Guest Post by Willis Eschenbach
Anthony has discussed a paywalled study in the new reality-based Nature Magazine production, Nature Climate Change magazine. Unlike Anthony, they approved my application for a free subscription … go figure. The study is called “Nonlinear heat effects on African maize (corn) as evidenced by historical yield trials”, Lobell et al. (hereinafter L2011). The study looked at the effect of heat on corn production. Here’s their Figure 1:
Figure 1. The opening figure in the L2011 study of maize production in southern Africa. I always enjoy rich visual presentation of data, note that this contains elevation information as well.
Their conclusion? When it gets above a certain temperature, maize growth quickly slows, and it’s worse when it’s dry. Of course with the obligatory links to global warming and the danger of large drops in corn production. Shocking news, I know. They provided a citation to other scientists saying the same thing, in case you doubted it — too much heat is bad for plants. I bet the farmers of the world were as amazed as I was.
Or as they put it in their abstract:
Each degree day spent above 30° C [86°F] reduced the final yield by 1% under optimal rain-fed conditions, and by 1.7% under drought conditions. These results are consistent with studies of temperate maize germplasm in other regions, and indicate the key role of moisture in maize’s ability to cope with heat.
Now, we need to be careful here. They are not talking about the number of days where the temperature goes above 30°C. They are discussing “degree days”. That is the sum of the average daily temperature (C) less 30 degrees, for all the days where the average temperature [defined as (daily max + min)/2] is above 30°C. The figure is written as “GDD30+”, for “growing season degree days over 30°C”. They figure the growing season as 150 days, which agrees with the Texas figures given below.
Are their numbers accurate? Is there a drop in yield of 1% for every degree day as they claim? I don’t know. Haven’t done my homework yet, just dug up the paper, gimme a minute. Where do they grow corn? Iowa? Let me look it up. OK, I find:
Figure 2. Major (dark green) and minor (light green) corn growing areas in the US, by county. Texas is the large state numbered “2”. Between 60-70% of Texas corn is irrigated.
Fascinating. I love doing this, I get to learn so much. Well, at first glance I’d say the following:
1. The major corn-growing areas are from about 37°N to 47°N. So clearly, corn prefers temperate weather.
2. Corn is only a minor crop in many regions within that general preferred temperature band. So obviously, there’s other factors. The usual suspect would be water, second would be soil.
3. Corn is grown in the California Central Valley, one county in Arizona (irrigated, no doubt), a number of counties in southern Texas (mostly irrigated), and one county in Florida. I looked at the temperature record for Hidalgo County, the left one of the counties at the south tip of Texas in Figure 2. I looked at the daily temperature record for Edinburgh, in the middle of the county.
Here’s the curious thing. During the corn-growing season of 1999, the total number of “degree day[s] spent above 30° C” (GDD30+) in the Texas corn-growing area was 136 … so if yield dropped by 1% for each degree-day over 30°C, we’re down below zero to a quarter of the original yield. Hmmm. Figure 3 shows the degree day analysis, from the excellent online calculator from Wolfram Alpha here:
Figure 3. Degree days over 30°C for 150-day 1999 corn-growing season, Edinburgh, Texas.
I got to thinking about what was happening. How could they be growing corn in that kind of heat, with a GDD30+ over a hundred and thirty? I thought about it a while, and looked around on the web a bit. Figure 4 shows part of the answer:
Figure 4. Corn planting and harvesting dates in Texas. The “Panhandle” is the most northerly square section of the state (see Figure 2). SOURCE.
I’m sure you see the pattern. In the south, like Hidalgo County above, they plant and harvest early. Their crop is three-quarters harvested before the rest of the state has even begun.
As for the other part of the answer, I don’t know. I don’t know why even with their early growing season (March 1 – August 1) the Texas farmers are still able to grow corn in that heat. The L2011 study says that’s impossible, but perhaps the Texas guys and gals didn’t get the memo, they’re a cactus-tough bunch down there, hard to get hold of. Thinking on it, though, it’s more likely they got the memo and shot it full of holes for target practice. In any case, during their growing season, the Texas farmers have no less than a hundred and thirty-six degree days over 30°C, which according to the L2011 results should reduce yield by 136% 75% … which means that either I or Wolfram or the climate scientists did something wrong. I’m open to any suggestions, I’ve been wrong before.
Now, if there were to be a general warming, say a degree on average over some long time, what do you think will happen to the planting and harvesting dates in Figure 4? Do you think those farmers would keep planting at the same time of year, year after year, in the face of increasing hot days summer and decreasing yield? Do we really face a 1% drop in yield for every degree day over 30°C?
Naw … in answer to the question in the title of this post, farmers are smarter than the L2011 climate scientists. If temperatures change, the farmers change their planting times … what do you do?
My best to everyone.
w.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Willis Eschenbach says:
“Thats what humans do, we’re not slaves to the weather such that we’ll sit and watch yields decline and not take action.”
Ironic. I’ve been pointing out that taking these actions (and increasing yields) are the main things that Norman Borlaug and CIMMYT have been doing for decades.
Here’s a link to an earlier (2007) study by Lobell on “global scale climate-crop yield relationships and the impacts of recent warming ”
http://iopscience.iop.org/1748-9326/2/1/014002/fulltext
It was something of a data trawling exercise looking at possible months of temperature data versus yield- however the best R2 combination for maize seems to be to behave rather differently to the lower R2 combinations. -[Their figure 5 ].
Physics Major says:
March 15, 2011 at 11:34 am
Why did they bother to go to Africa for this study? I would imagine that one could go the Ag departments in any of the Midwest state universities and find studies relating corn yields to every conceivable variable.
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
GOOD LORD YES!! And then some.
In reading through the post (which I much enjoyed) and the responses, I felt that something I had read damned the Lobell et al paper out of their own mouths. It was in Anthony’s previous discussion (linked to by Willis) and the three relevant paragraphs are repeated immediately below:-
Open quote.
While the crop trials have been run for many years throughout Africa, to identify promising varieties for release to farmers, nobody had previously examined the weather at the trial sites and studied the effect of weather on the yields, said Lobell, who is an assistant professor of environmental Earth system science.
“These trials were organized for completely different purposes than studying the effect of climate change on the crops,” he said. “They had a much shorter term goal, which was to get the overall best-performing strains into the hands of farmers growing maize under a broad range of conditions.”
The data recorded at the yield testing sites did not include weather information. Instead, the researchers used data gathered from weather stations all over sub-Saharan Africa. Although the stations were operated by different organizations, all data collection was organized by the World Meteorological Organization, so the methods used were consistent.
End quote.
The second quoted paragraph really says it all. It’s my experience that it is hard enough to get valid results from an experiment that you have actually designed to achieve a stated objective, let alone from one that was carried out for ‘completely different purposes’ and one in which the original observers did not bother to record any local weather information.
I also empathise with Anna. In my darker moments, I often feel that western civilisation is sowing the seeds of its own self-destruction especially where politicians are contaminating science for their own short-term gains and in feeling that the pursuit of knowledge etc. has ‘to be made easy, or at the very least, FUN’.
I could go on but that’s enough for now and it’s nearly time for bed, anyway.
Charlie
E.M.Smith says:
March 15, 2011 at 1:21 am
Jeff Alberts says:
2. Corn is only a minor crop in many regions within that general preferred temperature band. So obviously, there’s other factors.
For those who speak English, that would be “there are other factors”. 😉
Well, to quote my Texas Uncle:
“Well, son, we won’t go holten that ‘gainst yah. There’s been lotsa facturs ’roundt hearsabouts, and spiiken Anglish aint won we been known for holten ‘gainst a fella.”
Soes if yous ‘ill just stop t’ holler ‘n about it, we ken prolly get you’all ‘t pass with’n out too much trouble….
‘Corse, if you’all are gonna make a fuss ’bout it, well, let me introodoose you to some Texas “Culturall Art E Facts”… This ‘un here, it’s calt a “Three Fifty Sev’n Magnum” and this other ‘un, it’s calt a “Foorty Five Colt”…
😉
I “married in” to a Texas Family. It has been ‘interesting’…. ect
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Old Son lemme jest say that yer right as rain. Jest one little nit ta pick wit ya.
Stead of interesting….should’a been “INtrastin.” LOL
Denny
I should have added to my previous comment re the Lobell et al. paper that surely, to study the effects of ‘climate change’, you have to define somewhere what you mean by ‘climate change’
Perhaps the authors did as I have not had access yet to their paper.
Charlie
Willis — it appears that you used a different “degree day” than did the article.
They use some sort of hourly summation, assuming sinusoidal variation of temperature from min to max over the day.
Their method is definitely not the simple degree day calculation using (Tmax+Tmin)/2 – Tbase. It doesn’t seem to be the same as the Baskerville emin version of degree days, although it is closer to that.
See equation 2, on page 7 of http://iis-db.stanford.edu/pubs/23138/Lobell_2_11_NatureClimateChange.pdf
The equation seems a summation at hourly intervals over the entire 3600 hour (150 day) growing period. So the summation is really degree-hours as written.
——————————–
I note that the key supporting reference on effect of 30C+ temps on corn also appears to be climate change related: 13 Schlenker, W. & Roberts, M. J. Nonlinear temperature effects indicate severe damages to U.S.
231 crop yields under climate change. Proceedings of the National Academy of Sciences 106, 15594-15598 doi:10.1073/pnas.0906865106 (2009).
Physics Major says:
“Why did they bother to go to Africa for this study? I would imagine that one could go the Ag departments in any of the Midwest state universities and find studies relating corn yields to every conceivable variable.”
Two words: ground truth. How could an Ag department in any of the Midwest state universities duplicate tropical climates and lack of seasonality? You go to were these things occur to get the answers. See: http://www.cimmyt.org/
cirussell says:
March 15, 2011 at 11:13 am
If their formula were for growing degree days, you’d be correct. But they are calculating degree days over 30°C, which is why they call it GDD30+.
Thanks,
w.
We grow a lot of corn in Honduras, and corn is very popular in Mexico. Yields are low here, about 1/4 that of the U.S. Corn Belt. But poor rocky soil, low fertilizer and pesticide use surely contribute to a lot of the difference.
Charlie A says:
March 15, 2011 at 1:22 pm
Yes. I used the standard calculation, (daily temperature max + min)/2, because I didn’t have hourly temperatures for the Edinburg site in Texas. In their case, they fit a spline to the actual max and min and calculate it hourly … I don’t think it makes much difference, although I haven’t checked that. Hang on …
OK, I just took a look. It’s interesting, there is some difference. I made up some pseudo-data, 24 hourly pseudo-temps per day. Sinusoidal, of course. I added to it a much slower sinusoidal curve to represent the warming and cooling during summer. Then I looked at the degree-days figured two ways – the way I did it with average daily temperature, and their method.
The way they did it (calculate “degree-hours over X° C”, and divide by 24 hrs/day) gives a greater number of degree days than the usual method (using the daily mean temperature rather than hourly mean temperature). This is true for all values of X. The larger the number of degree-days, the smaller the effect. When temperatures are lower, of course, the mean temperature drops below 30°, so no degree days are recorded. But the hourly temperatures still can go over 30°, so using their method fractional degree-days are recorded. These can add up depending on the exact data in question.
So … where I measured 136 growing season degree days over 30°C in southern Texas, their method would give a larger number. How much larger? Unknown, but from my simulation likely around 25%.
Thanks for bringing that up, I’d been meaning to look into that.
w.
PS – I disagree with their method, in that it makes up data under the assumption that the daily temperature swings are sinusoidal. While in general this is not too far from true, often it is not the case at all. This is particularly true in the tropics. There, mornings are clear and the temperature rises fast. Afternoons are cloudy, the temperature rise flattens way out.
In addition, drawing a sine wave through a bunch of data can give you good numbers.
But drawing a sine wave through the extreme values is likely to give you bad numbers … and nature specializes in extreme extremes. Temperatures don’t vary smoothly. You have a day where it’s overcast all day. Then at around 2 PM, the clouds disappear and the hot sun shines down. For that hour it’s lovely and warm. Then the clouds come over again, and the party moves indoors to get out of the cold wind that springs up.
Now, if you draw a sine curve through the maximum temperature point of the day, it will look like a nice sunny warm day, warming to a peak and then slowly cooling down. But in fact, the temperature flatlined during the day, except for one large but short spike in temperature at 2 PM when the clouds broke.
As a result, putting a spline curve through the extremes of a dataset is pretty much guaranteed to distort the data. Better than nothing, I guess … but I just don’t like manufactured data.
safariman says:
March 15, 2011 at 8:04 pm
Thanks, safariman. The GAEZ study is great for looking at these questions. Take a look here at the GAEZ estimate of possible yields for maize in Central America. Honduras is inherently not a great spot for maize. The Yucatan is much better.
w.
The Romans must have starved to death while all around them empires and civilizations flourished. /sarc
Roman warm period as warm or warmer than the our current period.
http://www.jstor.org/pss/124215
http://www.pnas.org/content/early/2010/03/02/0902522107.full.pdf
Just to remind ourselves that there is nothing much new that comes around in human experience, I offer the following extract from published material.
My immediate source is ‘Planning of Experiments, D. R. Cox, John Wiley & Sons, 1958 (Fifth Printing December 1966), Library of Congress Catalog Card Number 58-13457’. I am somewhat concerned that I do not have written permission from the publisher to reproduce such material and I rather suspect that 44 years is not quite long enough for copyright to have lapsed.
The example given by David Cox is on page 10 of the referenced publication and is as follows.
“Student” (1931) mentions some experiments done by the Irish Department of Agriculture in connection with the introduction of Spratt-Archer barley. This was almost everywhere a great success; yet in one district the farmers refused to grow it, alleging that their own native race of barley was superior. After some time the Department, to demonstrate Spratt-Archer’s superiority, produced a single-line culture of the native barley and tested it against the Spratt-Archer in the district in question. “Student” reports that to the Department’s surprise the farmers were perfectly right; the native barley gave the higher yield. At the same time, the reason became clear; the barley in question grew more quickly and was able to smother the weeds, which flourished in that area; Spratt-Archer, growing less strongly to begin with, was, however, the victim of the weeds. Thus the original experiments, carried-out on well farmed land, were definitely misleading when their conclusions were applied elsewhere.
This little piece lends support to many of the points raised in this thread. It also shows a refreshing glimpse of the willingness of some bureaucrats in earlier times to attempt to demonstrate their case using a further properly organised study – rather than just shouting ever more loudly.
Some readers will no doubt recognise the name “Student” as he of ‘t-distribution’ fame. A century ago, as W. S. Gosset, he was working for the Guinness Brewery Company, but apparently in order to safeguard his employment there he felt obliged to publish his statistical writings under a pseudonym. Whether this was because of some dispute over intellectual property rights or other similar reason, I know not.
What does seem now to be likely is that many people are only able to work on projects, and to publish suitable results therefrom, as agreed by their employers. Step out of line and here’s your P45 (an HMRC tax document provided to people leaving employment in the UK). In principle, this has probably always been the case but it will be interesting to see if the work of the CAGW ‘climate scientist’ lobby stands the test of a century as has “Student’s” work.
Mod: Thanks for the note re latitude of Texas. My point was Swaziland is as far south as Testas is north. Couldn’t reply right away – been travelling from Ulaanbaatar to Waterloo. Takes time…
Tim Clark in his good piece says:
“This issue has been studied hundreds of times before, but perhaps not in Africa.”
As can be noted from seveal contributions above, there is a huge amount of agricultural research going on in Africa. For example, the local development of high lycene maize with a long shelf life in Ghana – global level contribution to the food supply.
One programme alone at the Malkerns Research Station in Swaziland 25 years ago produced 20 local MSc’s just for that task. The popular Western picture of African agriculture may locusts eating a shrivelled millet stalk in Mali, but it is very far from the truth in most places.
Dumping subsidised Western (including European) maize and milk products has been particularly damaging to farming throughout the continent. In a free market Africa could outproduce the ROW (rest of world).
Maybe they have failed to properly isolate the variabel?
grayman says:
“The scientist did not not use real farmers or agricutral science in the study, they just seemed to go to the farms take a few pictures, then go get the data from far off weather stations, and sit down and write.”
I don’t believe that is correct. For decades CIMMYT has set up a world wide network of trial farms to improve quality and quantity of grain. They have succeeded. The published article at http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate1043.html
states: “In this case, we focus on field trials for tropical maize conducted in Africa in 1999–2007 on a network of 123 research stations managed by the International Maize and Wheat Improvement Center (CIMMYT), National Agricultural Research Programs and private seed companies (Fig. 1). The original purpose of these trials was to test new varieties across a range of environmental conditions, to identify robust lines for release to farmers.”
Thus, these trial farms are in part set up by one of the author’s organization (CIMMYT).
They aren’t afraid to get their hands dirty. For more info see: http://en.wikipedia.org/wiki/Norman_Borlaug
Further to my earlier comments.
Wouldn’t you just know it?
“Student’s” 1931 paper about the differences in barley yields according to the Irish Department of Agriculture and farmers in a particular district was entitled ‘Agricultural field experiments’ and published in ‘Nature, 127,404’. It was later reprinted in “Student’s” collected papers. Cambridge, 1942.
I wonder if Lobell et al. cited “Student’s” paper in their study.
Loved this quote, “When it gets above a certain temperature, maize growth quickly slows, and it’s worse when it’s dry.”
Well, of course, quickly slowing is much worse than slowly slowing or quickly speeding up. But what about when it is slowly speeding up? Perhaps that’s as bad? Or maybe it’s better. I’m not sure if it makes me anxiously sanguine about the whole thing, or indifferently partial. Anyway, although though the observation is obviously profound, I find it to be seriously ludicrous.
R. Craigen says:
March 16, 2011 at 8:03 pm
Is this a serious complaint, or not? If so, we say things like “I was late noticing the police car, so I quickly slowed my car down to avoid a ticket. I would have slowed down faster, but I was afraid I’d squeal my tires.” The meaning of that seems quite clear … I’m afraid I don’t see the problem. If you’re serious. If not, ignore alien orders.
w.