Modeling tree leaves and branches in Feko

lukedj2

Hi,

I have been using Feko for radar simulations for a research project on remote sensing. I am trying to simulate the propagation of a 1-2GHz signal through tree canopy. I have been able to successfully simulate the radar backscatter from a complex tree model (picture attached), but I do not want to simulate every branch and leaf of the tree crown. Is there a way of simplifying this geometry without sacrificing too much simulation accuracy? Currently the tree mesh contains over 1 million triangles.

I was thinking of modeling the tree as a ellipsoid crown attached to a cylinder trunk, but I want the RF signal to be able to reflect from inside the tree crown as well, not just at the edges of the region. Is there a way of implementing a dielectric "fog" medium where the RF signal can scatter within the medium at random?

Jaehoon

Hello Luke,

Q1) Is there a way of simplifying this geometry without sacrificing too much simulation accuracy? Currently the tree mesh contains over 1 million triangles.

-> I believe you are using MLFMM for your Radar simulation. A possible way to simplify the geometry without sacrificing the accuracy is to remove the number of the extremely small sized meshes.  Normally, the small sized meshes are generated from very detailed geometry parts. 

Please 1) apply Unlinked meshes to your geometry after meshing, 2) Exclude the original geometry 3) Apply "Use model mesh" to the unlinked meshed 4) Select all the mesh faces from the detail tree 5) Merge all the mesh faces (The material properties of the mesh  faces should be the same) 6) Remesh the merged mesh face 

The above steps are the same in the following blog.

Q2) Is there a way of implementing a dielectric "fog" medium where the RF signal can scatter within the medium at random?

-> I don't fully understand your question, but any dielectric medium can be defined in Feko and assigned to the geometry region (3D).  You can solve the region with SEP or VEP.

 

Best regards,

Jaehoon

 

lukedj2

Jaehoon Kim_22316 said:

Hello Luke,

Q1) Is there a way of simplifying this geometry without sacrificing too much simulation accuracy? Currently the tree mesh contains over 1 million triangles.

-> I believe you are using MLFMM for your Radar simulation. A possible way to simplify the geometry without sacrificing the accuracy is to remove the number of the extremely small sized meshes.  Normally, the small sized meshes are generated from very detailed geometry parts. 

Please 1) apply Unlinked meshes to your geometry after meshing, 2) Exclude the original geometry 3) Apply "Use model mesh" to the unlinked meshed 4) Select all the mesh faces from the detail tree 5) Merge all the mesh faces (The material properties of the mesh  faces should be the same) 6) Remesh the merged mesh face 

The above steps are the same in the following blog.

Q2) Is there a way of implementing a dielectric "fog" medium where the RF signal can scatter within the medium at random?

-> I don't fully understand your question, but any dielectric medium can be defined in Feko and assigned to the geometry region (3D).  You can solve the region with SEP or VEP.

 

Best regards,

Jaehoon

 

Hi Jaehoon,

Thank you for your detailed response. I am using RL-GO as a solver. I can look into simplifying my mesh.

I can clarify my second question. By dielectric "fog," I am referring to how one might model a region with a large number of scatterers, like the canopy of a tree with many small branches and leaves. I do not want to specify the locations of all the scatterers, but instead I want to approximate their average backscatter response if they were randomly spaced out within a region. (Paper figure attached.) I am under the impression that Feko does not have the ability to  model a random region like this, but I was wondering if I was wrong.

Thanks,

Luke

Jaehoon

Luke Jacobs said:

Hi Jaehoon,

Thank you for your detailed response. I am using RL-GO as a solver. I can look into simplifying my mesh.

I can clarify my second question. By dielectric "fog," I am referring to how one might model a region with a large number of scatterers, like the canopy of a tree with many small branches and leaves. I do not want to specify the locations of all the scatterers, but instead I want to approximate their average backscatter response if they were randomly spaced out within a region. (Paper figure attached.) I am under the impression that Feko does not have the ability to  model a random region like this, but I was wondering if I was wrong.

Thanks,

Luke

Hello Luke,

Q: I am under the impression that Feko does not have the ability to  model a random region like this, but I was wondering if I was wrong.

-> Thank you for explaining about the fog material with the picture (above). 

If a tree can be assumed to be a periodic structure (unit cell) and the unit cell is smaller than lambda/2, you can use the characterized surface medium (.tr: transmission and reflection coefficient file) to model the whole trees.

The characterized surface medium is well explained in the Feko example guide, H.5 Characterised surfaces for FSS as shown below.

Best regards,

Jaehoon