Disease in Vineyard: best indice?

Agriculture General Ag Questions/Troubleshooting
1

I’ve surveyed a particular vineyard plot in which I know there is a strip of yellowing vine, due to disease. I have processed the plot through various indices including NDVI, GCI, ReCI and NDRE but the location of the stressed canopy is not really showing. Other than already known lower canopy cover (and hence soil showing) I can’t see anything specific that I would call out, if we didn’t already know there was a problem.

EDIT: the GRAPES are showing the yellowing, which is a disease indicator, believed to be Powdery Mildew. I would expect to pick this up on MS.

Is there a better / recommended index that I could use?

In the screen capture below, the known issue is roughly in the box. Note that the red colour in the map index is soil. I’m looking at the vine (green) to try and detect a variation in the reported colour and index value.
It would be really useful to be able to get the index value for a map point selected by the mouse…

Any advice would be hugely welcome!

Screenshot 2023-07-27 at 13.23.29
Screenshot 2023-07-27 at 13.23.291920×1522 689 KB

2

How does the RGB imagery look like at this place ?

3

Good question, I’ll run the RGB now. Normally I only run TIFF for multispectral.

4

Screenshot 2023-07-27 at 13.52.25
Screenshot 2023-07-27 at 13.52.251920×1276 513 KB

It’s difficult to see it clearly but you can start to see it here.

5

Zoomed out, RGB

Screenshot 2023-07-27 at 13.54.29
Screenshot 2023-07-27 at 13.54.291920×1295 494 KB

6

Adding to this: I found a ‘Powdery Mildew Index’ with the following formula:

PMI = (R520-R584)/(R520+R584)-R724

So think for Pix4D that would translate roughly as (Green- Orange?!)/(Green+Orange?!)-Red

Obviously ‘orange’ is an issue…any ideas folks?

7

Those are the wavelengths for the m3m bands:
image

So for R520 you could try with the Green band, but its still a couple of nm off.

8

In general for this kind of dataset, you biggest enemy is the area between the rows, because it will throw of the complete index visualization. You can try with MagicTool to detect it, and then Trim it away so that only the data from the actual rows will be left for analysis. The next version of PIX4Dfields will have tools to help with that.

1 Like
9

Perfect, thank you!

10

I’ve got around that by creating a boundary which has sub-boundaries per row. It’s time consuming but does remove the ‘in between’ areas, focussing the index on just the vine.

11

I would suggest to take sample on the field where the area got stressed due to disease. Then you overlay with indices (NDVI, NDVIRE, LCI, etc) and play with threshold value indices (classification). Hope you find correlation with that.

1 Like
12

Hi Julius, was there any update on this? Would love to know

13

Hi,

We introduced a new Zonation tool that might help you identify zones within the rows. If the field is rather flat, you can try the following:

  • Process the imagery using Accurate processing.

  • Go to Operations, Zonation, and choose the DSM layer.

  • Set the histogram so as to only leave pixels that are of vineyard rows
    image

  • Click continue.

  • Here you can:
    - Set 1 zone and save the Operation…
    - Export the Operation as Shapefile.
    - Import it again as boundary, now you can use it to trim the Index layer.

I hope this makes sense.
You can watch a similar workflow to identity weeds on fallow: Weeds on fallow detection (Green on Brown) - PIX4Dfields

14

Thanks Jose, I’ll try that!