exg <- function(img, r, g, b){
r <- img[[r]]
g <- img[[g]]
b <- img[[b]]
vi <- (2 * g - r- b) / (g + b + r)
return(vi)
}
ngrdi <- function(img, r, g, b){
r <- img[[r]]
g <- img[[g]]
b <- img[[b]]
vi <- (g -r) / (g + r)
return(vi)
}UAVs in ecology
UAVs are a valuable tool for ecologists and many companies and organisations have either their own vehicles or access to them. They are a relatively low cost and therefore accessible tool.
There are a wide range of software tools, platforms and more to help with collecting imagery and processing it, but some of these can be quite costly.
UAV imagery
Up to date colour (RGB) imagery is invaluable for habitat mapping, assessment and monitoring. Some proprietary software provide additional tools for further processing imagery, and vegetation indices are one of those. Vegetation indices may add value to the RGB imagery collected by UAVs for assessment and monitoring.
What are vegetation indices?
A vegetation index (VI) is a spectral imaging transformation of two or more image bands designed to enhance the contribution of vegetation properties and allow reliable spatial and temporal inter-comparisons of terrestrial photosynthetic activity and canopy structural variations.
Measurements of vegetation attributes include leaf area index (LAI), percent green cover, chlorophyll content, green biomass and absorbed photosynthetically active radiation (APAR).
Normalised Difference Vegetation Index (NDVI) is a well know vegetation index, but this relies on a Near Infra Red (NIR) sensor to calculate. So can we use RGB imagery to derive vegetation indices that may communicate the same information as NDVI
Hunt et al., [ 16] used the normalized green–red difference index (NGRDI) from RGB images to estimate the biomass of corn, alfalfa, and soybean, and found a linear correlation between the NGRDI and biomass
Woebbecke et al. [ 18] investigated the capability of several colour indices to distinguish vegetation from the background, and found that the excess green vegetation index (ExG) could provide a near-binary intensity image outlining a plant region of interest
Several indices can be used to estimate leaf N concentration in wheat, and possibly other plant species as well. Jian et al (2018)
Using open source tools to derive vegetation indices
As UAV imagery is raster data, it is relatively straight forward to calculate vegetation indices. The maths is relatively straight forward, so it is possible to code your own indices in your favourite programming language.
In this example I have used the programming language R, but the same is possible in python, or any other language that can read spatial raster data.
This is also very doable in QGIS usign the raster calculator, and I am sure ESRI has a similar facility.
Quick note on R
R is a programming language which provides a wealth of options for data reading, writing, analysis, stats visualisation and more. It is free open source software with a wealth of contributed libraries for doing an enormous array of data related operations.
R functions for vegetation indices
Here are two examples of vegetation indices turned into functions in R. As you can see the maths is pretty straightforward, even if the code looks somewhat more complex.
By reading a raster into R I can then apply these indices to the raster data and generate an output for each index.
Sonning
The example is from a small Local Nature Reserve near Reading. The imagery was acquired in September 2019, which you may remember was a very hot year. As such the vegetation was very parched, which is perhaps not the best conditions in which to be testing vegetation indices.
We can apply each index to the RGB imagery and view the outputs.
First the Excess Green Index:
and second the Normalised Red Green Difference Index:
Normalised Red Green Difference Index
We can tell it was very dry! The trees are the greenest things on the site, but not very green! We can see the pond well defined though and the paths stand out. We can also see some variation in the grassland (probably an other neutral grassland in the UKHab classification).
So we might be able to estimate biomass from the NGRDI and maybe help define areas of interest using the EXG index.
What about NIR?
Adding NIR bands to imagery would allow the calculation of NDVI and other vegetation indices that rely on NIR data. Cameras with NIR sensors on can bought for a relatively modest cost and added to existing drones. This can open up additional data and indices.
Advantages of using open source tools
How can I get this for my imagery? The code is simple to write and relatively quick to run, given sufficient amounts of space for the imagery, which can get quite large. It could easily be incorporated into a simple web or desktop app, or incorporated into other work flows and reporting. The outputs can also be read into GIS of all flavours for further analysis, as the data are GeoTIFFs, like the input data.
So it is a flexible and lightweight system. And it can be adapted and improved as required.
The key advantage is that as R / Python are free and open source, the cost is low and you are not tied into a single provider for your analytics. And as you learn more, you can develop new tools and analyses to suit your needs.
Summary
UAVs are valuable tools for ecological monitoring and assessment, relatively low cost and therefore accessible. Add in low cost open source tools can add further value to the imagery you collect.
Vegetation indices have been widely used to understand vegetation health and change, which will be increasingly important, not least in biodiversity net gain.