Skip to contents

Canonical schema for an `ossl_library` object

Source: R/spectra-ossl.R
ossl_library_template.Rd

predict_ossl_mbl and predict_ossl_plsr_local take an ossl_library argument that must be a list with two named elements:

Usage

ossl_library_template(
  wavelengths = 350:2500,
  properties = c("clay_pct", "sand_pct", "silt_pct", "cec_cmol", "bs_pct", "ph_h2o",
    "oc_pct", "fe_dcb_pct", "caco3_pct")
)

Arguments

wavelengths

Integer vector of wavelengths (default 350:2500 nm for Vis-NIR/SWIR).

properties

Character vector of property column names to seed the empty Yr data.frame with.

Value

A list with Xr (a 0-row matrix of the right column dimension) and Yr (an empty data.frame with the requested columns).

Details

  • Xr: numeric matrix, rows = OSSL training spectra, columns = wavelengths. Must align (after preprocessing) with the column space used by the spectra you predict on.

  • Yr: data.frame keyed by property name (e.g. clay_pct, cec_cmol), one row per training spectrum.

This function returns an empty template you can populate from a real OSSL extract (e.g. via the ossl-import Python package or the public S3 mirror at https://storage.googleapis.com/soilspec4gg-public/).

soilKey does not bundle OSSL data; until you populate this template with real values, all `predict_ossl_*` calls fall back to the deterministic synthetic predictor (which prints a warning).