UAV Imagery – Resolution

We have all heard the comment for the last few years. “UAV, you must have one or your ag business is done!” Well, it seems like a lot of business have grown and prospered without a UAV? The idea I am throwing out here is not DRONE bashing, but DRONE embracing. As I see it, after many hours of conversations with UAV users, there are 3 primary uses for the data. 

1. Scouting: Using a UAV is similar to boots on the ground or field visualization, providing data such as stand count, weed patch locations, possibly weed ID, drainage issues, drift issues and insect damage.

        2. Application: For example, acquiring high-resolution imagery and converting the spectral information in that image into a lower resolution variable rate application map.

        3. Topography: A great way to get updated topography data on your fields. Most of the data here is LiDAR “like” data. When done correctly, this is a great way to collect your topography.

  So, what does all of this mean as a UAV user or a potential UAV user.

  In the case of “scouting”, set the UAV system up to get the highest resolution you can. This data may not need to “cover the whole field”? This data may not even be pictures, it may be streaming video or just flying out to an area to see an event (no recording). Here are a couple examples. 
            -Looking at weed control when the fields are too wet to walk or drive. In this process you may find out your pesticide worked or that it did not work? Giving you the knowledge, in a timely manner, to make that next application decision. 
            -Second example has been the rally cry of why all farmers need a UAV, gathering stand counts. I need to elaborate on this topic, it is a bit of a sore spot for me. My opinion, if the field needs re-seeding, the farmer will park his butt in the tractor seat and re-seed it. Writing a VR re-seeding map would be great, but most of the time if it is a “spot” re-seed, you need to raise and lower the planter, so you do not destroy healthy areas of the field. I don’t know of many VR Controllers that can raise & lower the planter by a VR map. More so, the VR map in the background, that would be an awesome visual “suggestion” for the operator of the planter. Stand count maps are much more useful for in-season scouting and end-of-the-season analysis of seed, fertilizer and production. The need for “Stand Count” data for reseeding is nice, but not necessary. I see the biggest advantage in the analysis component of the data, looking at both agronomic and finical returns per plant. The main component “Scouting” is you want high resolution images and videos.

  There are many more uses out there, I just wanted to hit a couple that I see value in. 

  In the case of the “application” data, the need for the highest resolution is of less value. In my opinion, there is very little need for images that are more than 1-meter (100cm) resolution. The high-resolution data is going to contain too much information. Thus, requiring additional image processing to get rid of trivial data such as wheel track and stop/start of the planter. Again, this is for application and most application equipment only updates the rates once per second. Let’s put that into distance or resolution value. On a “Planter” at 5 mph, the rates are getting updated 7.3 feet and on a “Floater” fertilizing at 15 mph, rates are updated every 20 feet. Not to mention the width of the application sections. Now ask yourself, have I “Calibrated” my planter, seeder or fertilizer application equipment to make sure it is applying the rate in the correct spot (boom offsets and look ahead times). Now stop, ask yourself, do I need “Sub-Meter” resolution to write my VR map? Getting a VR application map of 10-foot to 20-foot resolution will be acceptable for most operations. 

  “Topography” is a very technical piece of data to collect with a UAV. Read and follow the manufactures instructions to collect best and most accurate data. Most UAV’s are not collecting actual “LiDAR” data but are taking several images of the same location and using a processing method know as photogrammetry (PG) to calculate elevation values. A quick explanation: LiDAR uses its own light source (laser) shooting it down and using a special camera (sensor) to pick up refection. LiDAR does a fair job of shooting through sparse canopy to see the ground. Photogrammetry (PG) relies on natural light and uses standard cameras. PG relies on photo overlap and the processing techniques to create “triangles” to calculate elevation. PG does not measure through sparse canopy very well, it needs bare soils or very low stubble. Here is a link to a very good article “LiDAR vs Photogrammetry”. Both systems can be used to create LAS files that can be used in ADMS. The option of creating LAS files is going to be a process of the stitching software. Neither system will measure through dense canopy and users may still need RTK readings to get elevations at outlets or culverts. Understand the processes being used and know the strengths and weaknesses of the equipment you are using.

  Next time you are out collecting UAV data, collect the best, highest resolution data you can (keeping in mind what you plan on using the data for). Process and stitch the data at the high resolutions up front. If needed, resample to a lower resolution later. At the point the data is moved from the stitching software over to the VR mapping software package such as ADMS for Application (zones), resample the data and export it out at lower resolution (suggest 1-meter). Also, if you are doing Topography, make sure you followed the procedures that best fit your need. Hope this gives you some ideas of how to use your UAV to the best of its ability.