Reading tree-ring data with dplPy¶

This notebook provides a tour through the capababilities of dplpy.readers(), which reads Tucson format .rwl files (and .csv) into a Pandas DataFrame. This dataframe has one column for each measurement series, indexed by year. The full description of readers can be found in the documentation here: https://opendendro.github.io/dplPy/reference/io/#readers and the User Guide here: https://opendendro.github.io/dplPy/guide/reading-writing-data/#ring-width-files-with-readers.

This notebook uses the sample files shipped in ../tests/data/rwl/ and ../tests/data/csv/, so it runs as-is from the notebooks/ folder of the repository.

First, we'll import dplPy:

In [2]:

import dplpy as dpl
import matplotlib.pyplot as plt

dpl.__version__

import dplpy as dpl import matplotlib.pyplot as plt dpl.__version__

Out[2]:

'0.8.0'

In order to be able to use this notebook 'out of the box', set the paths to where the Tucson .rwl and the .csv files live. These files reflect a range of different data, including several with known formating errors used to test the robustness of readers.py.

In [3]:

# Sample data shipped with the repo, relative to the notebooks/ folder.
RWL = "../tests/data/rwl/"
CSV = "../tests/data/csv/"

# Sample data shipped with the repo, relative to the notebooks/ folder. RWL = "../tests/data/rwl/" CSV = "../tests/data/csv/"

The Basics¶

At its most simple, you can pass a filename (and path) and get back a year-indexed DataFrame with one column per series. We'll call dpl.stats as well, to give us a quick look at the data that is now in our workspace.

In [4]:

data = dpl.readers(RWL + "ca533.rwl")
dpl.stats(data)

data = dpl.readers(RWL + "ca533.rwl") dpl.stats(data)

ca533.rwl successfully extracted as rwl file with 34 series covering the period from 626 to 1983

Out[4]:

	series	first	last	year	mean	median	stdev	skew	kurtosis	gini	ar1
1	CAM011	1530	1983	454	0.440	0.40	0.222	1.029	1.102	0.273	0.696
2	CAM021	1433	1983	551	0.424	0.40	0.185	0.946	1.110	0.237	0.701
3	CAM031	1356	1983	628	0.349	0.29	0.214	0.690	-0.366	0.341	0.808
4	CAM032	1435	1983	549	0.293	0.26	0.163	0.717	0.159	0.309	0.661
5	CAM041	1683	1983	301	0.526	0.53	0.223	0.488	0.098	0.238	0.690
6	CAM042	1538	1983	446	0.439	0.36	0.348	3.678	17.713	0.324	0.880
7	CAM051	1247	1983	737	0.273	0.25	0.140	1.836	6.678	0.262	0.705
8	CAM061	1357	1983	627	0.462	0.47	0.202	-0.111	-0.223	0.247	0.509
9	CAM062	1525	1983	459	0.442	0.45	0.188	-0.266	-0.196	0.240	0.529
10	CAM071	1037	1983	947	0.249	0.25	0.109	0.027	0.022	0.247	0.577
11	CAM072	1114	1983	870	0.309	0.29	0.163	0.698	1.181	0.292	0.735
12	CAM081	1081	1983	903	0.327	0.31	0.124	0.555	0.445	0.211	0.716
13	CAM082	977	1983	1007	0.285	0.29	0.114	0.312	0.406	0.223	0.767
14	CAM091	1460	1983	524	0.532	0.52	0.255	0.425	0.396	0.267	0.628
15	CAM092	1591	1983	393	0.349	0.34	0.226	0.337	-0.534	0.369	0.558
16	CAM101	1727	1983	257	0.568	0.56	0.260	0.254	-0.352	0.259	0.703
17	CAM102	1665	1983	319	0.604	0.62	0.261	0.082	0.092	0.243	0.677
18	CAM111	1446	1983	538	0.625	0.62	0.249	0.196	-0.097	0.225	0.624
19	CAM112	1471	1983	513	0.570	0.56	0.211	0.223	0.183	0.207	0.575
20	CAM121	1000	1983	984	0.259	0.26	0.106	0.042	-0.113	0.231	0.593
21	CAM122	1000	1983	984	0.271	0.27	0.109	0.346	0.601	0.223	0.646
22	CAM131	695	1970	1276	0.552	0.53	0.198	0.330	-0.118	0.202	0.787
23	CAM132	710	1232	523	0.397	0.38	0.148	0.871	1.001	0.203	0.808
24	CAM141	1030	1970	941	0.627	0.60	0.204	0.695	0.998	0.177	0.746
25	CAM151	1222	1970	749	0.446	0.39	0.273	1.068	1.159	0.332	0.765
26	CAM152	1221	1449	229	0.534	0.52	0.195	0.297	0.433	0.203	0.688
27	CAM161	1106	1609	504	0.339	0.33	0.149	0.633	0.648	0.243	0.792
28	CAM162	971	1970	1000	0.397	0.37	0.184	0.647	0.068	0.259	0.840
29	CAM171	1213	1970	758	0.450	0.40	0.210	1.250	2.326	0.250	0.797
30	CAM172	1174	1970	797	0.482	0.42	0.249	1.622	4.033	0.268	0.847
31	CAM181	1190	1970	781	0.283	0.25	0.149	0.706	0.119	0.293	0.801
32	CAM191	1180	1970	791	0.366	0.25	0.336	2.359	7.202	0.429	0.876
33	CAM201	990	1582	593	0.474	0.47	0.181	0.772	2.232	0.208	0.709
34	CAM211	626	1968	1343	0.357	0.34	0.182	0.513	0.127	0.286	0.681

You should see a success message from dpl.readers, indicating the file type, number of series, and the full time span of the data.

As data is a Pandas DataFrame, we can append a .plot method to it and use plt. methods from Matplotlib to add axes and a title:

In [5]:

data.plot(legend=False, figsize=(9, 4))
plt.xlabel("Year")
plt.ylabel("ring width (mm)")
plt.title("ca533 — raw ring-width series")
plt.show()

data.plot(legend=False, figsize=(9, 4)) plt.xlabel("Year") plt.ylabel("ring width (mm)") plt.title("ca533 — raw ring-width series") plt.show()

We can see the raw ring widths in units of mm arrayed through time. We can also use append other methods from Pandas. For instance, we can get a quick sample depth plot by first counting the number of non-NaN values in each row using .count, then using .plot again to display these:

In [6]:

data.count(axis=1).plot(figsize=(9, 3))
plt.xlabel("Year")
plt.ylabel("number of series")
plt.title("ca533 — sample depth")
plt.show()

data.count(axis=1).plot(figsize=(9, 3)) plt.xlabel("Year") plt.ylabel("number of series") plt.title("ca533 — sample depth") plt.show()

Headers are detected automatically and read attributes are recorded¶

dplPy automatically identifies where the data rows actually start and records how many header lines it skipped onin df.attrs. The .attrs is once again a feature of Pandas (https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.attrs.html) that can store attributes for the dataset, so that this information is carried along throughout your script. readers uses .attrs to store anything unusual during the import process as well as metadata:

In [7]:

tx = dpl.readers(RWL + "tx042.rwl")
print("header lines skipped:", tx.attrs["dplpy_header_lines_skipped"])

tx = dpl.readers(RWL + "tx042.rwl") print("header lines skipped:", tx.attrs["dplpy_header_lines_skipped"])

tx042.rwl successfully extracted as rwl file with 95 series covering the period from 1473 to 1992 (auto-detected 3 header lines)
header lines skipped: 3

Metadata in the header is identified and saved¶

The site and sample metadata is parsed as best as possible from the header by dpl.readers and is saved into df.attrs["dplpy_metadata"]. Alternatively we provide dpl.metadata() which allows you to simply extract the metadata into a Python dict structure:

In [8]:

meta = dpl.metadata(RWL + "tx042.rwl")
meta

meta = dpl.metadata(RWL + "tx042.rwl") meta

Out[8]:

{'site_id': 'BSC',
 'site_name': 'Big Bend National Park incl. Camp Springs',
 'species_code': 'PSME',
 'species_name': 'Douglas Fir',
 'country_region': 'Texas',
 'elevation_m': 2057,
 'latitude': 29.25,
 'longitude': -103.3,
 'first_year': 1473,
 'last_year': 1992,
 'investigators': 'Brewster  E.Cook  N.Montagu',
 'n_header_lines': 3,
 'header_raw': ['BSC    1 Big Bend National Park incl. Camp Springs           PSME               ',
  'BSC    2 Texas        Douglas Fir       2057M  2915-10318    __    1473 1992    ',
  'BSC    3 Brewster  E.Cook  N.Montagu                                            '],
 'hemisphere_verified': True}

Values in meta (or any dict) can then be extracted or printed to the screen (or added to a plot) by using square brackets and the name of the field (called the 'key' in Python) in quotation marks:

In [9]:

print(meta["site_name"], "|", meta["species_code"], "-", meta["species_name"])

print(meta["site_name"], "|", meta["species_code"], "-", meta["species_name"])

Big Bend National Park incl. Camp Springs | PSME - Douglas Fir

Data-quality warnings and heads-up¶

readers reads the file but flags spots worth a human's examination. For instance, a short interior gap (a missing year inside a series not indicated by a negative number), or characters past the last data column (indicating a possibly truncated value). Once again, this information is stored in df.attrs:

In [10]:

wa = dpl.readers(RWL + "wa082.rwl")
wa.attrs["dplpy_interior_gaps"]

wa = dpl.readers(RWL + "wa082.rwl") wa.attrs["dplpy_interior_gaps"]

wa082.rwl successfully extracted as rwl file with 23 series covering the period from 1698 to 1983 (auto-detected 3 header lines), with 1 unreadable value set to NaN
  heads-up: 1 short interior gap (a blank year between measured years -- possibly a dropped value) in 1 series:
    712011: 1900

/Users/kja/projects/opendendro/dplPy/src/dplpy/readers.py:1972: UserWarning: 1 anomalous negative value(s) (not the -9999 stop marker) were set to NaN [712011@1900]
  warnings.warn(

Out[10]:

[{'series': '712011',
  'gap_years': [1900],
  'short_gap_years': [1900],
  'max_run': 1}]

Reading directly from the ITRDB¶

readers allows you to directly read Tucson formatted tree-ring data from a URL, including from the ITRDB. In the example below, a Pinus echinata site from Oklahoma collected by Dave Stahle is fetched from the ITRDB, and readers warns the user that there is a likely formatting error that should be examined:

In [11]:

ok = dpl.readers("https://www.ncei.noaa.gov/pub/data/paleo/treering/measurements/northamerica/usa/ok049.rwl")

ok = dpl.readers("https://www.ncei.noaa.gov/pub/data/paleo/treering/measurements/northamerica/usa/ok049.rwl")

ok049.rwl successfully extracted as rwl file with 443 series covering the period from 1744 to 2020 (auto-detected 3 header lines)
  heads-up: 1 row had characters beyond the last data column (col 72) that were not parsed -- check for a truncated value:
    LIN152A 1839: read '303', ignored '5'  (looks like a split value)
  heads-up: 1 short interior gap (a blank year between measured years -- possibly a dropped value) in 1 series:
    LIN01A: 1939

A detective story: catching and repairing an oddity¶

Reading from the ITRDB is also where surprises turn up. Here is a real one. This European larch site from Mattertal, Switzerland (Hartl and colleagues, ITRDB che413) reads cleanly, but it carries a quirk that dpl.check is built to catch. It is important to practice safe ITRDBing!

dpl.check is the data-quality companion to readers: it screens a collection and returns problems as a table, one row per finding, each with a severity.

In [12]:

che = dpl.readers("https://www.ncei.noaa.gov/pub/data/paleo/treering/measurements/europe/che413.rwl")

che = dpl.readers("https://www.ncei.noaa.gov/pub/data/paleo/treering/measurements/europe/che413.rwl")

che413.rwl successfully extracted as rwl file with 38 series covering the period from 1596 to 2017 (auto-detected 3 header lines)

In [13]:

findings = dpl.check(che)
findings[["series", "check", "severity", "year_from", "year_to", "n", "message"]]

findings = dpl.check(che) findings[["series", "check", "severity", "year_from", "year_to", "n", "message"]]

Out[13]:

	series	check	severity	year_from	year_to	n	message
0	HCZ0106a	RWL_REPEATED_VALUE	warning	1769.0	1775.0	7.0	0.01 repeated in 7 consecutive rings, 1769-177...
1	HCZ0117a	RWL_REPEATED_VALUE	warning	1685.0	1690.0	6.0	0.01 repeated in 6 consecutive rings, 1685-169...
2	HCZ0101b	RWL_SERIES_OUTLIER	warning	NaN	NaN	NaN	correlates with the collection at r = 0.555 wh...
3	HCZ0112b	RWL_SERIES_OUTLIER	warning	NaN	NaN	NaN	correlates with the collection at r = 0.518 wh...
4	HCZ0102a	RWL_SHORT_OVERLAP	note	NaN	NaN	46.0	overlaps the rest of the collection by 46 ring...
5	HCZ0102b	RWL_SHORT_OVERLAP	note	NaN	NaN	46.0	overlaps the rest of the collection by 46 ring...
6	HCZ0104b	RWL_SHORT_OVERLAP	note	NaN	NaN	46.0	overlaps the rest of the collection by 46 ring...
7	HCZ0103b	RWL_SHORT_SERIES	note	NaN	NaN	29.0	29 rings, fewer than 30; too short to crossdat...

Two series carry a run of the same value, 0.01 mm, for several years in a row. A larch ring of 0.01 mm is not a real measurement. It is the smallest nonzero value the file can hold, and here it stands in for something else. Across the whole file the value 0.01 turns up 217 times in 24 series, so this is a convention, not a typo.

What does it mean? There are two likely readings, and the file does not say which:

• a not-measured value, a gap in the record, or
• a locally absent ring written as a tiny number because some detrending programs historically refused a literal 0.

Let us look at one of the flagged series.

In [ ]:

che["HCZ0106a"].loc[1766:1778] # this is a run of 0.01 between real rings measurements

che["HCZ0106a"].loc[1766:1778] # this is a run of 0.01 between real rings measurements

Out[ ]:

Year
1766    0.84
1767    0.87
1768    0.84
1769    0.01
1770    0.01
1771    0.01
1772    0.01
1773    0.01
1774    0.01
1775    0.01
1776    0.88
1777    0.01
1778    0.98
Name: HCZ0106a, dtype: float64

Real ring widths sit on both sides of this run of inplausible measurements. That is the case fill_internal is built for. First we mark every 0.01 placeholder as not-measured (NaN), which turns each run into an interior gap. Then fill_internal estimates the gap so the series can be used by tools that reject interior gaps, such as detrend and the crossdating checks. In the code below, we use fill="ARSTAN" to call on Ed Cook's spline based infilling method, which uses information from series with non-missing measurements to help estimate the missing values. Other options (inherited from dplR) include Spline,Mean,and Linear. See here for more: https://opendendro.github.io/dplPy/reference/utilities/#fill_internal

If instead we judged these to be locally absent rings, the right action would be to set them to 0, which dplPy keeps as real data rather than a gap. The choice is the analyst's. Here we take the not-measured reading.

In [36]:

# mark the 0.01 placeholder as not-measured, then fill the interior gaps
gaps = che.mask(che == 0.01)
filled_linear = dpl.fill_internal(gaps, fill="Linear") # linear interpolation infilling
filled_arstan = dpl.fill_internal(gaps, fill="ARSTAN") # use Ed Cook's spline-based ARSTAN infilling

# mark the 0.01 placeholder as not-measured, then fill the interior gaps gaps = che.mask(che == 0.01) filled_linear = dpl.fill_internal(gaps, fill="Linear") # linear interpolation infilling filled_arstan = dpl.fill_internal(gaps, fill="ARSTAN") # use Ed Cook's spline-based ARSTAN infilling

The placeholder run is gone, replaced by a smooth estimate that follows the neighboring rings. fill also accepts "Mean", "Linear", "ARSTAN", or a fixed number if you prefer a different estimate. A plot shows the repair in context.

In [39]:

ax = filled_linear["HCZ0106a"].loc[1699:1820].plot(figsize=(9, 3.5), color="b", label="after linear infilling")
ax = filled_arstan["HCZ0106a"].loc[1699:1820].plot(figsize=(9, 3.5), color="r", label="after ARSTAN infilling")
che["HCZ0106a"].loc[1699:1820].plot(ax=ax, color="k", lw=1, label="raw (0.01 placeholders)")
ax.set_xlabel("Year"); ax.set_ylabel("Ring width (mm)"); ax.set_title("HCZ0106a")
ax.legend()
plt.show()

ax = filled_linear["HCZ0106a"].loc[1699:1820].plot(figsize=(9, 3.5), color="b", label="after linear infilling") ax = filled_arstan["HCZ0106a"].loc[1699:1820].plot(figsize=(9, 3.5), color="r", label="after ARSTAN infilling") che["HCZ0106a"].loc[1699:1820].plot(ax=ax, color="k", lw=1, label="raw (0.01 placeholders)") ax.set_xlabel("Year"); ax.set_ylabel("Ring width (mm)"); ax.set_title("HCZ0106a") ax.legend() plt.show()

Duplicate IDs merged from separate segments¶

When a series ID appears more than once but do not have overlapping years of data, readers mimics dplR and joins them into a single series and reports the merge to the user. The idea of this default is that duplicate ID series that do not overlap in time are likely to reflect a single sample that needed to be measured in pieces (e.g due to damage, compression wood, etc.). However, if this is not the case, you can change this default behavior by setting join=False:

In [40]:

ca = dpl.readers(RWL + "ca667.rwl")

ca = dpl.readers(RWL + "ca667.rwl")

/Users/kja/projects/opendendro/dplPy/src/dplpy/readers.py:2051: UserWarning: Series ST850A had a duplicate ID in separate blocks with no overlapping years (joined into one series).
  warnings.warn("Series " + ids + " had a duplicate ID in separate blocks "

ca667.rwl successfully extracted as rwl file with 310 series covering the period from -2649 to 2005 (auto-detected 3 header lines)
  1 series combined from non-overlapping duplicate IDs (merged segments):
    ST850A: 2 segments spanning -662 to 453

Setting join=False will result in series with duplicate IDs being imported as separate series and renamed by appending a number to the series ID.

In [41]:

ca2 = dpl.readers(RWL + "ca667.rwl",join=False)

ca2 = dpl.readers(RWL + "ca667.rwl",join=False)

ca667.rwl successfully extracted as rwl file with 311 series covering the period from -2649 to 2005 (auto-detected 3 header lines)
  1 duplicate series ID kept as separate series (join=False):
    ST850A: disjoint block kept as ST850A2

Poorly formated Tucson files fail informatively (the strict=True default)¶

Rather than take heroic measures to read an .rwl file and potentially silently corrupt the data, in default mode readers fails to read the broken file and exits with a specific and actionable message identifying the questionable series and the year. This default behavior is set by strict=True:

In [42]:

for f in ["akfirmc.rwl", "cana326_error.rwl", "min.rwl", "viet001.rwl"]:
    try:
        dpl.readers(RWL + f)
    except ValueError as e:
        print(f"{f}:\n  {e}\n")

for f in ["akfirmc.rwl", "cana326_error.rwl", "min.rwl", "viet001.rwl"]: try: dpl.readers(RWL + f) except ValueError as e: print(f"{f}:\n {e}\n")

akfirmc.rwl:
  Cannot read file -- overlapping data:
  Series 'FAD23B' overlaps itself at year 1210: the row beginning 1205 has one value too many (only 5 fit before 1210) -- a misplaced value or mistyped start year.

cana326_error.rwl:
  Cannot read file -- data after the stop marker: series 'NRX14A' at year 1300. A block under an existing series ID appears after that series' stop marker (likely a mislabeled series ID).

min.rwl:
  Cannot read file -- misaligned fixed-width columns (an embedded space or TAB in a field): series 'SeriesA' at year 1 (field '10 30'); series 'SeriesB' at year 1 (field '200 4') (<TAB> marks a tab).

viet001.rwl:
  Cannot read file -- overlapping data:
  Duplicate series ID 'BDF02A' (two blocks overlap at year 1640) -- rename or remove the duplicate.

Salvage mode will do everything it can to read as much of the file (strict=False)¶

For batch processing, setting strict=False tells readers to recover whatever it reasonably can instead of failing. It may drop an unusable series and salvage the rest, it can join or rename duplicate IDs, and it records every decision and action in df.attrs["dplpy_salvage"].

In [43]:

kok = dpl.readers(RWL + "kyrg014.rwl", strict=False)
kok.attrs["dplpy_salvage"]

kok = dpl.readers(RWL + "kyrg014.rwl", strict=False) kok.attrs["dplpy_salvage"]

kyrg014.rwl successfully extracted as rwl file with 33 series covering the period from 1551 to 2005 (auto-detected 3 header lines)
  4 series combined from non-overlapping duplicate IDs (merged segments):
    kok2a: 2 segments spanning 1606 to 2004
    kok2b: 3 segments spanning 1552 to 2004
    kok5b: 2 segments spanning 1714 to 1989
    kok12a: 2 segments spanning 1653 to 2004

/Users/kja/projects/opendendro/dplPy/src/dplpy/readers.py:955: UserWarning: Salvaged kyrg014.rwl: dropped 1 series (kok3a:precision_shift)
  warnings.warn("Salvaged " + basename + ": " + "; ".join(parts))

Out[43]:

[{'series': 'kok3a',
  'issue': 'precision_shift',
  'action': 'dropped',
  'detail': 'stray -9999 near year 1898'},
 {'series': 'kok2a',
  'issue': 'duplicate_id',
  'action': 'joined',
  'detail': 'same ID in 2 separate blocks with no overlapping years, joined into one series'},
 {'series': 'kok2b',
  'issue': 'duplicate_id',
  'action': 'joined',
  'detail': 'same ID in 3 separate blocks with no overlapping years, joined into one series'},
 {'series': 'kok5b',
  'issue': 'duplicate_id',
  'action': 'joined',
  'detail': 'same ID in 2 separate blocks with no overlapping years, joined into one series'},
 {'series': 'kok12a',
  'issue': 'duplicate_id',
  'action': 'joined',
  'detail': 'same ID in 2 separate blocks with no overlapping years, joined into one series'}]

In [22]:

bdf = dpl.readers(RWL + "viet001.rwl", strict=False)

bdf = dpl.readers(RWL + "viet001.rwl", strict=False)

viet001.rwl successfully extracted as rwl file with 82 series covering the period from 1030 to 2008

/Users/kja/projects/opendendro/dplPy/src/dplpy/readers.py:955: UserWarning: Salvaged viet001.rwl: renamed 1 (renamed to BDF02A2)
  warnings.warn("Salvaged " + basename + ": " + "; ".join(parts))

readers handles mixed measurement precision just fine¶

A single Tucson file can have series measured at different precisions. In this example, all the TMS* cores at 0.001 mm and all the TWM* were measured at 0.01 mm many years earlier. readers automatically scales each correctly:

In [44]:

tms = dpl.readers(RWL + "TMScombined.rwl")
tms.plot(legend=False,figsize=(9, 4))
plt.xlabel("Year")
plt.ylabel("ring width (mm)")
plt.title("Twelvemile Summit — original and update")
plt.show()

tms = dpl.readers(RWL + "TMScombined.rwl") tms.plot(legend=False,figsize=(9, 4)) plt.xlabel("Year") plt.ylabel("ring width (mm)") plt.title("Twelvemile Summit — original and update") plt.show()

/Users/kja/projects/opendendro/dplPy/src/dplpy/readers.py:1972: UserWarning: 1 anomalous negative value(s) (not the -9999 stop marker) were set to NaN [TWM31a@1846]
  warnings.warn(

TMScombined.rwl successfully extracted as rwl file with 142 series covering the period from 1600 to 2021, with 1 unreadable value set to NaN
  heads-up: 1 short interior gap (a blank year between measured years -- possibly a dropped value) in 1 series:
    TWM31a: 1846

Non-ASCII series IDs are handled correctly¶

Series IDs containing non-ASCII characters are preserved as-is:

In [45]:

ru = dpl.readers(RWL + "russ301.rwl")
[c for c in ru.columns if any(ord(ch) > 127 for ch in c)][:6]

ru = dpl.readers(RWL + "russ301.rwl") [c for c in ru.columns if any(ord(ch) > 127 for ch in c)][:6]

russ301.rwl successfully extracted as rwl file with 80 series covering the period from 1847 to 2018 (auto-detected 3 header lines)

Out[45]:

['Áä89-06', 'Áä89-10', 'Áä89-12', 'Áä89-13', 'Áä89-14', 'Áä89-15']

readers also imports CSV files¶

readers also reads year-indexed CSV ring-width files:

In [46]:

csv = dpl.readers(CSV + "ca533.csv")
csv

csv = dpl.readers(CSV + "ca533.csv") csv

ca533.csv successfully extracted as csv file with 34 series covering the period from 626 to 1983

Out[46]:

	CAM011	CAM021	CAM031	CAM032	CAM041	CAM042	CAM051	CAM061	CAM062	CAM071	...	CAM151	CAM152	CAM161	CAM162	CAM171	CAM172	CAM181	CAM191	CAM201	CAM211
Year
626	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	0.17
627	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	0.13
628	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	0.14
629	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	0.19
630	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	0.22
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
1979	0.40	0.42	0.33	0.61	0.75	0.50	0.25	0.42	0.58	0.11	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1980	0.45	0.59	0.77	0.67	1.09	0.65	0.36	0.69	0.71	0.14	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1981	0.48	0.57	0.69	0.55	0.76	0.56	0.32	0.64	0.51	0.27	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1982	0.47	0.50	0.76	0.71	1.02	0.59	0.25	0.67	0.57	0.30	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1983	0.68	0.57	0.53	0.58	1.16	0.59	0.22	0.70	0.37	0.32	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

1358 rows × 34 columns

Documentation¶

The complete documentation is available in the dplPy documentation here: https://opendendro.github.io/dplPy/reference/io/#readers