Above-ground Biomass Raster Pre-Processing

The goal of this repository is to do some light pre-processing to above-ground biomass data products and save them to the “snow” server in a common format.

Pre-processing steps currently taken:

• Name all layers appropriately
  • If they are years, use the year (e.g. "2005")
  • If they represent a time range, use the time range (e.g. "2005-2010 mean" or "2009, 2012-2014 mean")
• Convert to units of Mg/ha
• Rasterize vector products
• Save as cloud optimized geotiff(s)

Other possible steps that could be added in the future:

• Re-project to a common CRS
• Include layers besides AGB. For example, some products have layers with standard deviation or quality control flags that are currently removed during this pre-processing.

Example Function

In R/ there is a collection of clean_*() functions defined. One of the simpler ones is for the Xu et al. dataset:

# Function has arguments for input and output path.  Input might be a directory in the case of multiple tiles
clean_xu <- function(
    input = "/Volumes/moore/Xu/test10a_cd_ab_pred_corr_2000_2019_v2.tif", 
    output = "/Volumes/moore/AGB_cleaned/xu/xu.tif"
) {
  # Read in file(s)
  xu_agb_raw <- terra::rast(input) 
  # Do any necessary conversions or subsetting
  xu_agb <- xu_agb_raw * 2.2 
  # Add metadata including layer names
  units(xu_agb) <- "Mg/ha"
  varnames(xu_agb) <- "AGB"
  names(xu_agb) <- 2000:2019
  
  # Write to COG
  terra::writeRaster(xu_agb, output, filetype = "COG", overwrite = TRUE)
  # Return output path
  output
}

If you have snow mounted to /Volumes/moore/ then all you need to do is run clean_xu() to read in the downloaded raw data, do the processing, and save it out as a COG to AGB_cleaned/xu.tif.

Note

This will actually write two files, xu.tiff and a “sidecar” file xu.tiff.aux.json containing some metadata (just the units in this example). terra::rast("path/to/xu.tiff") will automatically also read in the sidecar file if it is in the same directory.

Using targets

These functions are set up to all be run as part of a targets pipeline using just the command tar_make(). The pipeline (defined in _targets.R) tracks both input and output files as well as the code in the clean_*() functions so if there are any changes (e.g. adding a new tile for the ESA CCI dataset), running tar_make() should automatically update the output data in the AGB_cleaned/ folder on snow.

Setup with renv

The R package dependencies for this repository are tracked with renv. When you first open this project, you’ll be prompted to install the dependencies with renv::restore(). This will attempt to install the exact versions of all packages used. Sometimes, especially when a package needs to be built from source, this will fail. You have the option of turning off renv with renv::deactivate(clean = TRUE), installing the most recent versions of required R packages (the ones in _targets_packages.R, and then re-initializing renv with renv::init(). However, note that changes in R packages could require you to update code in the cleaning functions or in _targets.R.

---

Developed in collaboration with the University of Arizona CCT Data Science team