Check where the arable-land extension changes composition mid-panel.

Report, per term and area_code, whether a term of the fallow-inclusive arable land extension is present for part of a country's panel and absent for the rest. Two terms switch off inside the published panel and neither switch is a land-use change:

fodder — the FAOSTAT fodder items (Cat_1 == "Fodder_green" and not grass, i.e. CBS 2000-2003). FAOSTAT's fodder tonnage (faostat-production-old, production only) ends in 2013 and EU AgriDB, the only other source, ends in 2019, so a build reaching 2020 has no fodder at all from that year on and every ordinary arable crop silently absorbs fodder's share of the country's arable land (whep#938).
temp_grassland_netting — the temporary grassland netted out of the arable target. Modelled CBS 3002 comes from EU AgriDB alone, so it exists for 26 EU polities and stops at 2019; from 2020 the netting term is identically zero while FAO's arable land still contains temporary meadows, and the reconciliation changes method at the boundary (whep#937).

This is a diagnostic, not a correction: it flags the discontinuity so a series is not read across it. The treatments live behind build_fao_arable_fallow_extension()'s fodder_gap and temp_grassland_basis arguments, whose defaults reproduce the published behaviour.

Usage

check_arable_composition(extension, items_prod_full = whep::items_prod_full)

Arguments

extension: Tibble of the arable/permanent land extension as returned by build_fao_arable_fallow_extension(): year, area_code, item_cbs_code, impact_u, and optionally temp_grassland_source or temp_grassland_netted_ha. The temp_grassland_netting term is reported only when one of them is present. With temp_grassland_source a country-year counts as covered whenever a source supplied the term, zero included; with temp_grassland_netted_ha alone, only when it is positive.
items_prod_full: Crosswalk used to classify item_cbs_code as perennial via Herb_Woody. Defaults to items_prod_full.

Value

A tibble with one row per (term, area_code):

term: "fodder" or "temp_grassland_netting".
area_code: the country.
panel_first_year, panel_last_year: the years that country has arable rows for.
term_first_year, term_last_year: the years the term is present, as defined under extension (NA when it never is).
n_years_absent: panel years in which the term is absent.
break_year: the first panel year after term_last_year with no term (NA when the term runs to the end of the panel, or never appears).
broken: TRUE when the term is present in some panel year and absent in a later one.
never_present: TRUE when the term is absent for the whole panel.

Details

check_series_jumps() does not find either break by default. A term does not fall to a small value at the boundary, it stops having rows, and its min_value guard skips any pair involving a zero. With dropouts = TRUE it completes each series with zero and flags the stop, so check_series_jumps(extension, impact_u, .by = c("area_code", "item_cbs_code"), dropouts = TRUE) reports each fodder series at its break year. This function answers the coverage question per term and area instead, and also covers the netting term, which is not an item series.

Examples

extension <- tibble::tribble(
  ~year, ~area_code, ~item_cbs_code, ~impact_u, ~temp_grassland_netted_ha,
  2019L, 10L, 2003L, 100, 50, # fodder mix, netting fires
  2019L, 10L, 2511L, 900, 50,
  2020L, 10L, 2511L, 1000, 0 # fodder gone, netting off
)
check_arable_composition(extension)
#> # A tibble: 2 × 10
#>   term                area_code panel_first_year panel_last_year term_first_year
#>   <chr>                   <int>            <int>           <int>           <int>
#> 1 fodder                     10             2019            2020            2019
#> 2 temp_grassland_net…        10             2019            2020            2019
#> # ℹ 5 more variables: term_last_year <int>, n_years_absent <int>,
#> #   break_year <int>, broken <lgl>, never_present <lgl>