Coming from dplR¶

dplPy is a Python companion to the R library dplR, and most workflows translate almost one-to-one. The biggest differences are ergonomic rather than conceptual:

Data structure. dplR uses a data.frame of class "rwl" (years as row names); dplPy uses a pandas DataFrame indexed by year. Same idea, same orientation (years down, series across).
Naming. R's dotted names become Python underscores (read.crn → read_crn, chron.ars → chron_ars).
Calling style. library(dplR); chron(rwi) becomes import dplpy as dpl; dpl.chron(rwi).
Return values. Some functions that print in R return a dict or DataFrame in dplPy (for example xdate returns a dict of segment correlations and flags).

Function map¶

Reading & writing¶

dplR	dplPy	Notes
`read.rwl`, `read.tucson`, `read.compact`	`readers`	Auto-detects format; `readers_url` for URLs
`read.crn`	`read_crn`	Single, stacked, and combined multi-site files
`read.ids`	`read_ids`	Site–tree–core parsing
`write.rwl` / `write.tucson`	`writers(..., format="rwl")`	Returns the path written
`write.crn`	`writers(..., format="crn")`
`combine.rwl`	`combine_rwl`
`read.tridas` / `write.tridas`	`read_tridas` / `write_tridas`, or `writers(..., format="tridas")`	TRiDaS 1.2.2 XML

Descriptive statistics¶

dplR	dplPy	Notes
`rwl.stats`	`stats`	Per-series mean, stdev, skew, kurtosis, gini, AR1, …
`rwl.report`	`report`
`rwl.check`	`check`	Findings-as-data QC checker (dplR 1.8.0 `rwl.check`)
`summary`	`summary`
`sens1`, `sens2`	`sens1`, `sens2`	Mean sensitivity (kept out of `stats`, as in dplR)
`gini.coef`	(column in `stats`)	Reported in the `stats` table

Detrending & standardization¶

dplR	dplPy	Notes
`detrend`, `detrend.series`	`detrend`	`fit=` selects Spline / ModNegExp / ModHugershoff / Mean / AgeDepSpline; `spline_period=` is dplR's `nyrs`; `plot=False` by default, like `make.plot=FALSE` for a data frame
`rcs`	`rcs`	Curve via `fit=`, arithmetic via `method=`, stiffness via `spline_period=`; `preset="crust"` for the CRUST curve. Pith offset differs — see below.
`ssf`	`ssf`	Signal-free; `preset="crust"` variant
`ads` (age-dependent spline)	`ads`	Also `detrend(fit="AgeDepSpline")`
`powt`	`powt`	Power transformation

Chronology¶

dplR	dplPy	Notes
`chron`	`chron`	`prewhiten=True` adds the residual chronology
`chron.ci`	`chron(ci=True)`	Bootstrap confidence interval on the standard chronology (adds `lower_ci`/`upper_ci`)
`chron.ars`	`chron_ars`	ARSTAN-style
`chron.stabilized`	`chron_stabilized`	Variance stabilization; `winLength` is `window_length`
`ar` / AR modelling	`autoreg`, `ar_func`	Yule-Walker by default (`method="yw"`), as in R's `ar()`, with `ar_max=10`. Order selection differs: dplPy takes the first local AIC minimum (`first_aic_min=True`, the ARSTAN/COFECHA rule); R's `ar()` takes the global minimum. Pass `first_aic_min=False` to match R.

Crossdating¶

dplR	dplPy	Notes
`corr.rwl.seg`	`xdate`	dplR-faithful default; returns a dict. `seg.length` is `window_length`, `lag.max` is `lag`
`corr.series.seg`	`series_corr`	Single-series diagnostics
`ccf.series.rwl`	`series_corr`	Cross-correlation view of one series
`interseries.cor`	`interseries_corr`	Overall interseries correlation; returns `{"mean", "table"}`
`xdate.floater`	`xdate_floater`	Extended with Wilson (2026) t-statistics
(COFECHA UFLOAT)	`xdate_undated`	Relative dating of a set of undated series against one another
`xdate.report`	`xdate_report`	COFECHA-style batch report (`xdate.report` arrived in dplR 1.8.0)
(COFECHA itself has no dplR equivalent)	`xdate(preset="COFECHA")`	Full COFECHA emulation mode

Editing series¶

dplR	dplPy	Notes
`insert.ring`, `delete.ring`	`edit_rwl`	One function inserts, deletes, or changes a ring; it pairs with the crossdating checks

Signal strength, agreement & indices¶

dplR	dplPy	Notes
`rwi.stats`, `rwi.stats.running`	`rwi_stats`, `rwi_stats_running`
`sss`	`sss`	Subsample signal strength
`glk`, `sgc`	`glk`, `sgc`	Gleichläufigkeit / synchronous growth changes
`treeMean`	`tree_mean`	Average cores to tree level
`common.interval`	`common_interval`
`bai.out`, `bai.in`	`bai_out`, `bai_in`	Basal area increment
`po.to.wc`, `wc.to.po`	`po_to_wc`, `wc_to_po`	Pith-offset conversion. dplPy's `pith_offset` is a true offset (`0` = pith present); dplR's is the cambial age of the first ring (`1` = pith present), so subtract 1 from a dplR `po` table.
`fill.internal.NA`	`fill_internal`	Interior-gap infilling

Plotting¶

dplR	dplPy	Notes
`plot.rwl`, `plot.crn`, `spag.plot`, `seg.plot`	`plot`	`type=` selects the plot style
crossdating plots	`xdate(plot=True)`, `xdate_plot`, `series_corr(plot=True)`

Beyond dplR¶

dplPy adds LiPD interchange, which dplR does not provide:

dplPy	Notes
`to_lipd`, `from_lipd`, `lipd_metadata`	Requires the `[lipd]` extra

Not yet ported¶

dplPy targets the core standardization, chronology, and crossdating workflow. A few dplR capabilities are intentionally out of scope: superposed epoch analysis and the spectral and wavelet methods are better served by dedicated Python time-series packages such as Pyleoclim, so dplPy does not duplicate them. Skeleton-plot crossdating and interactive (point-and-click) detrending are not currently available. Of dplR's tabular I/O, read.sheet / write.sheet (tidy/long and delimited spreadsheet formats) are not yet ported. Check the API Reference for the current surface, and the project repository to request or contribute a port.

A note on fidelity¶

Where dplPy claims to reproduce a dplR (or COFECHA, or ARSTAN) method, the behavior is checked against those reference implementations, and deliberate departures are documented in the relevant function's docstring. The clearest example is crossdating: xdate's A flag reproduces dplR's corr.rwl.seg exactly. dplR 1.8.0 added a lag search of its own (corr.rwl.seg(lag.max=)) but reports it separately and still flags on significance alone; dplPy's B (lag-shift) flag folds that lag result into an explicit COFECHA-style flag.