Does the time of day you plant matter?

Working with a couple of other local area farmers in the very early days of mapping not only harvest/yield data but also planting data.  (For the younger generation, this was before there were fancy (and expensive) data-ready monitors in cabs.  I have tried to squeeze an old laptop in a tractor cab way more times than I care to remember.  Does anyone remember the Panasonic toughbooks?  I think we could find them used on Ebay for $300-$400.  But I digress from the topic at hand.

One data piece resulting from putting a GPS logging device in the planter tractor is an exact time each point in the field was planted, captured by the GPS timestamp.  This allows for some interesting possibilities.  I credit Dr. Randy Taylor (at Kansas State University at the time) for driving a lot of my initial interest in data collection and analysis.  

In this case, as a farmer group we were having a roundtable discussion and the subject came up about planting early in the morning - if that would possibly affect final yield. Randy jumped on it.  Gathering the combined corn seeding data from a single year for the three of us farmers, he worked his regression/modeling magic and came back with a result that corn planted at 6am yielded about 6.5 bushel per acre less than the optimum time - which peaked around 1pm.

So..  planting time does matter.  At least on a single year data set from three northeast Kansas producers.  As a data person though, I always cast a wary eye towards results, esp. results that represent knowledge I have not seen before.  Unfortunately in this case I do not have the regression model or raw data that Randy used, and considering how often he probably worked on this kind of thing I doubt he does either.  So I can’t really properly discuss our analysis methods or results.

It was before I even understood the basics of modeling in a multivariate regression framework and I don’t have other useful info like the p-values of the coefficents or which coefficients were used.  I think we had about six or seven.  

As with the planting direction example, at the time the effort required to pull all this data together so that it can be effectively analyzed was not feasible at a commercial level, but it sure was fun!   One of my biggest hopes a decade later is that we have made solid progress as an industry on the data collection side (tools like Fieldview/MyJohnDeere/SMS’s batch export abilities) give me hope this kind of analysis and what we may be able to learn from it will become more routine.