Two kinds of number appear on this site. The models name the papers they implement and the archive they run against, and those are listed below. The crop reference is a table of planning ranges, and it does not yet record where its figures came from — which is stated here as a count rather than left to be discovered.
Counted across 538 rows and the 8 fields in each that carry a figure somebody could plant on. Computed on this request, not written down.
None of them, today. The reference describes itself as indicative planning ranges for Indian conditions, and that is all this site can honestly say about it: nobody recorded which package of practices, extension bulletin or trial each figure came from, and this page will not invent one. Every number on it still resolves to a row you can read, and the pages that run a model show their method.
To close the gap, register a source and map it to a field with the an_sx_sources and an_sx_crop_citations filters (inc/sources.php). A column can be cited at once, or a single crop where it differs. This count follows.
The daily soil water balance on /decide/: the Kc curve, root-zone depletion, readily available water and the stress coefficient. The coefficients used are FAO group defaults, and every run names the FAO crop they were measured on.
The yield fraction on /decide/. The run stops reporting one below ETa/ETc = 0.5, because the relation is fitted to moderate deficits and carrying it further turns a failed crop into a confident percentage.
The fallback reference evapotranspiration, used only where the weather archive has no computed ET0 for a day.
The rainfall and evaporation each run is measured against: the same calendar window in each of the last five years at the place named. The spread in a run is observed variance at that location, never a simulated distribution.
Listening…
It is Kharif — the monsoon crop. Answers are weighted for it.
Software, not a person. Nothing is recorded.