Create derived variables

Crease derived variables such as NDVI and NDRE, upscaled images, and canopy height estimates.

Usage

derive(
  site,
  pattern1 = "",
  pattern2 = NULL,
  metrics = c("NDVI", "NDWIg", "NDRE"),
  window = 3,
  resources = NULL,
  local = FALSE,
  trap = TRUE,
  comment = NULL
)

Arguments

site

One or more site names, using 3 letter abbreviation. Use all to process all sites. In batch mode, each named site will be run in a separate job.

pattern1

File names, portable names, regex matching either, or search names selecting source for derived variables. See Image naming in README for details. See details.

pattern2

A second pattern or vector of layer names, used for bivariate metrics. See details.

metrics

A list of metrics to apply. Univariate metrics include:

NDVI

Normalized difference vegetation index, (NIR - red) / (NIR + red), an index of biomass

NDWIg

Normalized difference water index (green, commonly known as McFeeter's NDWI), (green - NIR) / (green + NIR), primarily helps distinguish waterbodies

NDRE

Normalized difference red edge index, (NIR - RE) / (NIR + RE), an index of the amount of chlorophyll in a plant

mean

mean of each band in a window, size defined by window

std

standard deviation of each band in a window, size defined by window

NDVImean

mean of NDVI in a window, size defined by window

NDVIstd

standard deviation of NDVI in a window, size defined by window

Bivariate metrics include:

NDWIswir

Normalized difference water index (SWIR, commonly known as Gao's NDWI), (NIR - SWIR) / (NIR + SWIR), an index of water content in leaves; requires a Mica layer for pattern1, and a matched SWIR layer for pattern2

delta

The difference between pattern1 and pattern2, may be useful for taking a difference between late-season and early-season DEMs to represent vegetation canopy height

window

Window size for mean, std, NDVImean, and NDVIstd, in cells; windows are square, so just specify a single number. Bonus points if you remember to make it odd.

resources

Slurm launch resources. See launch. These take priority #' over the function's defaults.

local

If TRUE, run locally; otherwise, spawn a batch run on Unity

trap

If TRUE, trap errors in local mode; if FALSE, use normal R error handling. Use this for debugging. If you get unrecovered errors, the job won't be added to the jobs database. Has no effect if local = FALSE.

comment

Optional slurmcollie comment

Details

This function creates any of a number of univariate and bivariate metrics derived from raster data, such as NDVI and NDRE. Results are written as rasters to the flights directory, with the metric included in the name. These metrics will be treated like source layers in subsequent processing and modeling.

For univariate metrics, supply one or more layer names via pattern1. All metrics will be calculated for each layer specified by pattern1. Results will be named <layer>__<metric>.

For bivariate metrics, specify matched pairs of layers with pattern1 and pattern2. It's best to specify complete names (you can use vectors for each) so the layers are paired properly. If you're crazy enough to use regular expressions here, scrutinize the result names carefully. Results will be named <layer1>__<layer2>__metric. At the moment, NDWIswir and delta are the only bivariate metrics.

Note that all normalized difference (NDxx) metrics require five-band Mica data.

Note that derived metrics get two underscores in their names, e.g., <layer>__NDVI. This is used to distinguish primary from derived data.

This fits in the workflow after gather and before sample.