saltbush

saltbush processes drone imagery to calculate spectral diversity values as part of the assessment of the ‘spectral variability hypothesis’ or the ‘spectral-biodiversity relationship’. This was specifically implemented to connect from-the-sky diversity with on-the-ground diversity as measured in the AusPlots project. There are also functions to calculate on-the-ground diversity built on the ausplotsR and vegan packages. The statistical methods for connecting diversity sampled on the ground to diversity sampled from the sky are still developing. This package is aimed at making from-the-sky diversity methods development easier for researchers across the world.

Specifically we have used this package in a test in the Australian arid zone. The results suggest that some of the methods below perform much better than others. See our paper Placeholder to link to the preprint once that’s posted.

Installation

 install.packages("remotes")
 remotes::install_github("traitecoevo/saltbush")

Usage

1. Direct the package to the input files in this case we use a drone image from Fowlers Gap, NSW, Australia

The example data in this case is 5 bands sampled from a drone

create_multiband_image("inst/extdata/create_multiband_image/",
                       c('blue', 'green', 'red', 'red_edge', 'nir'), 
                       output_dir = tempdir(),
                       make_plot = TRUE, return_raster = TRUE)

The resulting multiband image can be combined with an area of interest shapefile:

raster_files <- list.files("inst/extdata/example",
    pattern = '.tif$', full.names = TRUE)

aoi_files <- list.files("inst/extdata/aoi",
    pattern = 'NSABHC0009_aoi.shp$', full.names = TRUE)

2. Extract pixel values from the area of interest in the image

pixel_values <- extract_pixel_values(raster_files, 
                                     aoi_files,
                                     c('blue','green','red','red_edge','nir'))

head(pixel_values)  
#>    site_name       blue      green        red   red_edge        nir aoi_id
#> 1 NSABHC0009 0.05511368 0.05600401 0.07633078 0.06699516 0.08160288      1
#> 2 NSABHC0009 0.06224341 0.06684195 0.09760323 0.08534835 0.10119278      1
#> 3 NSABHC0009 0.05985198 0.06332091 0.08991133 0.07894940 0.09560681      1
#> 4 NSABHC0009 0.05927841 0.06296582 0.08709560 0.07646009 0.09347322      1
#> 5 NSABHC0009 0.04356062 0.04587397 0.06008397 0.05540059 0.07216450      1
#> 6 NSABHC0009 0.03945594 0.03824322 0.04755034 0.04384791 0.05770193      1

3. Calculate spectral metrics

metrics <- calculate_spectral_metrics(pixel_values, 
                                      wavelengths = colnames(pixel_values[, 2:6]),
                                      masked = F,
                                      rarefaction = F)

head(metrics)
#>          site aoi_id        CV          SV          CHV image_type
#>        <char>  <num>     <num>       <num>        <num>     <char>
#> 1: NSABHC0009      1 0.3314709 0.001708346 3.302738e-10   unmasked

• co-efficient of variance (CV)
• spectral variance (SV)
• convex hull volume (CHV)

For a full discussion of these metrics see the manuscript

4. Download plot data from AusPlots. The veg.PI part extracts the point intercept data from the AusPlots data structure. In this case we use the same AusplotID as in the drone images above. This gets us data for two on-the-ground sampling visits to one particular AusPlot where we happen to have drone imagery.

my.data <- ausplotsR::get_ausplots(my.Plot_IDs=c("NSABHC0009"), veg.PI=TRUE)
my.data$site.info$visit_date
#> [1] "2012-09-03" "2016-09-13"

5. Calculate on-the-ground diversity from the point intercepts using different diversity metrics. The output is a list which includes taxonomic metrics, and also community matrices for checks. Not that this takes the PI part of the AusPlot object which stands for point-intercept. For a general function to calculate on-the-ground diversity, see the diversity function from the vegan package.

field_diversity <- calculate_field_diversity(my.data$veg.PI)
field_diversity$taxonomic_diversity
#>        site_unique site_location_name species_richness shannon_diversity
#> 1 NSABHC0009-58026         NSABHC0009               52          3.041263
#> 2 NSABHC0009-53604         NSABHC0009               38          2.833114
#>   simpson_diversity pielou_evenness exp_shannon inv_simpson
#> 1         0.9195892       0.7696978    20.93167    12.43614
#> 2         0.9039217       0.7788445    16.99831    10.40818

2012 was wetter than 2016 and there are a number of rain ephemeral species at this site so the higher species richness makes sense.

The calculated taxonomic diversity metrics are:

• species richness
• shannon’s diversity index
• simpson’s diversity index
• pielou’s evenness
• exponential shannon’s index
• inverse simpson’s index

6. Compare on-the-ground and from-the-sky diversity

High-resolution drone images are very large, and the example data included with the package on github are too small to do this meaningfully. We recommend doing this locally following the above steps.