Climate Change Threatens the Corn Belt

While American farmers are still suffering from recent severe flooding, historically our “corn belt” has done a remarkable job increasing food production. The main reason is of course the advances of farming chemistry, genetics, and technology, but throughout the 20th century the U.S. corn belt went farther and faster than other regions of the world (even other regions of the USA).

New research from Partridge et al. might explain why. Climate change has been raising temperatures around the world and across the USA, but while other areas had to contend with the bulk of it, during the growing season our corn belt has heated up much less than most places, while precipitation increased slightly. In other words, we got lucky.

Because climate change isn’t good for agriculture. Partridge et al. put it this way:

The Corn Belt of the United States, one of the most agriculturally productive regions in the world, experienced a globally anomalous decrease in annual temperatures and a concurrent increase in precipitation during the mid- to late-20th century. Here, we quantify the impact of this ‘warming hole’ onmaize yields by developing alternative, nowarming hole, climate scenarios that are used to drive both statistical and process-based crop models. We show that the warming hole increased maize yields by 5%–10% per year, with lower temperatures responsible for 62% of the simulated yield increase and greater precipitation responsible for the rest. The observed cooling and wetting associated with the warming hole produced increased yields through two complementary mechanisms: slower crop development which resulted in prolonged time to maturity, and lower drought stress. Our results underscore the relative lack of climate change impacts on central US maize production to date, and the potential compounded challenge that a collapse of the warming hole and climate change would create for farmers across the Corn Belt.

What is this “warming hole” of which they speak?

They refer to the “warming hole” I’ve blogged about earlier. That was a sudden shift in temperature, a sharp cooling mid-century (right around 1958) which some said was focused on the American southeast. Partridge and others had previously extended its geographic range to include more of the corn belt, when defining it as “persistently cool stations 1961-2015”:

The range does tend further north when considering the summer season:

However, it doesn’t extend as far north as the corn belt’s area of fastest-growing agricultural production. In particular, it doesn’t cover North Dakota much, despite its showing the biggest climate impact of the “warming hole” on food production, according to the new study.

I have a hypothesis, just an idea: that there may be two different “warming hole” regions at work here. One shows the sign of the sudden temperature shift around 1958 associated with the “warmhole” discussed in my previous post. This is one way to get “persistently cool stations 1961-2015.” The other is more associated, not with a suddend shift in 1958, but with a lower warming trend from 1975 (or thereabouts) to the present. This too can can lead to “persistently cool stations 1961-2015” and hence meet Partridge’s definition of “warming hole.”

Here are the warming rates in the USA from 1975 to the present during the “growing season” which I’ve set to May through September (on a climate-division scale rather than the county scale used in Partridge et al.):

You can probably see that in a large region of the midwest, the warming has been slower than in the rest of the country. We can emphasize it by showing, not the warming rate, but its difference from the USA average. I’ll use upward-pointing red triangles for regions warming faster than average, downward-pointing blue triangles for those warming slower than average:

The first warming hole, the sudden shift around 1958, shows up in the average warming rate during the earlier period 1930 to 1975:

It seems to me that the earlier warmhole was concentrated in the southeast, manifesting as a sudden shift in the lates 1950s. I also concluded, in my earlier post, that a reduced warming rate afterward was not part of the pattern. It’s the later warmhole that shows that behavior, and seems to me a better match with the U.S. corn belt.

I’ve indicated those areas by mapping the “warming rate minus USA average” in mean temperature (the average of the day’s high and low temperatures). But similar behavior is shown by high temperatures (the metric used to correlate with food production by Partridge et al.); here’s the warming rate of high temperature from 1930 to 1975, to detect the first warmhole event:

Here’s the same result for the period 1975-present, to define the second warmhole event:

Perhaps the most important aspect of the new research isn’t how they define the warming hole, but how they quantify its impact on agriculture, and how they estimate what its effect would have been without the “warming hole” keeping things cool (and wet) in the corn belt.

They use two methods to estimate the impact of climate change on food production. One is a process model, which simulates how climate affects corn growth. The other is a machine learning method to model climate’s effect on corn production. The method they chose is the “algorithm du jour” in machine learning (and with good reason), known as “random forests.” I’m not here to critique machine learning methods, just to say that this is one of the best and that I’m glad such methods are making their way into the physical sciences.

They also use two methods to simulate what conditions would have been like without the warming hole. One is to treat it as a sudden shift, and just add a constant to post-1958 values to simulate what is expected without the warming hole. The other is to compare the distributions before and after, and shift later values to their corresponding valus by matching the quantiles of the distributions.

In all cases — using a sudden shift or quantile matching to simulate climate without the warming hole, and using process models or machine learning to estimate its impact on food production — the corn belt in the U.S. gained dramatically by not experiencing the climate changes we’ve seen recently nearly as strongly as other agricultural regions.

And they caution us: the corn belt is not immune to climate change, and will “catch up” with the rest of us before too long. That can put a serious dent in their food production. And other climate factors can be damaging too: the increase in average rainfall in the U.S. has, they say, long been of benefit to the corn belt, until this year when too much precipitation has caused such severe flooding problems for farmers.

All in all, it’s an object lesson in the real danger of climate change. We have long been pestered by climate deniers with the idiotic question “What’s the ideal climate anyway?” The answer is: stable. We have built our society, our cities and our farmlands, our infrastructure and technology, around the climate we grew up with. Changes mean our adaptations no longer match our needs.

This is not your father’s climate. It’s not your grandfather’s climate, or his grandfather’s.

Let’s do what we can to make sure our daughters’ climate is still something they can deal with.

This blog is made possible by readers like you; join others by donating at My Wee Dragon.


3 responses to “Climate Change Threatens the Corn Belt

  1. Thanks for the analysis. I worry for our children and grandchildren. My generation may be feeling the impact of global warming, but the younger generations are going to struggle with the impact. There is nothing fair about this and I don’t blame smart young folks for looking at the situation and saying essentially WTF?

    Have you done any large analysis of global warming with an eye to determining if global warming is happening faster than scientists expected as Oreskes and Stern indicate in their recent opinion piece at the NYT?

    I think the warming is clearly happening faster the consensus scientific timeline expected, but others are not convinced.

    Cheers and warm regards


    [Response: No, I haven’t addressed that question specifically, but I have looked at some aspects of it. Sea ice decline in the Arctic is one of those things that has happened faster than expected, by quite a bit. Temperature itself is following model expectations very well. As for sea level, I don’t know what the models say but I would guess that very recent acceleration exceeds expectations.

    What I hear from other scientists that worries me, but that I don’t know enough about personally, is the ecosystem impact. Dangerously close to ecosystem collapse events really worries them, and it worries me too.]

  2. This seems like a really impressive paper. Thanks for bringing it to everyone’s attention.

    I have had repeated arguments with “contrarians” who insisted that the ever-increasing production of the US maize belt is proof that AGW is good for agriculture. This paper seems like convincing evidence for the opposite conclusion.

  3. I have two questions, mostly based on looking at the final map (Growing Season high temps 1975-2018. First, does the area of blue triangles map onto the major part of the Mississippi drainage? And is this a coincidence or is there some connection? Second, do you know what’s been happening in the area north of the US border in Canada, especially in the “blue” zone?