Five decisions to make before a multispectral drone survey
Most problems in a vegetation map were fixed, or caused, before the aircraft left the ground.
Multispectral surveys fail quietly. The flight looks fine, the mosaic stitches, and only later does someone notice that the index values in the east of the map do not match those in the west. These are the choices that prevent it.
1. What size of thing are you trying to see?
Ground sampling distance is set by flying height. Individual plants need a low flight and many more images. Field-scale patterns do not. Decide the smallest feature that matters, then choose the height.
2. How much overlap?
Uniform crops and water are hard for photogrammetry software to match. Higher overlap costs flight time but saves the survey. Over dense, uniform canopy we plan more overlap than the default.
3. When to fly
Fly close to solar noon when you can, to keep shadows short and light steady. Broken cloud is the worst condition: the illumination changes from frame to frame.
4. How will reflectance be calibrated?
Raw pixel values are not reflectance. A sunlight sensor on the aircraft records the incoming light for every frame, and a calibrated reflectance panel photographed before and after the flight anchors the values. Without calibration, surveys from different days cannot be compared.
5. How will position be controlled?
RTK positioning gives each image a precise location. A few surveyed check points on the ground tell you whether to believe it. If the map will be compared with a later survey, this is not optional.
The short version
Write down the answer to each of these before the survey, and keep it with the data. In a year's time, when someone wants to repeat the survey and measure change, it is the most valuable page in the report.