Table recipes

A DATABASE block names the tables a run may write, and naming one is never enough on its own. The option is accepted whether or not the run does what the writer reports on, and the table is then simply absent: the projection succeeds, the database is written, and nothing says the table is missing. This page gives the second thing each table needs.

Every recipe here has been run, and every table in the Checked column has been compared against a standalone FVSbc built from the same source and found to match value for value.

The keyword set that produces each table is in the repository note notes/reference/oracle-coverage.md; this page is the contract, and that note is the evidence.

Always

These need only their option, plus the keyword where one is named.

Table	Option	Also needs
`FVS_Summary_Metric`	`Summary`
`FVS_Compute`	`ComputDB`	a `COMPUTE` block, or there is nothing to write
`FVS_TreeList_Metric`	`TREELIDB`	a `TREELIST` keyword
`FVS_StrClass_Metric`	`STRCLSDB`	a `STRCLASS` keyword
`FVS_Stats_Species_Metric`	`INVSTATS`	a `STATS` keyword
`FVS_Stats_Stand`	`INVSTATS`	a `STATS` keyword and ``DESIGN``’s fourth field above 1: it is the distribution among sample points, and a one-plot stand has none

The fire-model block

Put ``FMIN`` before ``INVYEAR``. FUELOUT and POTFIRE capture the inventory year when they are read, and FVS treats a positive value as “suppress this report” — read after INVYEAR they switch their own reports off and the tables are silently absent. The validator warns.

Table	Option	Also needs
`FVS_Fuels_Metric`	`FuelsOut`	`FUELOUT`
`FVS_PotFire_Metric`, `FVS_PotFire_Cond_Metric`	`PotfirDB`	`POTFIRE`
`FVS_Consumption_Metric`	`FuelReDB`	`FUELREPT` and a fire
`FVS_SnagDet_Metric`, `FVS_SnagSum_Metric`	`SnagOuDB`, `SnagSuDB`	`SNAGOUT`, `SNAGSUM`
`FVS_Down_Wood_Vol_Metric`, `FVS_Down_Wood_Cov_Metric`	`DWDVLDB`, `DWDCVDB`	`DWDVLOUT`, `DWDCVOUT`
`FVS_Carbon_Metric`, `FVS_Hrv_Carbon`	`CarbReDB`	`CARBREPT`
`FVS_Mortality_Metric`	`MORTREDB`	`MORTREPT` and a fire: the writer is called from `fmburn.f`
`FVS_CanProfile`	—	`CANFPROF`

SIMFIRE schedules a fire: SIMFIRE=1,20,1,21,1,100,1 is cycle 1, wind 20 km/h, moisture condition 1, 21 °C, FFE mortality, the whole stand, early spring. Without one, the fire tables report the initial state and the mortality table never appears.

The event tables

Table	Option	Also needs
`FVS_CutList_Metric`	`CUTLIDB`	a `CUTLIST` keyword and a removal — the writer is called from `cuts.f`
`FVS_ATRTList_Metric`	`ATRTLIDB`	an `ATRTLIST` keyword and a removal — it is the after-treatment listing
`FVS_Regen_Tally_Metric`, `FVS_Regen_SitePrep`	`REGREPTS`	an `ESTAB` block with a `PLANT` or `NATURAL` line carrying stems per acre; a line with zero is refused and takes the whole keyword set with it

The blocks

RDIN, MISTOE and ESTAB are blocks of keywords, written by the same writer as everything else and checked against their readers’ own tables. A hand-spaced line is read as different numbers with no error, which is why they are not lines any more.

Table	Block	Option
`FVS_RD_Det_Metric`, `FVS_RD_Sum_Metric`	`RDIN` with `RRINIT` and `RRDOUT`	`RDDETAIL`, `RDSUM`
`FVS_DM_Spp_Sum_Metric`, `FVS_DM_Stnd_Sum_Metric`, `FVS_DM_Sz_Sum_Metric`, `FVS_DM_TreeList_Metric`	`MISTOE` with `MISTPINF`, `MISTPRT` and `NEWSPRED`	`MISRPTS`

The dwarf-mistletoe tables also need an infection. The rating enters through the damage columns — code 30 with the Hawksworth rating as the severity — and NEWSPRED is what makes the per-tree table appear. Without it, three of the four are written and the fourth is not.

`FVS_InvReference`

It needs no keyword at all and is written on every run: species limits, site indices and volume equations, one row per species. It is produced by DBSREFERENCE, which fvs.f calls immediately after STATS.

Before `0.18.0` no run of this package produced it. The extension is driven by PyFVS’s src/api/fvs_step.f90, an extraction of fvs.f’s initialisation that calls STATS and omitted DBSREFERENCE; the original fvs_ is dead code in the extension, so the writer was compiled in and never called. The build recipe’s PyFVS patch set now carries the missing call.

The two that are not recipes

FVS_Error is written only when FVS logs something. A run that logs nothing has no such table, and that is a clean run.

Note

FVS_Regen_SitePrep’s PcntNone, PcntMech and PcntBurn cannot be relied on. FVS decides whether to sample site preparation by branching on a local, LOAD, that it only assigns when the stand is more than twenty years past inventory or a site-preparation keyword is present — so an ordinary run reads an uninitialised value. Two builds of the same source disagree. Every other table in the same run is unaffected; notes/triage.md carries the evidence.