeatherBell anomaly maps differs so much from other temperature series? I found other source (GFS-CFSR):
http://www.karstenhaustein.com/climate.php
and this is much closer estimate to GISS/NCDC/HadCRUT. WeatherBell maps show much lower values.
p (Comment #121519) First guess? Different base periods. Pretty innocent a reason, really. When dealing with anomalies, remember to ask “relative to what?”
@Andrew
I rebaselined all series to 1981-2010 which match WeatherBell base period.
p (Comment #121521)-Hm, can you link to the weatherbell maps? Could be different measurement sources.
Another possibility is you are looking at their hourly real time anomalies compared to monthly means of climate data. Not very comparable, and I wouldn’t expect good agreement.
Not sure if that’s what it is, though.
http://models.weatherbell.com/temperature.php
Of course, sources are different – they using CFS-CFSR differences. However differences between their calculations and any other are really big.
Okay, I’m looking at September, making a comparison to GISS-gotta turn down their smoothing to 250 km, infilled data will of course disagree with the reanalysis model estimation. In the area of overlap with actual observations, the map looks good.
In Sep 2013 WeatherBell calculated +0.17 in respect to 1981-2010 base period.
http://models.weatherbell.com/climate/ncep_cfsr_t2m_anom_092013.png
If I didn’t made any mistake, GISS = 0.37, HadCRUT4 = 0.27, NCDC = 0.25, UAH = 0.37, RSS = 0.15 (1981-2010).
Last two months (aug & sep) are almost equal to RSS, however before august 2013 differences was higher (0.18 in june).
Lucia: I put two guesses in. The first for zero quatloos….
A good recipe to go with turkey:
Mashed and baked sweet potatoes with walnut crust.
Boil a large sweet potatoe (enough for two people);
mash as you would potataoes (for something different, use orange juice instead of milk when mashing);
Spoon into a casserole dish;
use walnut bits (not crushed) and mix with brown sugar;
sprinkle the mixed brown sugar and nuts on the sweet potatoe; bake for 5-10 minutes.
the thermal conductivity of this is quite low. It stays hot longer than mashed spuds, so you are warned. I burn my mouth every thanksgiving….
p (Comment #121528) I was under the impression you were looking at maps, not the averages. The difference with surface data is from missing regions in the surface products. In areas where model estimation is constrained by observations, the anomaly maps look comparable.
As for UAH and RSS, they aren’t 2 meter temps, so comparison is somewhat (actually very) misleading.
Wild for me is more like the racoon clan that patrols our neighborhood. You only see them by surprise. Our power was out over night in mid October, and I found out about the momma racoon and her three pups when I turned on the flashlight and found momma supervising her young ones washing and playing in our rain water buckets. When they noticed me noticing them, they slowly and deliberately withdrew fromthe back porch, with momma acting as over all guard. the pups were cute but momma made it clear in her body language that I was fortunate to have a nice glass door between me and her brood. The ‘possums are less subtle, and do not get so ominous when caught out. Both will end up, fairly often, in people’s homes thanks to cat doors, or open garage doors. that can turn into an expensive nightmare.
Coyotes are repopulating Houston, following our bayous, railroad and powerline easements into the heart of the city. Feral cats, a yummy favorite of coyotes, are in decline.
@Andrew
Missing regions are generally always the same, so it is weird that WB series is always cooler than others.
p (Comment #121528)-I was a bit interested in weather the NCEP CFSR data trended differently from surface temp datasets, but KNMI only goes to 2009.
At any rate, over that period, the reanalysis has a trend of 0.0185661850475979 per year, for comparison, GISS 1200 km over the same period is 0.017311216709603 per year. If you are under the impression that the reanalysis has a long term cooling bias relative to the surface datasets, you are sadly mistaken.
p (Comment #121534) -As you can see, it is not “always” cooler. That’s just your confirmation bias.
@Andrew
I have no idea if CFSR shows cooling bias or not. I only wonder, why WB calculations shows lower values in any month from jan 2012 than all surface datasets.
If CFSR really shows stronger trend than GISTEMP, it’s even more weird.
p (Comment #121537)-It’s simple, not difficult to understand at all. In 2012 the areas without data in surface datasets where generally below average in the reanalysis. If you looked at some other years they would generally be above average in some, generally nuetral in others, or all over the map. I really don’t think that should be that hard to grasp.
@Andrew
So in 21 months in a row all that regions was cooler than other regions? I really don’t think so.
http://models.weatherbell.com/climate/ncep_cfsr_t2m_anom_2012.png
p (Comment #121540)-You may find it improbable, and yet it happens. And it happens in the other direction, too. Go to KNMI, download the reanalysis data, see what the global anomalies do: sometimes they are way higher, sometimes way lower than GISS and others on the same baseline.
It’s easy to insinuate malfeasance by people you don’t like, but it’s not that much harder to show you’re just wrong.
@Andrew
“Go to KNMI”
Veni, Vide and still don’t know why WB goes such low, especially there is no such effect in ERA Interim and NCEP/NCAR.
http://postimg.org/image/ttt5pqpk3/
Re: Andrew_FL (Nov 25 11:15),
That assumes that you think reanalysis data are useful. I should put scare quotes around data, as they are actually results, not data. I’m not convinced that they are very useful.
p (Comment #121544)-That’s because you aren’t look at the reanalysis being used by Weatherbell, NCEP CFSR. I said this above, I have no clue why you would look at ERA Interim, except a deliberate attempt to confuse yourself and others.
DeWitt Payne (Comment #121545)-We are discussing someone else’s use of a reanalysis product. p is alleging a conspiracy by Weatherbell to hide the hot hot hot earth by faking data. Except it’s not “Weatherbell’s calculation,” it’s a reanalysis product they are using. And it is hardly a conspiracy.
Now as for whether the reanalysis is useful, that is a separate question for whether those horrible horrible people-Ryan Maue, Joe Bastardi, and Joe D’Aleo-are conspiring to lie to people that the actual anomalies are lower than they are. They aren’t. Generally, over the long term, NCEP CSFR will *overstate* anomalies. The fact that for several months at a time it occasionally does the opposite is hardly this grand conspiracy.
Jim Bouldin, I’m intrigued by the idea of using a more lenient p value when dealing with a complex system. Instinctively, I’d think to do the opposite. The more complex a system, the more opportunities there are for errors (which won’t inherently cancel out).
I know I’d use a lenient p value if “no opinion” wasn’t an option, such as having to pick between paths to take (when staying still/going back wasn’t possible, perhaps because of a fire), but otherwise, I’m not sure why we’d want to do it.
By the way, this is an off-topic post, so feel free to discuss whatever here. Don’t feel burdened though.
Brandon, I assumed I was supposed to be discussing turkeys here. But then again, maybe I’m doing exactly that 🙂
The topic of whether to favor Type I or Type II errors, whether based on general and theoretical considerations, or on context-specific ones, is pretty darn interesting and important I think (at least for me, might well be boring as hell for others). One of the comments I read in a critique of the Johnson paper I linked to is simply that as you drop the Type I error rate you raise the Type II error rate. No news there of course, but in a complex system in which your dependent variable of interest is consistently influenced by a number of independent vars., each of which has a relatively small impact, you really do run the risk of missing these small but consistent effects if your p value is set too low. You will consistently come to the conclusion that the signal from any particular, single driver cannot confidently be separated from the noise, when this is in fact incorrect–there is a real signal but the R squared value is relatively small.
Not that I’m saying anything that hasn’t been said before, but I think, based on my admittedly extremely cursory reading, that the Johnson study doesn’t address this point.
“You will consistently come to the conclusion that the signal from any particular, single driver cannot confidently be separated from the noise, when this is in fact incorrect–there is a real signal but the R squared value is relatively small.”
Jim, would you be thinking in terms of the noisy temperature signal in TRW/MXD?
“If that’s the case then it’s not unreasonable to use a simple majority (i.e. p = 0.50) to determine which of the two has the greater weight of evidence behind it. Granted, you pay for this in an elevated Type I error rate, but I think there are philosophical reasons for being willing to accept that, especially in complex systems having many variables and with a poor understanding of the cause and effect relationships among them…but that’s for another place I suppose.”
Are we thinking TRW/MXD here, Jim?
Jim Bouldin, it doesn’t sound like what you’re looking for is really captured by the test you’re looking at. It’s true a more lenient p value can lead to picking up signals that would otherwise be lost, but unless the influence of each tested influence is the same, there’s no way to be sure errors introduced won’t bias your results in a way identical to a faint signal.
I think you need a more nuanced approach. There are a whole class of methods designed for handling (testing of) complex correlation. I’d use one of them unless there’s a specific reason for r squared values. I can’t see why there would be. Maybe if I knew more of what you have in mind, I would.
One thing I will say is if you’re willing to use a p value of .5, you could probably make it more lenient too. There is no inherent reason something has to be more probable than not to be accepted. It’d just depend on the cost/benefit of Type I/II errors.
Kenneth,
Yes and no. I didn’t have any particular process or variable in mind with my comments. I was arguing from a more general perspective. Having said that, I would answer yes and no to your question. Yes, you can definitely find situations in which tree ring response variables (TRW, MXD, mean density, etc) are influenced by several variables, including non-climatic ones. This is especially the case in closed canopy stands, where changing above- and below-ground competition–which can be slow–has to be considered. But even aside from that, we have the very thorny problem of separating out biogeochemical effects from climatic ones, most importantly the CO2 fert. effect, but also, in more spatially specific cases, N fert., and tropospheric ozone effects. But “no” in the sense that one can, in some locations, reduce the effects of at least some (but not all) potentially important variables, e.g. by sampling open, sub-alpine stands that are not typically limited by precipitation–such as at the boral forest N range limits in the sub-arctic–or at least are limited by that factor much less often, or less severly, than they are by temperature.
Brandon,
When you use the term “error” are you referring to sampling error, or to systematic bias(es) in the estimate(s) of the effect size of the various possible drivers of the response variable of interest? Some combination of these? Something else?
And wrt more complex methods for complex correlations, do you have in mind methods like structural equation modeling or path analysis? Or something different altogether?
I do think there is an inherent reason for arguing p = 0.50 is useful, because that’s the p value at which the evidence for, and against, a particular null hypothesis, are equal, i.e. the point at which the odds ratio = 1.0. Since we’re making a dichotomous decision regarding the likelihood of the null hypothesis, this is useful.
“One thing I will say is if you’re willing to use a p value of .5, you could probably make it more lenient too.”
Since the p value relates the null hypothesis to a probability of obtaining that result by pure chance, I cannot see how one would ever consider using the value of 0.5 for hypothesis testing or a value close to it. You can do random generations of a known distribution and get a result a lot smaller than 0.5 and with high frequency.
Like you say Brandon, other types of testing should be applied as an alternative to accepting large p values.
There is always the larger number of degrees of freedom that can be applied in some of these cases, but I would still want to refer to the value of R^2 since one can get a low p value and R^2 indicating that the independent variable only explains a small amount of the dependent variable’s variability – like in TRW and MXD and other temperature proxies.
Kenneth, suppose you conducted 20 independent tests of a relationship between two variables and each time got a p value somewhere between, say, 0.35 and 0.45, mean of say, 0.40. Would you conclude that there is a relationship between those two variables?
Jim Bouldin (Comment #121608)
November 27th, 2013 at 11:20 am
If you are doing it the frequentist way I would suppose you would run a larger test and then calculate a p value based on a larger number of degrees of freedom and also look at R^2 (coefficient of determination) values to determine the ratio of the explained variance to the total variance of the dependent variable to insure that given a significant probability that the result did not occur by chance the relationship between variables can be applied in a practical manner.
Doing it the Bayesian way implies multiplication of probabilities, but the 1 − p-value is not the probability of the alternative hypothesis being true. Bayesians use likelihoods in their calculations and worry about the assumptions made a prior about distributions, strive for non informative priors and end with a resulting posterior probability distribution – as I recall. Given my lack of experience in doing Bayesian calculations, I would be presumptuous to attempt here to demonstrate a Bayesian approach, but I somehow think in your push for using of a p value of 0.50 that you are improperly mixing frequentist and Bayesian concepts.
Jim Bouldin, I use “error” to refer to any divergence from the actual value.
Aye. Those, PCA and the like.
I’d agree to setting the minimum p to .5 if we are only comparing “evidence for” to “evidence against.” I don’t see when that would happen though. We should always have the “uncertain” category to consider.
Kennet Fritsch, I don’t see what you say as an inherent obstacle. Suppose, for instance, the hypothesis you wanted to check was whether or not a computer was infected with a rootkit (the most serious type of malware). If you believe it is, you’ll reformat the harddrive at the cost of a couple hours effort.
Would you require a p value no greater than .5? Of course not. If you thought there was a 10% chance of a rootkit, you’d reformat the drive.
Since this blog is so interested in playing games and betting, as that apparently is what attracts the readership, here is a fun game for all you to play.
The following link is to a chart of global average temperature since 1880. Try to guess which of the two curves is the model and which is the data :
http://img845.imageshack.us/img845/5758/6do.gif
Everyone is the winner in this game because science advances relentlessly.
WHT,
An excellent example of curve fitting as means of confirmation bias, of which you appear to have an almost unlimited amount. Curve fitting is not science, it is rubbish.
Not even sure what WHT is trying to prove. Are we supposed to be impressed? Give mea maxima culpas? I don’t get it.
WHT,
You may be right.
http://contextearth.com/2013/10/26/csalt-model/
“We then have enough information to try to solve the first equation as a variational problem, where each of the terms needs to be scaled to get the best fit.
or rearranging to make it more convenient to a solver
That turns into multiple linear regression problem as we use data from each month from 1880-2013 to give a best fit with respect to the coefficients ci.”
You could replace CO2 with something like world population or US government spending and het about as good a fit.
WHT my guess is that the CSALT model series is the blue trace in your graph with observations beginning to diverge nearing the end point.
WHT you refer to your CSALT model as being a “pause buster” in your (very frequent) posts at JC’s blog. Forgive me I have not had time to study your blog posts. In order to confidently explain the pause you would need to fit up to 1997 or so and then run your model projections against observations correct?
SteveF (Comment #121641) November 28th, 2013 at 9:28 am
“An excellent example of curve fitting as means of confirmation bias, of which you appear to have an almost unlimited amount. Curve fitting is not science, it is rubbish.”
CSALT seems to me to be a multiple regression, very similar to
yours. It just uses two extra variables, LOD and aerosols.
Nick Stokes,
Any resemblance is superficial. LOD? Aerosols? Please, aerosols are too uncertain to base anything.
Nick,
There are lots of ways to arm wave away ‘the pause’. But here is the key issue: nobody predicted ‘the pause’. Ex post-facto explanations are quite a lot like a good Australian wine that has been diluted 1:5 with water… weak and unconvincing. William of Ockham was right. The simple answer is the most likely: the damned models are most likely too sensitive.
A happy Thanksgiving to all my American friends. I trust the turkey was hot, the beer was cold, and the football great.
I guess this means WHT is going to win all the bets from now on. Why don’t we just dispense with the formalities and roll all the quatloos over to his house!
Les–
Thanks! The turkey was smokin’. I’m a wine gal. Today we are doing Turkey Alfredo for the family.
Here is wishing everyone a great day after thanksgiving.
The turkey experiment turned out great. The trick to curried turkey turns out be heating a nice mix of butter, grapeseed oil and curry powder, spooning this carefully into the space between skin and meat, after filling the cavity with peeled and sliced oranges, allspice, onion chunks and sealling off with small red potatoes.
INitially heating at 350 o uncovered, then covering with aluminum foil when skin starts to brown. Occasional basting with the same oil/butter/curry mixture. The meat was really moist, and the curry flaovor was (thankfully) not too strong.
Add to that mashed cauliflower, an Italian sausage stuffing, fresh salad green bean casserole and of course lots of Beaujolais Nouveau, home made pies and Bluebell Ice Cream and good time was had by all.
A belated happy Thanksgiving wish to all the denizens here. May we all delve a little deeper into the climate maize in the coming year.
The blue curve is the CSALT model, so the educated guess by Layman Lurker is correct.
The model diverges in the last few years for reasons perhaps related to the Cowtan & Way correction factor. I have no doubt that this divergence will eventually close up, as all the contributing factors other than CO2 have a noticeable reversion to the mean property.
I wouldn’t joke too much about the LOD factor, which is what Judith Curry has staked her reputation on, ala her “Stadium Waves” paper. Remove the LOD and the modeled pause from 1940 to 1980 disappears. The current pause seems inconsequential in comparison.
WHT-“My faith is too strong, burn the heretic.”
Andrew_FL,
WHT will have to hoard some of the oil he thinks we are running out of before he can burn any heretics.
“May we all delve a little deeper into the climate maize in the coming year”
That’s a real turkey of a comment Harold. You are hereby relegated to a maize maze which has no exit.
hunter (Comment #121664)-Okay, that was actually pretty funny!
The CSALT model as I see it uses 5 tunable (to observed monthly temperature data) parameters to describe the time series. That is not a serious modeling attempt in my view.
Jim Bouldin (#121665)
Please, no Sartre! Okay, that was too corny. Let me change that to “may you always have a sweet potato beside you”.
Harold,
OK no more Sartre, I promise. Instead I give you Descartes’ version of existentialism:
“I think, therefore I yam”
@ Jim Bouldin (comment#121672)
After our huge feast last night, one of our guests noticed we had neglected serving sweet potatoes. They told us, as they left, that it was a great meal, except for the silence of the yams….
Because of the chaotic nature of climate as shown with the variations of multiple runs of serious climate models and over periods as long as 50 years when varying only the initial conditions, a model tuned to a past temperature series is going to fail to predict the future for at least the duration of the weather noise and probably longer as the true test of an in-sample derived model is out-of-sample results.
Kenneth Fritsch (Comment #121668)-Yup, just enough to wiggle the elephant’s trunk.
John Von Neumann was really great, gotta say.
I have heard that over indulgence at Thanksgiving can cause sleepiness, but that it would induce a Cornycopia of Thanksgiving comments is something new. Perhaps it is related to climate change.
hunter,
Yes, I can imagine their disappointment.
Moreover, botanists, horticulturists and pharmacologists have long been complicit in the science of the yams I’m afraid.
Yea, it is pretty cool ain’t it?
The CSALT model is an example of everyone being right to a partial degree — Curry is right with the Stadium wave, Scafetta is right with his orbital parameters, Crowley is right with the Aerosols, Bob Carter is right with his SOI contribution, sunspots are partially right, and of course whoever figured out the importance of CO2 is right.
Like a blender, we apply the variational principle to thermodynamics and see what pops out. Observational evidence such as the long-term trend, pauses, and natural fluctuations are all characterized.
Or you know. Not at all. For a number of reasons, actually.
Very cool. We have: (1) a climate model that is personalized whereby we can provide tunable parameters with names we all recognize from climate science, (2) a climate model that allows us to say that everyone is right (to a degree), (3) a climate model that blends and pops out a result like magic, if not by magic, and finally (4) a model moderator who describes the model as though from lyrics from a Rock and Roll song. Can it get any better than this? Totally awesome.
and really awesome.
Jim Bouldin & hunter,
You have squashed my poultry efforts. [And by the way, I’m going to recycle those. Much better word than stealing.]
Kenneth,
” Can it get any better than this?”
.
Did you mean to write ‘worse’ in place of ‘better”?
.
“Totally awesome.”
.
Did you man to write ‘awful’?
“With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.” John von Neuman
The complication is that many researchers think that their own contribution is the main factor in the global AGW signal.
Scafetta thinks that his complex Moon-Sun-Earth-Jupiter orbital tug is the main contribution.
Curry thinks that her Stadium Wave theory is a contributor.
Bob Carter was one of the first to note that the ENSO SOI value followed the temperature excursions.
The “It’s the Sun, stupid” people think that sunspots are the key.
Many think that non-specific red noise contributes to the fluctuations.
No doubt that volcanic aerosols contribute significant transient effects given a large enough eruption.
And of course, there is the obvious log(CO2) sensitivity that looks like it is contributing to the bulk of the secular trend.
Has anyone put all these factors together to see what shakes out?
This is essentially factoring out all the nuisance variables and trying to figure out the main contributing factor. Whatever is left over is a residual that one can analyze for red or white noise.
I am trying to be very objective of what parameters I add to the mix. The volcanic aerosols are obviously subjective but the large eruptions (Volcanic Explosivity Index > 4) are very rare and that is one criteria that I employed, but for a couple of events. Another is the lag on the Stadium Wave, as it is not clear which modal phase controls the multi-decadal peaks. The other factor is a correction that effects temperature measurements during the WWII years.
I also include the Hybrid C&W to see how this effects the recent pause.
Virtual everything you just said is wrong. Earliest paper I can find in ENSO is (IIRC) by John Christy, who wanted to isolate the volcanic signal better. Then there is Michaels and Knappenberger in 2000. ENSO effects have been seen in the data well before Bob Carter.
The “war data correction” is a naked fudge. Picked up from Lean and Rind, no doubt, whose similar nonsense model failed during the world wars leading them to postulate cold crap data in the first, warm crap data in the second.
It would be funny if you weren’t so smug and self assured.
SteveF, ““With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.†John von Neuman”
I think he has a new 4 parameter version in the works, Carbon, Radiation, Aerosols and Pirates. I am waiting on the commodities app.
Thanks to my smugness and self-assuredness, I am getting close:
http://imageshack.us/a/img577/8173/6pu.gif
I will look into some of the work of Christy, Michaels, and Knappenberger to see what I can use. Thanks.
I agree that the WWII correction looks like a fudge but the U-boats were nastier than any fudge.
This is what the factorization looks like near the latest pause
http://img801.imageshack.us/img801/3483/bhb.gif
Well. I’ve literally had more productive conversations with unreconstructed Marxists. Anyone else?
Andrew FL, I tend to agree. How about a subject shift 🙂
http://redneckphysics.blogspot.com/2013/12/2000-years-of-climate.html
dallas (Comment #121699)-Hm, interesting. I’ll admit I am not too familiar with the reconstruction from the IPWP. At any rate, one could perhaps use these various datasets to estimate the sort of variation around the smoothed data we would expect. I think I’ll look into an analysis of the data myself. Could be interesting!
I think it is time for me to tune out here as I was convinced that WHT was doing a parody on over-fitting models and the more he comments the more I think that is what he has to be doing or at least appears to be doing in my view.
I sense you are right Kenneth. I believe we have an over-cooked turkey here.
I have come up with a neat demonstration of how stupid his fit is. Maybe it is worthy of a guest post?
Basically, I can get a good fit, too, using obvious junk variables.
It would be nice if some one explained to him the difference between a fit and a model.
Strictly speaking it *is* a model. A *statistical* model.
It’s about as meaningful as a GCM fit, IMAO.
No, LOD and air pressure ratios are phenomenological system interrogators and so cannot be forecast; fit.
DocMartyn (Comment #121709)-You seem to have something of a semantic confusion that stems from not understanding what a regression model is.
It’s not really that big a deal though.
Hehehe, I have this great demonstration though, seriously.
Typically a model is a turn-key emulator, give the model the ingredients and it will make the cake.
A fit just lists the ingredients and tells you how long it was in the oven.
Nope, you don’t understand what a regression model is.
Look, a regression model is a model, and it is the same thing as a fit. It is not whatever you are describing, which is not a definition of *all* models, it’s a definition that excludes certain statistical models.
Andrew FL, a regression model is a model but like all models it is pretty much useless unless you know why the model fits anything. One of Doc’s pet peeves is the use of (Tmax +Tmin)/2 when they are measures of different things. That is one of the reasons I used BEST and CET scaled to “fit” the IPWP proxy. Scaled, the fit is pretty good right?
http://redneckphysics.blogspot.com/2013/12/climate-sensitivity-for-instrumental.html
They have to be scaled because they include Tmin. I don’t have a problem with using all the data available provided people understand what they are using. Since Tmin needs to be scale by 0.5 to fit SST, it seems that would have been a topic in the climate discussion somewhere along all the attribution estimates that along with the long term persistence.
I understand you to mean that if you have a variety of possible fitting algorithms to chose from you call each of them a ‘model’, and state that your have chosen a particular ‘model’. Rather like when one buys a car, you have a number of ‘models’ to chose from and you purchase as particular ‘model’.
People who look at kinetics always try to make the distinction between a fit and a model; a fit does not necessarily inform you about the underlying process, where as a model generally does. some fits are useful for their informationlessness, like polynomials, and their ability to be transformed by integration, like polynomials.
Model ‘models’ are simulations/emulations of processes that have intrinsic mechanistic units., which typically are tied to actual ‘nuts and bolts’ of whatever you are studying.
Dallas, I know the model is useless. I’m being pedantic.
Doc Martyn, I don’t think we disagree, except on semantics. If you wish to say your definition of a model is narrower, that’s fine. Either way, calling a fit a model does not magically confer it the properties of a physical model.
I’ll be quite busy for the rest of the day, but I have a “model/fit” I’d like to share with everyone here if that’s copacetic with everyone. The trick is that the parameters are proprietary until such time as I feel everyone will be sufficiently impressed with it so I can spill the beans to maximum effect. 😉
Read this and links below for multiple updated versions of “predicting” S&P 500 returns based on the number of 9-year olds in the U.S. as the author keeps updating each failed model with a new yet-to-be-failed one. Remarkable looking fits are obtained though each time. I just checked and the author continues to have further updates on blogs etc. with revised models I’m sure, which continue to fail. Some people never learn.
RB,
I can’t tell if that article is meant to be a spoof?
When David Leinweber (who showed a “link” with Bangladesh butter production) emailed them, he got flamed apparently. Other “interesting” studies there that might land me in the spam folder.
Roy has posted…. 0.190C, RickA wins.
I OTOH, came no closer than usual. The group has apparently changed from diehard ‘coolers’, always under predicting warming, to flaming warmers… since the group average was 0.226C this month, well over the actual value.
SteveF,
You are almost right. It’s actually +0.193. Roy generally only shows the anomaly in his headline to two digits. The correct third digit can be seen further down in his list of monthly anomalies. Same winner tho’.
I was happy I missed to the “warm” side. But instead I was actually a little “cool”. Darn.
I have a dummy red noise data set that never gets close to as good a fit as the actual data. Sure, it can capture detail that is coincidental but that is part of the challenge. And I agree the values for correlation are inflated when an underlying trend exists, but one can still discriminate over the details in the yearly fluctuations.
Blame Scafetta for picking the periods. I tested each one of the time periods, and changing the cycle time by +/- 10% on each value substantiated the fact that they form a local optimum fit. He and others did the heavy lifting and I simply applied the variational approach that someone else was eventually going to get around doing:
http://contextearth.com/2013/12/02/orbital-forcings-in-the-csalt-model-explain-the-pause/
Dammit, I bet 0.21. So close.
Damn! All my fault!
After the comment about the cool bias, Lucia worked it out and said this:
So I
guestimatedcalculated 0.19, and added the 0.03 to get the 0.22 value I voted for. And was too high by exactly 0.03!I guesstimated 0.144 and then added a Roy factor of 0.07 but I should have only added 0.05
Nope, my model isn’t red noise based, but the fit is good. Some of the variables are the same because yeah, obviously you can match wiggles if you include ENSO. But that’s part of the point, that is literally the only thing that makes it look so good visually.
Didn’t even get the sign of the change correct. Dang.
I am not using ENSO. I am using SOI, which is a measure of a pressure differential, not a temperature. This is what some refer to as an energy variational approach.
Consider the expansion of the ideal gas law, nRT = PV
dT = 1/nR( PdV + VdP)
SOI is the dP term which relates to a differential change in temperature dT.
In terms of free energy we can add more terms such as Scafetta’s gravitational forces and then minimize the change in energy in the full expression. Just so happens that a statistical multiple regression solver functions well as a variational principle solver.
Fine to take potshots at the approach but this stuff is what any student can glean from textbooks on thermodynamics and it is something that any curious person would try — i.e. how well does the earth’s atmosphere perform as a classical free energy thermodynamic entity? I am not a climate scientist, but I am a student of statistical mechanics and thermodynamics, and energy conservation is still the key.
WebHubTelescope (Comment #121728)-No, now you are spouting gibberish. First of all, SOI is ENSO. That you aren’t a climate scientist, or even the least bit well read on climate is actually self evident, as otherwise you would know that the pressure differential between Tahiti and Darwin is, in fact, strongly negatively correlated with the temperature anomaly in the NINO3.4 region:that’s the whole god damn point, they are both measures of the same phenomenon-The El Niño Southern Oscillation. Second, ENSO’s effects have nothing whatsoever to do with the Ideal Gas Law. There is nothing about your model that involves energy conservation at all.
I wanted to make sure you realized that I was not using the temperature index but the pressure index. Continue to flame away.
It wouldn’t make a difference. I could also substitute NINO3.4 with SOI and get about the result I’ve already gotten for mine. And it would still be meaningless.
Ciao for now, everyone, gotta be going. I’ll be busy today, but have some fun stuff to share later.
We are characterizing fluctuations here. Fluctuations in temperature at 300K of 0.06K correspond to fluctuations in pressure of 0.2 mbar at constant volume. This becomes a transient energy reservoir according to the fluctuation-dissipation theorem. So we add it to the global free energy mix.
Re: WebHubTelescope (Dec 4 13:24),
Since when did the atmosphere have constant volume? Sure, if you increase the temperature of the whole atmosphere there’s some PdV work done because the volume increases, the pressure doesn’t. But, as I remember, it’s small compared to the increase in enthalpy from the change in temperature. ENSO/SO doesn’t increase the temperature of the whole atmosphere either.
Over large areas, hydrostatic equilibrium applies, so absent a temperature change, a change pressure somewhere must be matched by a change in the opposite direction somewhere else. The total mass of the atmosphere doesn’t change.
Actually, I think there is some evidence for *decreasing* pressure at sea level. But I think that could easily be measurement error.
Andrew FL Yeah, but not much. What Webster seems to miss is that surface pressure is a dynamic measurement whether he likes it or not. Changes in average surface wind velocity change the average surface pressure. Since winds have both a temperature and mechanical, Coriolis effect dependence, shifts in the ITCZ change surface pressure. With all the interdependency you can get caught in a what causes what circle. Since he mentioned 0.2mb, that would be a 12mph wind speed if nothing else changed and the pressure difference increases with the square of velocity. The “average” surface pressure is higher at the poles so the polar warming associated with warm NH ie AMO, would tend to reduce pressure a touch both with temperature and average wind velocity which are joined at the hip so to speak.
“DeWitt Payne (Comment #121733)
The total mass of the atmosphere doesn’t change.”
Are you sure about that? I am pretty sure that the amount of water vapour in the Earths atmosphere will be proportional to the amount of solar flux over water. During Southern Hemisphere summer the amount of water in the airs should be greater than in the winter.
DocMartyn (Comment #121739)-Hm, interesting question.
I was going to say at some point that the Earth system is not *perfectly* closed, I think there is a net loss of mass over time, but it is *extremely* tiny, at least compared to the amount there is to begin with (the atmosphere loses, apparently, some 95 thousand metric tons of hydrogen every year to space. I think the net of all mass flow is 50 thousand. For reference, that net is ~8.37*10^-16% of the whole mass of the Earth. Of course, it’s a larger percentage of the atmosphere. That hydrogen mass loss is ~1.85*10^-9% of the atmosphere.
Oh, found it! Mass variation due to water vapor over the course of a year is about ~2.91*10^-2%:
http://en.wikipedia.org/wiki/Atmosphere_of_Earth#Density_and_mass
Which, I guess if the atmosphere got *much* more humid, could be significant.
OK, want to work out the loss rate of mass, due to radioactive decay? Don’t forget we lose He too.
DocMartyn (Comment #121741)-That’s included in the net 50 thousand metric ton figure. But it’s only 16 metric tons-not thousands, just units.
Oh, and loss of He is 1.6 thousand metric tons.
The Earth system is not *perfectly* closed, I think there is a net loss of mass over time, but it is *extremely* tiny, at least compared to the amount there is to begin with (the atmosphere loses, apparently, some 95 thousand metric tons of hydrogen every year to space”
According to some other theories there is an expanding earth as it picks up space dust, debris, meteors [one big meteor would square up a lot of that mass loss never mind the 10,000 smaller bits that hit every day] and a bit of the spare hydrogen and helium that the sun loses.
So is it a net loss of mass or a net gain of mass guys?
Hey guys this is meteorology 101. Surface areas with the greatest volume of air over them have the lowest pressures and relatively lower temperatures.
This is all due to the fact that warm moist air is less dense than dry air and as it rises it cools because it expands and it does not radiate the heat away or warm the surrounding air.
Just look at the IR images of the big hurricanes, it is easy to see the volume and the extremely low temperature.
Lower lapse rates create lower surface temperatures.
Angech, space debris is included in te net 50 thousand metric ton figure.