Classify many profiles from one horizon table

Builds one PedonRecord per profile from a table with one row per horizon, and classifies each with classify_all. The column named by id says which profile a horizon belongs to; the other columns use soilKey's names (horizon_attribute_spec), or are renamed and converted by map. Site properties (site_attribute_spec), such as soil_moisture_regime, come from sites.

Usage

classify_profiles(
  horizons,
  sites = NULL,
  id = "profile_id",
  map = NULL,
  systems = "all",
  on_missing = "silent",
  ...
)

Arguments

horizons

A data.frame with one row per horizon and the column named by id.

sites

Optional site properties: a data.frame with one row per profile and the same id column, or a list of lists named by profile id. A missing value (NA) is read as not recorded. Profiles without an entry get only their id.

id

Name of the column that identifies the profile, in horizons and in a sites table (default "profile_id").

map

Optional data.frame with the columns from and to, and optionally factor: the column from of horizons becomes the column to, multiplied by factor (default 1) to convert its unit. For example, organic carbon in g/kg: from = "ORGC", to = "oc_pct", factor = 0.1.

systems, on_missing

Forwarded to classify_all; the default on_missing = "silent" keeps a batch quiet, and the missing column reports what the keys did not find.

...

Other arguments forwarded to classify_all, such as include_family.

Value

A data.frame with one row per profile, in the order of their first horizon in horizons: the id column, one column per system with the name of the class (as in the summary of classify_all), missing (the attributes the keys asked for and did not find, comma-separated) and error (NA, or why the profile could not be classified). The attributes "results" and "pedons" hold the full classify_all results and the pedons, in lists named by profile id.

Details

The names are checked once for the whole table: a horizon column or site property that no key reads is named in one warning (see unrecognised_attributes), not once per profile. Within a profile the horizons are put in order of top_cm. A profile that cannot be built or classified gets its reason in the error column, and the others are classified as usual.

See also

classify_all for one pedon, classify_csv for one profile in a spreadsheet, unrecognised_attributes to check the names first.

Examples

h <- rbind(
  cbind(profile_id = "p1", as.data.frame(make_ferralsol_canonical()$horizons)),
  cbind(profile_id = "p2", as.data.frame(make_phaeozem_canonical()$horizons)))
s <- data.frame(profile_id = c("p1", "p2"),
                soil_moisture_regime = c("udic", "ustic"))
res <- classify_profiles(h, sites = s, systems = "usda")
res[, c("profile_id", "usda", "error")]
#>   profile_id              usda error
#> 1         p1 Rhodic Haploperox  <NA>
#> 2         p2 Typic Haplustolls  <NA>

# \donttest{
# Columns named and measured differently: rename and convert with `map`.
h2 <- h
names(h2)[names(h2) == "oc_pct"] <- "ORGC"
h2$ORGC <- h2$ORGC * 10                      # g/kg
m <- data.frame(from = "ORGC", to = "oc_pct", factor = 0.1)
classify_profiles(h2, sites = s, map = m)[, c("profile_id", "wrb", "sibcs", "usda")]
#>   profile_id                                                  wrb
#> 1         p1 Geric Rhodic Ferralsol (Clayic, Epic, Eutric, Humic)
#> 2         p2             Chernic Phaeozem (Loamic, Humic, Pachic)
#>                                                                    sibcs
#> 1 Latossolos Vermelhos Distróficos típicos, textura argilosa, A moderado
#> 2                   Chernossolos Háplicos Órticos típicos, textura média
#>                usda
#> 1 Rhodic Haploperox
#> 2 Typic Haplustolls
# }