Tuesday, April 10, 2012

Spatial Coherence of Illinois Corn Yields


I'm not really sure where I'm going with this, but it took a long time to make so I'm inflicting it on readers regardless!

The above shows, for a sample of thirty counties in Illinois, the departure of the corn yield from its eleven year moving average in that county. (Data: NASS) Obviously corn yields, in Illinois as elsewhere, have been increasing as a result of technology - but we can use the departure from the moving average to measure the degree to which the circumstances of a particular year were markedly better or worse than the average.  I picked Illinois for no better reason than being a fairly representative midwestern state that I happened to visit recently.

You can see that there is substantial covariance - a bad year tends to be bad for most counties, while vice versa in a good year.  However, there is not complete covariance - for example 1954 was a very bad year, but if we look at the cumulative fraction of all 102 counties with a particular yield fraction (relative to the moving average) we get the following (blue line):


It was pretty bad in about half the state - some counties had less than a quarter of the normal yield - but the other half was within 10% either way of the moving average.  1934 and 1983 were more consistently bad - but even in 1983 the fraction of the moving average crop yield ranged from in the 25% range in the worst hit counties to over 80% in the least hit counties.  In short, the scale of even the most consistent crop failure in recorded history is not so much larger than the state of Illinois that it pretty much hits the whole state equally.  And most crop failures left some portion of Illinois untouched.

If I get the time to do some more heavy-duty number crunching I'm interested in understanding the spatial structure of yield failures like this, and whether there's any way to detect any signature of recent changes in the weather patterns causing more extensive or more correlated yield failures.

8 comments:

Stuart Staniford said...

For future reference - there is a file of county centroid positions at http://home.comcast.net/~rickking04/gis/spcmeta.htm

Susan Kraemer said...

Can you hang on till the 2080s? That is when corn and soy crop yields throughout the farm belt are expected (by climate scientists) to drop by over 80%: when there are simply too many days a year when the temperature is over 86 F (IIRC: or 82 F)

Michael R said...

May be interesting to crosstab that with property tax records.

"On the first settling of a country, in which there is an abundance of rich and fertile land, a very small proportion of which is required to be cultivated for the support of the actual population, or indeed can be cultivated with the capital which the population can command, there will be no rent; for no one would pay for the use of land, when there was an abundant quantity not yet appropriated, and, therefore, at the disposal of whosoever might choose to cultivate it.

"On the common principles of supply and demand, no rent could be paid for such land, for the reason stated why nothing is given for the use of air and water, or for any other of the gifts of nature which exist in boundless quantity. With a given quantity of materials, and with the assistance of the pressure of the atmosphere, and the elasticity of steam, engines may perform work, and abridge human labour to a very great extent; but no charge is made for the use of these natural aids, because they are inexhaustible, and at every man's disposal. In the same manner the brewer, the distiller, the dyer, make incessant use of the air and water for the production of their commodities; but as the supply is boundless, they bear no price. If all land had the same properties, if it were unlimited in quantity, and uniform in quality, no charge could be made for its use, unless where it possessed peculiar advantages of situation. It is only, then, because land is not unlimited in quantity and uniform in quality, and because in the progress of population, land of an inferior quality, or less advantageously situated, is called into cultivation, that rent is ever paid for the use of it. When in the progress of society, land of the second degree of fertility is taken into cultivation, rent immediately commences on that of the first quality, and the amount of that rent will depend on the difference in the quality of these two portions of land.

"When land of the third quality is taken into cultivation, rent immediately commences on the second, and it is regulated as before, by the difference in their productive powers. At the same time, the rent of the first quality will rise, for that must always be above the rent of the second, by the difference between the produce which they yield with a given quantity of capital and labour. With every step in the progress of population, which shall oblige a country to have recourse to land of a worse quality, to enable it to raise its supply of food, rent, on all the more fertile land, will rise."


http://www.econlib.org/library/Ricardo/ricP1a.html#Ch.2,+On+Rent

Michael R said...

Also, I thought this was interesting:

"New England also experienced great consequences from the eruption of Tambora. The corn crop was grown significantly in New England and the eruption caused the crop to fail. It was reported that in the summer of 1816 corn ripened so badly that no more than a quarter of it was usable for food."

http://en.wikipedia.org/wiki/Year_without_a_summer#Effects

Even in a "nuclear winter" scenario, 25% of the corn crop survived.

sunbeam said...

Since you have the data (at least the temperature portion I'm sure) is there any way you could do a graph of something like yields per county on average on the vertical scale, with temperature on the horizontal axis? For different values of average rainfall?

I hope that makes sense, and I know you would have many different data points for temperature. I am hoping you have a statistical trick for dealing with it.

I think I need to define more in my mind what I'm looking for. I believe in climate change, but my mind works better when I see exactly how crop yields are affected by moisture and temperature changes.

I ask this because it seems like you have most of the data already.

Stuart Staniford said...

Susan - yeah, but 2080 is so far off in the infinite future as far as biotechnology is concerned - there's absolutely no way to know what corn plants will look like by then and therefore I don't see how we can say anything useful about what will happen to yields in any particular climate.

Up till now my view of the data has been that there is no negative climate signal discernible in yields. However, I'd wondering if there may be something detectible in the correlation structure as a result of more weather extremes - probably not, but I'm interested enough to check.

Stuart Staniford said...

Sunbeam - my sense from a little googling around this morning is that at least the big crop failure years are due to drought (at least 1934 and 1954 are that way) So probably our friend the PDSI is the most useful thing

Unknown said...

Crop yields are the sum of all the factors which impinged on the crop from the time of planting thru harvest. The main overriding factor is weather. Weather can impact planting dates, spraying dates, insect hatching, slope of growth curve, and harvest dates to name a few. Weather can be spatially variable from year to year - some years affecting a whole state and only a few counties in other years. Thus the only way to sort this all out is to get the weather data by County and start plugging away. THEN, the problem is how useful is this to a farmer in this years crop? I live in Oregon where we say "If you don't like the weather, wait 5 minutes." In other states the weather is highly predictable.

Tom Thomson