OUTLOOK: Potential Weather Impacts To Midwest Corn/Soybean Harvest

Back on August 29th, I made a post, using some cutting edge weather research projects, about the risk of frost/freeze in Upper Midwest circa 10-20 September time frame that could be better ascertained  in 5 to 7 days.   This risk can be now more specifically clarified.

I'll state at the onset this risk has been confirmed of at least a low end to moderate magnitude event(s).   A several percent loss of crop yields appears plausable with 10 percent or more not of the question in sections of the Upper Midwest farm belt. 

First, we need to outline how slow the corn and soybean crops are maturing from normal year.

• Larger scale outlooks and agencies state that overall...crops are running 10 to possibly 15 days behind schedule.
• Several local contacts (farmers, etc) and analysis for Eastern Iowa and NW Illinois suggest  were running 14 to 21 days behind schedule.
• My analysis and assessment - overall...the corn and soybeans are closer to at least 14 days behind schedule.
• This indicates a signficant impact to crop yields would occur with a heavy frost before 1 October based on near normal weather temperatures.

Second, temperatures at or below 34F for a few to several hours will negatively impact harvest yields with below 32F having major impacts.   These impacts will become increasingly less after ~20 October as crops approach harvest.  Minor impacts are suggested after 1 October based on temperatures remaining near normal the next 3 to 4 weeks.

Below are two charts, courtesy of Iowa Environmental Mesonet Web Page that provide some useful climatological information we will use.

This picture illustrates that no matter what occurs (i.e. record warmth...red line), there is no way we can recover to a near normal date of maturity for harvest.    Local analysis and data I have suggests were closer to being behind by ~500 Growing Degree Days.

Next, below is a chart that illustrates the climatological probability of 34 degrees at Ames Iowa.  Note on an average year, the first 34 degree day is 30 September.    

The problem is...the regime were in (which current local, regional teleconnections suggest little or no change next 3-4 weeks) data indicates for the past 3 months were running almost 2 standard deviations (or roughly 95%) colder than normal based on our climatological data.   So...if this changes little or none...the first frost could occur roughly in the 15-20 September time period from second chart above.    This is consistent with my earlier post on August 29.   Analysis output below continues to compare favorably with this time frame.

Most of the methodology and data of analysis for creating energy spectral analysis is beyond the scope of this audience.  However, I will broadly describe the process.

Visualize a rough sea surface with wave tops, bottoms, crests and bottoms protruding and oscillating up and down.  These can be akin to various energy waves in the atmosphere (which is a gaseous fluid).  

1. We can attempt to separate these interesecting wave crests (beyond model ensembles...if you've heard of this term) in 4 dimensional framework with time used as T-2T...T-1T...T...T+1T...T+2T (those statistics savvy folks may note this approximates 95% of the energy variance in time) .   
2. Then, we can correlate this output with current existing, proven weather teleconnections (these include ENSO...plus dozens more).
3. Finally...we can compare and potentialy integrate what are described as forecast analog database of past similar flow pattern that match our teleconnections  and create statistical matches that can have value out to roughly day 15 in certain circumstances.  This step can aid in confidence at times and even potentially intensity/magnitude.
4. Applying Human analysis (I'll leave it at that):  The key is incorporating human capacity to identify signature feature/pattern recognition and interactions, integrate the assess smoothing of the data from the shapes and sizes of the energy spectrum.  This is roughly analogous to the observer looking at a picture, lets use the Mona Lisa from, say a distance of 150 feet.  One can make out this a picture of person...possibly a lady.   This is like looking a few weeks out (e.g. 29 August post).   As the event approaches...the observer is walking closer to picture and makes out more of the features.   Through education and training, we can diagnose more features and sooner of what were looking at than a typical person.

EVENT 1

So...what does this mean for our situation?   Within 12 days...there is enough skill we can apply some techniques to show the reader some supporting graphs outlining what may occur.   Below is such...called a model ensemble (a series of perturbed solutions from slightly different initial conditions)...this is a poor but initial marginally adequate description of us trying to account for the sea state we described above.   Below is a chart suggesting some supporting evidence (can't stress enough DO NOT TAKE THE DATA LITERALLY) of a upper trough circa ( 1.5 days) of 12 September).   The actual output used routinely is considerably different than this.

This picture, if interested (couldn't get to load properly here) is at Penn State Electronic E-Wall.

Here's the key facts based on analysis of data through 18Z today (early afternoon)...for ~11-13 September.

There is slight chance (~15-25%) of areas of frost 34 degrees or cooler over portions of the Upper Midwest farmbelt.

There is a less than a 10 percent chance of 32 degrees or cooler over portions of the Upper Midwest farmbelt.

KEY CAVEAT:  We are in the fall transition period...the atmosphere routniely makes major step function changes (energy radiation imbalances as sun is moving south at 25 miles per day) which is suggested may occur in this time period.  This is noted as output suggest  a very large standard deviation of output.  

Specific clarification of the intensity and magnitude of midwestern cold should be able to be diagnosed within next 1.5 days.

This first event has the potential that it could reduce crop yields by a few percent in certain areas.

EVENT 2.

A more potentially significant event may occur 3 days of 18 September.   This risk of a moderate (or even greater) frost/freeze event should be clarified within next 4 days.   Statistical analysis suggests a 5 percent loss is quite possible based on temps AOB 32F and the anticipated stage of crops.  It is also possible isolated double digits losses could happen.

Here's a picture that very, very, very roughly outlines the energy centre.  Again ensembles data is only broadly used in this analysis process.

If there is still enough interest, I can post updates on this risk the next 3 to 5 days.   FWIW.  I'm told the markets are keenly aware of this risk.

Past experiences indicate the markets often react 1-3 days after this forecast research technique...as these are then picked up by specific model output and applied in more established and convectional methods.  

Nichoman