Mapping trees is a tricky thing

[Hawkwind]

I've done some flights to map wooded areas. Trees are difficult to create orthomosaics. This time of year here, the leaves are off of the trees, making it easier to see the ground.

Areas where there were clearings, roads, or buildings turned out great, but the photogrammetry processing software couldn't handle the photos with only trees.

I started with default overlap settings in dronedeploy's flight app, then used suggested 80% front and 80% side overlap for doing trees, which was somewhat better, but not great. Anyone here had good success mapping forests? If so, what seems to work best for you?

Additional details:
Platform: Mavic Pro
Software: dronedeploy for flight automation and mapsmadeeasy for processing.
Altitude: ~390ft AGL

 

[Dave Pitman]

The thing to remember is that the overlap setting is applied at the take-off point elevation. 80/80 overlap at that elevation is not 80/80 at the top of the canopy. The higher the object is, the less overlap.

I have not used DD for flight for awhile, so I don't know if they have an offset setting to allow for tall objects. If not, you might want to try 85/85 to get a higher overlap on the trees.

 

[Hawkwind]

Thanks @Dave Pitman

I retested with my sample woods and 85%/85% worked well. The terrain is flat, so if you introduce hills, you may not get the same results.

 

[Pawel]

You may also like to fly higher, provided uour target GSD allows. This will result in a bigger ground imsges footprint and more details for processing software to play with.

 

[KentuckyAbove]

I’m new to drone mapping as well and didn’t know what to do for this. I’ve got an area I’ve been trying to map for a local emergency management that’s all hills and trees. I didn’t know what to do to make it come out right. Thank you all for the informative replies.

 

[Dave Pitman]

You may also like to fly higher,

In the US, we are generally limited to 400'agl. The OP is at that limit.

 

[Hawkwind]

@Pawel Yes, that is an important consideration, flying higher will give you more overlap, but I'm limited as @Dave Pitman said to 400ft AGL. Unfortunately trees don't count as structures, so I cannot fly 400ft above the trees. This would help, since I'm dealing with hills. Maybe I should apply for a waiver to fly 400ft above trees...

@KentuckyAbove Yes, this forum is a great source to share information.

 

[Hawkwind]

Interesting observation: I was able to capture 85% front and 85% side overlap, then I culled every other row prior to processing and still got excellent results. This was effectively 85% front and 72% side- very good final map. Good to know if you are looking for ways to reduce the amount of data being processed without performing another flight...

 

[Hawkwind]

Follow up to the original post, has anyone found other photogrammetry/mapping software worked better or worse than MME for generating a 2D orthomosaic of bare trees? ODM, DroneDeploy, Pix4D, etc.?

 

[BigAl07]

Follow up to the original post, has anyone found other photogrammetry/mapping software worked better or worse than MME for generating a 2D orthomosaic of bare trees? ODM, DroneDeploy, Pix4D, etc.?

I use AgiSoft MetaShape Pro and process my own. If you'd like to share your Data Set I will process it for you so you have a comparison to work with.

 

[Hawkwind]

Aha!

I found that DroneDeploy was able to process the images and successfully create a complete 2D orthomosaic using 80%/80% overlap.

So, here is a clear example where DD's algorithm performed significantly better than MME.