Isolating red leaf pigments with multispectral imagery

Agriculture
1

Hello,

I have a DJI Mavic 3M drone and I’m dealing with an experimental crop that has red leaves at maturity. I am attempting to track the transition from green pigmentation of the leaves to red among different varieties once the crop approaches maturity. Is there an index that will highlight the red leaves over the other colors? For example, my understanding of NDRE is that it shows the amount of chlorophyll (i.e. greenness) a plant has. Is there a way to measure the display the amount of redness a plant has? Essentially I want to create a map highlighting the different spectrum of redness observed in the field by the plants.

Thanks,

Marcus

1 Like
2

Hi Marcus,

great question! NDRE is not directly showing the “greenness” of the plant but rather the amount of red light it absorbs for photosynthesis. Normal (green) plants don’t use much of the green light which they just reflect that’s why they appear green to our eyes. The color of a leaf is determined by the pigments present in the leaf cells. Green leaves appear green because chlorophyll, the primary pigment involved in photosynthesis, absorbs light in the red and blue parts of the spectrum and reflects green light.

image
image1707×1101 108 KB

For your red plant its different, as this is using more of the green wavelength of the light for its photosynthesis and reflects more of the red light, that’s why appearing red to our eyes. Red leaves, on the other hand, contain pigments such as anthocyanins in addition to chlorophyll. Anthocyanins can absorb light across a broad range of the spectrum, including the green part of the spectrum. This means that red leaves can absorb and use more green light for photosynthesis than green leaves can.

You can start with a Red / Green Ratio in the Custom Index Calculator:

image
image1016×758 119 KB

And as you can see that highlights the red plants in my example:

image
image1353×1020 158 KB

3

You can also try a variation of the NDRE= (NIR+RedEdge)/ (NIR−RedEdge)
which would swap the RedEdge with the Green band:

GNDVI = (nir - green) / (nir + green)
​

image
image1091×963 101 KB

RGB vs GNDVI:

image
image1353×966 145 KB

NDRE vs GNDVI:

image
image1348×1018 154 KB

4

Julius,

Thank you so much! That was super helpful, particularly the red/green ratio.

A follow up question I have: is there a way in Pix4D Fields to calculate percent cover or area of a field that is covered by a given range within the index? Essentially I would like to be able to determine what percent of the crop has turned red and track that through the growing season to optimize harvest planning.

Additionally if I may make a suggestion, it would be super helpful if the scale of the slider bar on the “Visual” tab of a given index automatically adjusted based on the range of values observed. For the red/green ratio, most of my targeted falls between 1 and 2, but the scale goes from 0-24, making it hard to toggle the upper and lower end of the ranges. NDVI, for example, is displayed between -1 and 1 since it is normalized. I know I can manually input the numbers if I know what my range is, but at this stage I am still trying to make sense of what the raw values actually mean to understand what range I expect to observe.

5

A follow up question I have: is there a way in Pix4D Fields to calculate percent cover or area of a field that is covered by a given range within the index? Essentially I would like to be able to determine what percent of the crop has turned red and track that through the growing season to optimize harvest planning.

Yes but you need to use the statistics export for annotation. Meaning:

  1. Draw an area annotation over the index area you are interested in
  2. Export → Statistics and you will get the visible are like so : https://support.pix4d.com/hc/en-us/articles/4409231699345-Exporting-statistics-PIX4Dfields#:~:text=the%20histogram%20range.-,Layer-statistics.csv,-This%20file%20contains
6

Maybe you can create your own maturity index based on the green and red bands like a normalized red/green index (NRGI) = ((red - green)/(red+green)) ?