Flight Procedures

Best Practices for Flying Over Large Plots

In large and homogeneous plots, some recommended flight practices are the following:

  1. Fly as high as possible. The loss of resolution is not decisive in this type of crop.

  2. Not sure about the overlaps used for the flight, but it should be at least 60% sidelap and 80 or more overlap if the flight speed allows it (I don’t think there will be any problems when increasing height).

  3. Split the plot into several zones. Clear areas must be found to divide, e.g. by paths. If this type of division is not possible, consider dividing the flights into groups of 3 or 4, in case there is any problem in joint processing, having a second alternative with that group division.

  4. Configure the flight perpendicular to the crop rows whenever possible, although the geometry and dimensions of the plot may not allow it or may complicate the operation. In this case, being able to complete the flight should take priority.

  5. In that plot that requires multiple flights, try to make the flights symmetrical to solar noon (this is what should always happen when there are multiple flights), although we know this is not always possible.

  6. As a rule, calibration plates must be correctly photographed and should occupy most of the image. The camera should not be rotated in relation to the plate.

About calibration plates

In the example imageset, after looking at the data, there are apparently no issues with the flight, time, date, and consequently the process is launched. The result is shown below:

The mosaic looks good, but those strange red lines appear. After looking at it, it can be seen that it’s due to the calibration plates. One of those images was photographed like this:

Where you can clearly see a shadow on the calibration plate. That, added to the fact that the program has miscalculated the mask (partly due to the shadow) is what causes those red lines to appear.

In many cases, numerous photos of calibration panels are taken for fear of getting a bad photo, but this excess of images can cause more problems than benefits. For example, in the plot in question, 7 photos of the panels were taken at the beginning of the flight:

and 15 pictures at the end:

When 2 or 3 at the beginning and another two or three at the end would really be enough. Excess, especially if there are errors, makes automated processes more likely to fail and therefore require repeating tasks or doing them manually.

Hubiese bastado con las fotos 1, 2, y 3 al principio, y las 910, 911 y 912 al final. El resto puede ser eliminadas. Esta decisión debe ser tomada por el piloto/concesionario antes de subir las fotos y sólo mantener aquellas que sean correctas. Si no se hace así, muchos procesos pueden fallar. También es importante que las imágenes de las placas (las que se van a usar) sean tomadas en las misma condiciones de iluminación. Se ve que las fotos 906 y 907 son muy diferentes, lo que hace que la calibración radiométrica tampoco sea correcta si intervienen las dos imágenes.