NEWS

---

forestTIME 2.2.0.9000

• Now uses the annual inventory start years from Appendix J of the FIADB user guide to filter both raw data (in fia_tidy()) and EVALIDs (in fia_assign_strata()) to include only annual inventory years and exclude earlier periodic inventories.
• Fix to internal function inter_extra_polate() which was not handling leading NAs correctly in linear interpolation (#225).
• EVALIDs specific to East or West Texas won't be matched to plots by fia_assign_strata().
• Early EVALIDs in NM and WY that "don't work" no longer assigned to plots by fia_assign_strata().
• Changed behavior of fia_assign_strata() so that for each plot x year, the first EVALID that contains the YEAR (i.e. between START_INVYR and END_INVYR) is matched.
• Changed package license to MIT.
• Fixed a bug where for woodland species, carbon was being replaced with biomass.
• As part of the workaround for woodland species (#163), any negative interpolated carbon and biomass values were set to 0.

forestTIME 2.2.0

• The internal function fia_eval_info() was added.
• fia_assign_strata() was added to match plots in each year of the annualized data to an EVALID, estimation unit, and stratum along with associated values necessary for population level estimation with variance calculations.
• fia_tidy() now adds several columns from the POP_* tables relating to EVALIDs. Each plot in each INVYR is matched to the EVALID of each type that has the latest end year. Then, for EVALIDs ending in 01 and 02 ("EXPCURR" and "EXPVOL", respectively), two columns are created for each of EVALID, ESTN_UNIT, STRATUMCD, P1POINTCNT, P1PNTCNT_EU, and AREA_USED suffixed with _EXPVOL or _EXPCURR. There are also two logical columns EVAL_TYPE_EXPCURR and EVAL_TYPE_EXPVOL. These get filled down and switch at the next inventory (rather than at the midpoint like everything else) (#192).
• fia_tidy() no longer filters to base intensity plots and reverts to just filtering to INVYR≥2000.
• Changeed the way composit IDs plot_ID and tree_ID are constructed by changing the order of STATECD and UNITCD so that now it is UNITCD_STATECD_COUNTYCD_PLOT (#189)
• Fixed a bug in fia_tidy() resulting in duplicated SUBCYCLE columns (#173).
• A temporary workaround was implmented to deal with missing code for calculating biomass and carbon for woodland species (#163). Eventually, we will modify fia_estimate() to be able to produce carbon and biomass estimates for woodland species and this can be reverted to avoid confusion.
  • fia_tidy() now keeps the CARBON_AG and DRYBIO_AG columns
  • interpolate_data() (and therefore fia_annualize()) now linearly interpolates CARBON_AG and DRYBIO_AG
  • fia_estimate() now overwrites CARBON_AG and DRYBIO_AG columns in the input unless the result of carbon estimation is NA (as is the case with woodland species).
• The calculation of the EXPNS column no longer occurs as part of interpolate_data().

forestTIME 2.1.0

• This package has been re-named from forestTIME.builder to forestTIME
• fia_tidy() now only keeps base-intensity plots (INTENSITY == 1 & SUBCYCLE != 99 & SUBCYCLE != 0).
• Fixed a bug that was causing TPA_UNADJ to not be populated for all trees in states that use macroplots (#160).
• Fixed a bug in fia_split_composite_ids() that caused it to fail when tree_ID was NA (as it is in conditions with no observations). It now falls back on the information in plot_ID when tree_ID is in the data frame but NA (#149).
• fia_estimate() now returns the additional variables DRIBIO_FOLIAGE, VOLTSGRS, and VOLTSSND in addition to DRYBIO_AG and CARBON_AG.
• Code to deal with negative extrapolated values has moved to adjust_mortality(). Therefore, the results of interpolate_data() may now contain negative numbers, which are non-sensible. Use fia_annualize() whenever possible to ensure sensible results.
• fia_download() arguments have changed. keep_zip now defaults to TRUE. extract options have changed from TRUE/FALSE to "forestTIME", "rFIA", "all", or "none" to extract just files needed by forestTIME, those used by rFIA (for compatibility), all files, or none.
• Fixed a bug in interpolation of CONDPROP_UNAJ. Now the workflow retains all "empty" conditions (i.e. CONDIDs with no trees in them) and properly interpolates CONDPROP_UNADJ so the proportion for all conditions in a plot in a year sum to 1 (within rounding error) (#64).
• fia_annualize() now adds and EXPNS column calculated as the total land area of the state in acres divided by the number of plots in the interpolated data. It should be usable in the same ways the EXPNS column in the "raw" FIA data can be used.
• Renames the state_codes dataset to state_areas and adds a column for state land area in acres.

forestTIME 2.0.0

• The separate prep_carbon() and estimate_carbon() functions are no longer exported and are replaced by the combined fia_estimate() function.
• Added fia_annualize() which is a wrapper for df |> expand_data() |> interpolate_data |> adjust_mortality() and prefered over running each step separately as the individual steps contain artifacts of the annualization process.
• Renamed split_composite_ids() to fia_split_composite_ids()
• Renamed add_composite_ids() to fia_add_composite_ids()
• Renamed prep_data() to fia_tidy()
• Renamed read_fia() to fia_load()
• Renamed get_fia_tables() to fia_download()

forestTIME 1.1.0

• Fixed a bug causing the INTENSITY column (and possibly other plot-level variables) to be filled incorrectly by expand_data() (#122, reported by @brian-f-walters-usfs)
• expand_data() now converts NAs for CULL to 0s (this is what the carbon estimation code in predictCRM2() does already anyways) so that they are better interpolated. CULL values are converted back to NA if DIA is < 5 after interpolation by interpolate_data() (#77).
• prep_data() no longer filters out any rows (un-doing #59 and addressing #99). If you want to remove certain rows, do this between prep_data() and expand_data().
• Trees with only a single measurement have their single measurement carried forward during extrapolation rather than getting dropped from the data (#94, #99).
• In the case when MORTYR is an inventory year where the tree is alive (STATUSCD 1), it is now assumed by adjust_mortality() that the tree died in the year following MORTYR in order to keep the observation (#61).
• interpolate_data() no longer produces negative values for HT, DIA, or ACTUALHT. Instead, trees that get extrapolated to have DIA < 1 or HT or ACTUALHT < 4.5 (or < 1 for woodland species) are assumed to be fallen and dead (STATUSCD 2 and STANDING_DEAD_CD 0). These fallen dead trees then have their measurements set to NA by adjust_mortality(). Therefore, prep_carbon() no longer filters out trees with negative values for HT. (Fixes #60).
• adjust_mortality() now assures trees in non-sampled conditions (COND_STATUS_CD != 1) don't have interpolated values.
• estimate_carbon() no longer modifies any columns and no longer filters out any rows. It only adds columns for biomass and carbon estimates (which may be NA if they couldn't be estimated) (finally fixes #63)
• expand_data() now fills MORTYR so it is constant for a particular tree. NOTE this is different from how this column is populated in the raw data.
• Fixed a bug in adjust_mortality() that was causing trees that go from STATUSCD 2 to STATUSCD 0 (move to non-sampled area) to inapropriately have extrapolated values (#100 reported by @dnsteinberg)
• Fixed a bug in expand_data() that was caused STANDING_DEAD_CD and DECAYCD to not be interpolated correctly, resulting in extrapolated measurments for fallen dead trees (#101 reported by @dnsteinberg)
• prep_data() now converts PLT_CN from numeric to character for better readability in the output.
• Empty plots are no longer dropped silently by prep_data() and should be handled correctly by the rest of the workflow through interpolate_data().
• expand_data() now adds a column, interpolated, that marks whether an observation was interpolated (TRUE) or in the original data (FALSE).
• Trees that have always been fallen and have no measurements are now removed by prep_data()
• Trees that change species (more than one SPCD value) are assumed to have always been their last recorded species. prep_data() now overwrites SPCD with the last recorded SPCD for each tree.
• Additional columns PLT_CN, COND_STATUS_CD are kept for the interpolated data.
• Added a vignette (WIP) on how to use outputs of forestTIME.builder to get population level estimates.
• forestTIME.builder is now an R package
• Added functions add_composite_ids() and split_composite_ids() to deal with the composite ID columns tree_ID and plot_ID. This should make it easier to join to other FIA tables.

forestTIME 1.0.0

• Refactored to not use databases and instead to produce a single table of interpolated data
• Added new "main" functions get_fia_tables(), read_fia(), prep_data(), expand_data(), interpolate_data(), adjust_mortality(), prep_carbon(), and estimate_carbon()
• Outlined the process of creating annualized data in docs/pop_scaling.qmd
• Added "null and length 0 coalescing operator" %|||% to R/utils.R which is used to suppress warnings that come from adjusting for mortality when a tree hasn't yet died.
• fixed bug where TPA_UNADJ wasn't getting joined for trees with DIA between 4.9 and 5 (#68)
• fixed a bug where categorical vars weren't interpolated correctly when switching from NAs (#72)
• trees that have had RECONCILECD 7 ("Cruiser error") or 8 ("Procedural change") at any inventory are now removed in prep_data() (#59)
• interpolates trees with STATUSCD 0 and RECONCILECD 5, 6 or 9 at t2 to midpoint between t1 and t2 and then removes them from sample (#59)

forestTIME 0.1.0

• Initial release with "pre-carbon" code to annualize the tree table, but not estimate carbon or biomass