
Load the bundled WoSIS stratified RSG-balanced sample (v0.9.73)
Source:R/wosis-cache.R
load_wosis_stratified_sample.RdReturns a 130-profile snapshot of WoSIS GraphQL data pulled on 2026-05-09 with **stratified sampling by WRB Reference Soil Group**: 5 profiles per RSG across 26 RSGs (Acrisol, Andosol, Arenosol, Calcisol, Cambisol, Chernozem, Cryosol, Ferralsol, Fluvisol, Gleysol, Gypsisol, Histosol, Kastanozem, Leptosol, Luvisol, Nitisol, Phaeozem, Planosol, Plinthosol, Podzol, Regosol, Solonchak, Solonetz, Stagnosol, Umbrisol, Vertisol).
Value
A list with:
pedons: list of 130PedonRecordobjects.meta: named integer vector of profiles per RSG.pulled_on: pull date.endpoint: ISRIC GraphQL endpoint URL.filter: pull strategy metadata.n_pulled: 130.
Details
This is the recommended cache for global WRB benchmarking. Compared
to load_wosis_sample() (40 SA-only profiles, mostly Solonetz
and Phaeozem from Argentina), the stratified sample provides:
Even coverage across the 26 most important RSGs.
Richer analytical attributes – CEC available on 26 ECEC on 37 in the SA snapshot).
Geographic diversity (Angola, Brazil, USA, China, Russia, South Africa, Indonesia, Argentina, etc.).
First-ever benchmark on this sample (soilKey v0.9.73, full v0.9.69-72 fallback stack):
Overall top-1 accuracy: 16.2\
Histosol 100\ from 20\ Cambisol 60\
18 RSGs at 0\ expose (Munsell colours, base saturation, sodium for Solonetz, slickensides for Vertisols, etc.). Documented data ceiling.
For the live API, call run_wosis_benchmark_graphql() or
the read_wosis_profiles_graphql(wrb_rsg = "...", n_max = N)
helper (small RSG-filtered queries are tractable; large unfiltered
pulls time out as of 2026-05).
Reference
Batjes, N. H., Ribeiro, E., van Oostrum, A. (2020). Standardised soil profile data to support global mapping and modelling (WoSIS snapshot 2019). Earth System Science Data, 12, 299-320. doi:10.5194/essd-12-299-2020 .