Precision ag has plenty of room for error

Learning from your mistakes is an important part of education, especially in precision agriculture, so I thought it would be entertaining to hear about one of the more stupid precision farming mistakes I have made.

One of the fields at Kirkwood Community College, where I taught for 15 years, was known as the Beef field. I’m not sure why, but that’s what it had always been called.

It was close to campus and therefore was easy for students to participate in field activities, such as soil sampling or yield mapping.

It was in a corn-soybean rotation with a variety of chemical and nutrient trials and demonstrations.

The best part of this field was that it had been in consistent use for many years, which meant more than five years of yield maps.

Too often the college would buy farmland for use in its farm and within three years the boundary would change as the college built a new horse arena or hotel or sell it to the local school district for a new middle school.

Mistakes in agriculture come in many forms. In precision agriculture there are more opportunities to goof up, but often they aren’t ones the neighbours will see. | File photo

This meant that we did not have many fields with a consistent boundary for more than three years of yield data.

But the Beef field, with five years of yield data and good variability, made an excellent field for demonstrating analytical techniques. It had a distinctive rectangular shape that made it easier to recognize for students and supposedly by me, the instructor.

One fall after all the fields had been harvested, I was responsible for the initial yield data processing. This included exporting the yield map file from the combine display and processing it in the desktop software.

The software worked well and allowed me to process one field at a time. Most systems allow the field entry of a field name, operator and other pertinent information, but in this case the raw data did not have a field name associated with it, so it was up to me to identify the field.

When I got to one specific field, the characteristic rectangular shape told me automatically that it was the Beef field. I did a review of the yield data, deleted some outliers, checked the statistics and classified the data to a legend based on our local yield standards.

The problem occurred when I added the Beef field yield layer to the farm map. It was on the wrong side of the road.

The Beef field is directly north of a main paved road, so not only was it easy to identify by its shape, but also by its location.

However, the newly processed field was directly south of the road, about 30 metres south of its “correct” location.

In these early days of GIS and mapping, there was something known as datum shift.

This occurred when the wrong datum was assigned to a data layer.

This sometimes expressed itself when a road incorrectly positioned itself running through a field or two field boundaries did not line up.

The most common datum shift was usually an error of about 30 metres.

Knowing that this was possible, I made the assumption that this was the problem. How else could the Beef field end up on the wrong side of the road?

My first task was to call software support to ask how this could have happened. In talking with the tech support, the gentleman assured me that it was not possible. The software automatically assigned a correct datum to all data layers.

His only response was, “maybe it’s the wrong field.” Well, that couldn’t be, since I recognized the Beef field, and the field on the south side of the road was not even owned by the college.

I actually edited all yield points for the entire field so it fit where I thought it should be.

The field in question happened to be on my way home from work. I looked that night and noticed that the field to the south was about the same size and shape of Beef and it was a cornfield. The next day I mentioned the situation to the agronomy instructor.

“Oh, sorry,” was his response.

He had forgot to mention that we had indeed harvested that field to the south.

The neighbouring farmer wanted a yield map so we helped out by harvesting his field.

After reprocessing the raw data, I left the yield where it was supposed to be on the south side of the road. I also found the raw data for the real Beef field and processed it to the north side of the road.

I always wondered about calling that tech support person up and explaining but never did. This was a case that precision happens … in spite of me.

About the author

Terry Brase Pag's recent articles



Stories from our other publications