Publish PAH-ENV-001 research edition

.gitattributes

@@ -1,60 +1,3 @@
-*.7z filter=lfs diff=lfs merge=lfs -text
-*.arrow filter=lfs diff=lfs merge=lfs -text
-*.avro filter=lfs diff=lfs merge=lfs -text
-*.bin filter=lfs diff=lfs merge=lfs -text
-*.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
-*.ftz filter=lfs diff=lfs merge=lfs -text
-*.gz filter=lfs diff=lfs merge=lfs -text
-*.h5 filter=lfs diff=lfs merge=lfs -text
-*.joblib filter=lfs diff=lfs merge=lfs -text
-*.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.lz4 filter=lfs diff=lfs merge=lfs -text
-*.mds filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
-*.model filter=lfs diff=lfs merge=lfs -text
-*.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
-*.onnx filter=lfs diff=lfs merge=lfs -text
-*.ot filter=lfs diff=lfs merge=lfs -text
 *.parquet filter=lfs diff=lfs merge=lfs -text
-*.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
-*.pt filter=lfs diff=lfs merge=lfs -text
-*.pth filter=lfs diff=lfs merge=lfs -text
-*.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text
-*.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
-*.tflite filter=lfs diff=lfs merge=lfs -text
-*.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
-*.xz filter=lfs diff=lfs merge=lfs -text
-*.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
-# Audio files - uncompressed
-*.pcm filter=lfs diff=lfs merge=lfs -text
-*.sam filter=lfs diff=lfs merge=lfs -text
-*.raw filter=lfs diff=lfs merge=lfs -text
-# Audio files - compressed
-*.aac filter=lfs diff=lfs merge=lfs -text
-*.flac filter=lfs diff=lfs merge=lfs -text
-*.mp3 filter=lfs diff=lfs merge=lfs -text
-*.ogg filter=lfs diff=lfs merge=lfs -text
-*.wav filter=lfs diff=lfs merge=lfs -text
-# Image files - uncompressed
-*.bmp filter=lfs diff=lfs merge=lfs -text
-*.gif filter=lfs diff=lfs merge=lfs -text
+*.csv.gz filter=lfs diff=lfs merge=lfs -text
 *.png filter=lfs diff=lfs merge=lfs -text
-*.tiff filter=lfs diff=lfs merge=lfs -text
-# Image files - compressed
-*.jpg filter=lfs diff=lfs merge=lfs -text
-*.jpeg filter=lfs diff=lfs merge=lfs -text
-*.webp filter=lfs diff=lfs merge=lfs -text
-# Video files - compressed
-*.mp4 filter=lfs diff=lfs merge=lfs -text
-*.webm filter=lfs diff=lfs merge=lfs -text

CHECKSUMS.sha256

@@ -0,0 +1,13 @@
+68e9289e9a7ffd0cdf8a57c357685f21ae8bcee54121171178e9bd460d70e01f  .gitattributes
+1ab0acbfc920e33e8db781c8e8f0f22f8c1ae1473cacc04ddb15638a0a35db55  assets/paired_signals_plot1_6cm_2022.csv
+ab9472d776fd6748a59d6f1f32d4367e00ac735190f86a453b72301b05acb42b  assets/paired_signals_plot1_6cm_2022.png
+7e6de57684030c9ff5d25dc0dbd799bb53b477f5096284dbc5d75b47f4293a09  data/PAH-ENV-001_soil_moisture_30min.parquet
+f4ab39bb4cf7fb8e67795aa855d635f1de3fb0fe27725732903836852c715d4e  data/PAH-ENV-001_soil_temperature_30min.parquet
+8735072a352201daf6bdf04df95d14d665bc40714461ae33f176c0e7c8e4ad11  docs/ATTRIBUTION.md
+58440a64e23994bfab78ebadabdf8f98c81450b15cf68110b3f36d8fb4d69b89  docs/DATA_DICTIONARY.md
+e37c64e50516333b8b9a132eac7cdb0949d63028d43795f501907cb66fb4a951  docs/DATASET_CARD.md
+00023a205c6372ff0a36b453b674a817edded2402d446b95b3fdb7442ec2a34f  docs/METHODOLOGY.md
+8db621f845c37931415a9eaa67018bfedf80b9786d5de8ca2eaec2e6a1a9a1b1  docs/PAH-ENV-001_gap_report.json
+19661d094a05c2d896c12e2ce7f109bfd1260d0b3a07fe9b13eb5bd90d87a168  docs/PROVENANCE.jsonl
+000181e6e298874baabd246a90dd46fb6d49cb3c58fa78eabaa07823222882ef  examples/quick_start.ipynb
+9e1bdd3313961107a42fc8011a602f20f71dc845e695d912261eb2f5b820b961  README.md

README.md

@@ -0,0 +1,121 @@
+---
+license: cc0-1.0
+language:
+  - en
+tags:
+  - environmental-monitoring
+  - soil-moisture
+  - soil-temperature
+  - sensor-data
+  - time-series
+  - agriculture
+  - earth-science
+  - neon
+  - parquet
+size_categories:
+  - 10M<n<100M
+task_categories:
+  - time-series-forecasting
+  - tabular-regression
+pretty_name: "Pahzuzu SERC Paired Soil Moisture & Temperature (2016–2022)"
+configs:
+  - config_name: soil_moisture
+    data_files:
+      - split: train
+        path: data/PAH-ENV-001_soil_moisture_30min.parquet
+    default: true
+  - config_name: soil_temperature
+    data_files:
+      - split: train
+        path: data/PAH-ENV-001_soil_temperature_30min.parquet
+---
+
+# Pahzuzu SERC Paired Soil Moisture & Temperature (2016–2022)
+
+More than 10 million paired, 30-minute soil observations from five plots at the Smithsonian Environmental Research Center (SERC), cleaned into two analysis-ready Parquet tables with preserved quality flags and machine-readable provenance.
+
+This free research edition contains the complete moisture and temperature tables. It exists so researchers and engineers can inspect the actual data, limitations, schema and provenance before deciding whether the supported commercial bundle is useful to them.
+
+## Dataset at a glance
+
+| | Soil moisture | Soil temperature |
+|---|---:|---:|
+| Rows | 4,721,280 | 5,311,440 |
+| Depths | 8 (6–141 cm) | 9 (6–191 cm) |
+| Plots | 5 | 5 |
+| Time range | 2016-04-08 to 2022-12-31 | 2016-04-08 to 2022-12-31 |
+| Resolution | 30 minutes, UTC | 30 minutes, UTC |
+| NEON QF=0 | 40.2% | 96.7% |
+| Rows containing the primary sensor value | 68.0% | 97.0% |
+
+The moisture gaps are real and important. They are documented rather than interpolated away. See the [gap report](docs/PAH-ENV-001_gap_report.json) and [dataset card](docs/DATASET_CARD.md) before modelling.
+
+## Viewer subsets
+
+Use the Dataset Viewer subset selector to switch between:
+
+- `soil_moisture`: water content, ion content, uncertainties and NEON flags.
+- `soil_temperature`: temperature, uncertainty and NEON flags.
+
+Both tables join on `timestamp_utc`, `site_code`, `plot_id` and `depth_cm`. Temperature includes one extra depth at 191 cm.
+
+## Five-minute quick start
+
+```python
+from datasets import load_dataset
+
+moisture = load_dataset(
+    "Pahzuzu/PAH-ENV-001-serc-soil-signals",
+    "soil_moisture",
+    split="train",
+)
+temperature = load_dataset(
+    "Pahzuzu/PAH-ENV-001-serc-soil-signals",
+    "soil_temperature",
+    split="train",
+)
+
+print(moisture.features)
+print(temperature.features)
+```
+
+For a reproducible local workflow, open [`examples/quick_start.ipynb`](examples/quick_start.ipynb). It loads the Parquet tables, filters NEON quality flags, aligns matching plot/depth observations and creates a paired daily signal.
+
+![Daily soil moisture and temperature for SERC Plot 1 at 6 cm during 2022](assets/paired_signals_plot1_6cm_2022.png)
+
+## Files
+
+- `data/PAH-ENV-001_soil_moisture_30min.parquet`
+- `data/PAH-ENV-001_soil_temperature_30min.parquet`
+- `docs/DATASET_CARD.md`
+- `docs/DATA_DICTIONARY.md`
+- `docs/METHODOLOGY.md`
+- `docs/ATTRIBUTION.md`
+- `docs/PROVENANCE.jsonl`
+- `docs/PAH-ENV-001_gap_report.json`
+- `examples/quick_start.ipynb`
+
+## Quality guidance
+
+- Start strict: retain rows where the relevant NEON final quality flag equals `0`.
+- Inspect flagged, non-null moisture values separately; a flag is not the same thing as a missing measurement.
+- Choose plots and years from the gap report rather than assuming uniform completeness.
+- Do not treat this as a spatially representative survey of the entire SERC site; it is a five-plot sensor record.
+- No measurement values were interpolated, gap-filled or synthetically generated by Pahzuzu.
+
+## Source, licence and provenance
+
+The measurements derive from NEON products DP1.00094.001 and DP1.00041.001 at SERC. The source files were acquired in May 2026 under NEON's CC0-era terms, as recorded in `docs/PROVENANCE.jsonl`. NEON later moved newly released data to CC BY 4.0. Read [`docs/ATTRIBUTION.md`](docs/ATTRIBUTION.md) for the full transition note and citation guidance.
+
+- Cleaned data tables: CC0 1.0 Universal.
+- Pahzuzu-authored research documentation: CC BY 4.0.
+
+This is not legal advice; users with strict compliance requirements should independently confirm the source-data terms that apply to their use.
+
+## Supported commercial bundle
+
+The underlying public measurements are not exclusive. The paid bundle is for teams that want a single downloadable package, CSV.gz compatibility files, a commercial support commitment and update notices.
+
+[View the supported Pahzuzu bundle](https://pahzuzu.gumroad.com/l/pahzuzu-fortunate-finds-serc-soil-moisture-temperature?utm_source=huggingface&utm_medium=dataset&utm_campaign=pah_env_001_launch)
+
+Questions about schema, provenance or custom variants: `pahzuzu.data@proton.me`

assets/paired_signals_plot1_6cm_2022.csv

@@ -0,0 +1,343 @@
+date,soil_moisture_m3_m3,soil_temperature_c
+2022-01-01,0.1696222222222222,10.596958333333333
+2022-01-02,0.1744860465116279,12.136562499999998
+2022-01-03,0.16831481481481483,8.298864864864864
+2022-01-04,0.17231000000000002,5.0607999999999995
+2022-01-05,0.17254848484848484,4.44925
+2022-01-06,0.17575625,4.111354166666667
+2022-01-07,0.1763703703703704,3.8319791666666667
+2022-01-08,0.1715642857142857,3.3913541666666664
+2022-01-09,0.17652,3.266458333333333
+2022-01-10,0.18398085106382978,3.324145833333333
+2022-01-11,0.1729939393939394,1.5955833333333331
+2022-01-12,0.16985151515151517,1.3736249999999999
+2022-01-13,0.1702590909090909,2.9231702127659576
+2022-01-14,0.1711035714285714,4.2253541666666665
+2022-01-15,0.16785,1.9837083333333334
+2022-01-16,0.16493076923076921,0.20879166666666668
+2022-01-17,0.19514864864864867,2.687127659574468
+2022-01-18,0.180496875,2.9907500000000002
+2022-01-19,0.175125,3.0140208333333334
+2022-01-20,0.18347857142857144,4.923979166666666
+2022-01-21,0.17692105263157895,1.398625
+2022-01-22,0.17406666666666668,0.9086041666666667
+2022-01-23,0.1726088888888889,1.1446458333333334
+2022-01-24,0.17164642857142856,2.4771875
+2022-01-25,0.17047391304347825,3.416541666666667
+2022-01-26,0.1708806451612903,2.3794375
+2022-01-27,0.1678605263157895,0.519
+2022-01-28,0.16859166666666667,2.1257708333333336
+2022-01-29,0.169605,2.2655833333333333
+2022-01-30,0.1689304347826087,0.7658541666666667
+2022-01-31,0.16838285714285714,1.4215625
+2022-02-01,0.1682479166666667,1.1266458333333333
+2022-02-02,0.17208958333333335,1.7603541666666667
+2022-02-03,0.18692916666666667,3.6603125
+2022-02-04,0.2013309523809524,6.15475
+2022-02-05,0.1911657142857143,3.6494375
+2022-02-06,0.18133030303030304,1.49225
+2022-02-07,0.17969148936170212,3.0975625
+2022-02-08,0.18168,4.125708333333333
+2022-02-09,0.17779166666666668,3.4173958333333334
+2022-02-10,0.17644137931034484,5.134895833333333
+2022-02-11,0.17525,5.4057708333333325
+2022-02-12,0.1745857142857143,7.4224375
+2022-02-13,0.17458888888888888,5.426583333333333
+2022-02-14,0.17429999999999998,2.3691666666666666
+2022-02-15,0.17201249999999998,1.1008297872340425
+2022-02-16,0.16714411764705883,2.9241458333333337
+2022-02-17,0.16650689655172413,6.249604166666667
+2022-02-18,0.17143333333333333,8.287791666666667
+2022-02-19,0.17045000000000002,3.9378124999999997
+2022-02-20,0.16843684210526316,2.3996875
+2022-02-21,0.16730243902439024,4.2754375
+2022-02-22,0.16651395348837209,6.358666666666667
+2022-02-23,0.16726388888888888,11.084479166666666
+2022-02-24,0.1662965517241379,6.905708333333333
+2022-02-25,0.1839925,5.534916666666667
+2022-02-26,0.17754642857142858,5.335270833333333
+2022-02-27,0.17531315789473684,5.492562500000001
+2022-02-28,0.17306363636363636,5.613812500000001
+2022-03-01,0.17164166666666666,5.209604166666667
+2022-03-02,0.17137750000000002,6.749041666666667
+2022-03-03,0.17026341463414635,7.5734375
+2022-03-04,0.16912258064516128,4.988979166666667
+2022-03-05,0.16951666666666665,5.596395833333333
+2022-03-06,0.17159117647058825,9.485479166666666
+2022-03-07,0.17099,13.118604166666666
+2022-03-08,0.16994255319148938,10.665687499999999
+2022-03-09,0.19258666666666666,7.5601875000000005
+2022-03-10,0.18781666666666666,6.8205625
+2022-03-11,0.18479375,7.623895833333333
+2022-03-12,0.2006282608695652,6.674166666666667
+2022-03-13,0.19481162790697673,3.398270833333333
+2022-03-14,0.19037142857142858,5.71125
+2022-03-15,0.18828421052631578,7.917854166666667
+2022-03-16,0.18559714285714285,9.392270833333333
+2022-03-17,0.20289166666666666,9.592333333333334
+2022-03-18,0.20303125,11.244791666666666
+2022-03-19,0.19564468085106382,12.759708333333334
+2022-03-20,0.19358723404255318,12.073104166666667
+2022-03-21,0.18959318181818183,10.750041666666666
+2022-03-22,0.18784375,11.182229166666666
+2022-03-23,0.18731521739130436,9.824666666666667
+2022-03-24,0.2069291666666667,11.080479166666665
+2022-03-25,0.19789268292682927,11.199708333333334
+2022-03-26,0.19695333333333334,10.685375
+2022-03-27,0.1938423076923077,7.661520833333333
+2022-03-29,0.190625,5.268020833333334
+2022-03-30,0.19117499999999998,6.303541666666667
+2022-04-02,0.1974,9.22125
+2022-04-03,0.18904,10.353541666666667
+2022-04-04,0.18873333333333334,9.277270833333334
+2022-04-05,0.1899111111111111,9.667208333333333
+2022-04-06,0.22965454545454547,11.459687500000001
+2022-04-07,0.22136,11.046229166666668
+2022-04-08,0.21763658536585365,10.903479166666665
+2022-04-09,0.21595555555555557,10.730895833333333
+2022-04-10,0.21209545454545456,9.466708333333333
+2022-04-11,0.2077794871794872,9.283916666666666
+2022-04-12,0.2058,12.659479166666666
+2022-04-13,0.2017,14.793978723404255
+2022-04-14,0.20108888888888887,17.011791666666667
+2022-04-15,0.19825581395348837,14.531895833333335
+2022-04-16,0.1952630434782609,15.045895833333333
+2022-04-17,0.19933589743589744,13.787833333333333
+2022-04-18,0.20722413793103447,10.267604166666667
+2022-04-19,0.24099545454545454,9.6249375
+2022-04-20,0.23125,10.5358125
+2022-04-21,0.2252,11.370333333333333
+2022-04-22,0.212805,13.630916666666666
+2022-04-23,0.21121612903225806,13.739166666666668
+2022-04-24,0.209675,14.821083333333334
+2022-04-26,0.2049,14.485833333333334
+2022-04-27,0.20965,13.243895833333333
+2022-04-28,0.2137,11.316895833333334
+2022-04-29,0.2011,11.444604166666666
+2022-05-01,0.1944,11.706458333333332
+2022-05-02,0.1989,13.304770833333334
+2022-05-03,0.19240249999999998,14.1934375
+2022-05-04,0.21555384615384615,14.077833333333333
+2022-05-05,0.20413846153846155,14.316125
+2022-05-06,0.20845675675675676,13.909999999999998
+2022-05-07,0.23418510638297874,12.497895833333333
+2022-05-08,0.23547037037037039,10.965916666666667
+2022-05-09,0.22585,11.666666666666666
+2022-05-10,0.2206307692307692,12.843333333333334
+2022-05-11,0.21291612903225807,13.592145833333333
+2022-05-12,0.20952093023255813,14.171333333333335
+2022-05-13,0.21466666666666667,15.203395833333333
+2022-05-14,0.21660000000000001,15.953833333333334
+2022-05-15,0.21263777777777776,16.3081875
+2022-05-16,0.2111333333333333,16.923916666666667
+2022-05-17,0.2091186046511628,16.526374999999998
+2022-05-18,0.20339756097560974,16.097666666666665
+2022-05-19,0.20220833333333332,16.7214375
+2022-05-20,0.19918333333333335,17.7259375
+2022-05-21,0.19567317073170734,19.848895833333334
+2022-05-22,0.1922388888888889,20.283104166666664
+2022-05-23,0.2194093023255814,18.72658695652174
+2022-05-24,0.22259791666666665,16.638729166666668
+2022-05-25,0.21665652173913044,16.036478260869565
+2022-05-26,0.21201578947368419,16.455208333333335
+2022-05-27,0.21651666666666666,17.617458333333335
+2022-05-28,0.2340375,18.049729166666665
+2022-05-29,0.2222891891891892,18.1618125
+2022-05-30,0.21624705882352943,18.935208333333332
+2022-05-31,0.21225588235294118,20.275125
+2022-06-01,0.20765714285714287,21.191229166666666
+2022-06-02,0.216775,20.889458333333334
+2022-06-03,0.23284722222222223,19.94448936170213
+2022-06-04,0.221888,18.839187499999998
+2022-06-05,0.2140177777777778,18.575229166666666
+2022-06-06,0.20913947368421054,17.871104166666665
+2022-06-07,0.2058945945945946,18.730770833333334
+2022-06-08,0.204440625,19.69966666666667
+2022-06-09,0.20383846153846155,20.469541666666668
+2022-06-10,0.19550731707317073,19.006083333333333
+2022-06-11,0.1929741935483871,18.58325
+2022-06-12,0.19143076923076924,18.6571875
+2022-06-13,0.18809428571428569,20.3383125
+2022-06-14,0.2072225,20.55452083333333
+2022-06-15,0.20509444444444444,20.43883333333333
+2022-06-16,0.20023863636363634,20.83427083333333
+2022-06-17,0.19514864864864867,21.721125
+2022-06-18,0.1850113636363636,20.772854166666665
+2022-06-19,0.1736711111111111,18.675124999999998
+2022-06-20,0.16407659574468084,18.008458333333333
+2022-06-21,0.155195,18.163065217391303
+2022-06-22,0.14755,19.3164
+2022-06-23,0.19514782608695655,19.123041666666666
+2022-06-24,0.18947857142857144,18.662979166666666
+2022-06-25,0.17776,19.99083333333333
+2022-06-26,0.1677063829787234,20.854854166666666
+2022-06-27,0.1706452380952381,21.187583333333333
+2022-06-28,0.176125,19.8425
+2022-06-29,0.15960227272727273,19.533333333333335
+2022-06-30,0.15067826086956523,21.08760416666667
+2022-07-01,0.14023529411764704,22.119645833333333
+2022-07-02,0.13016666666666665,22.760229166666665
+2022-07-03,0.18103541666666667,22.21066666666667
+2022-07-04,0.16813809523809523,21.359854166666665
+2022-07-05,0.17192432432432433,21.069854166666666
+2022-07-06,0.19857499999999997,22.590541666666667
+2022-07-07,0.20503541666666666,22.596895833333335
+2022-07-08,0.19411063829787237,22.0378125
+2022-07-09,0.22253541666666665,21.925479166666666
+2022-07-10,0.21750555555555554,21.320020833333334
+2022-07-11,0.20856451612903226,20.474416666666666
+2022-07-12,0.20448636363636366,21.4344375
+2022-07-13,0.21155333333333332,22.075354166666667
+2022-07-14,0.2016260869565217,22.591520833333334
+2022-07-15,0.19860416666666666,22.462999999999997
+2022-07-16,0.20263488372093025,22.145145833333334
+2022-07-17,0.2380893617021277,21.8989375
+2022-07-18,0.22528378378378378,23.194104166666666
+2022-07-19,0.21774878048780488,23.448270833333336
+2022-07-20,0.21186595744680853,23.616874999999997
+2022-07-21,0.21080232558139536,24.133395833333335
+2022-07-22,0.22576595744680852,23.858625
+2022-07-23,0.21416363636363636,24.212979166666667
+2022-07-24,0.2060818181818182,23.87908333333333
+2022-07-25,0.20719722222222223,24.576125
+2022-07-26,0.2172175,23.362770833333332
+2022-07-27,0.20820540540540539,23.124875
+2022-07-28,0.20713170731707317,24.0436875
+2022-07-29,0.23097,23.28033333333333
+2022-07-30,0.2128625,23.244
+2022-07-31,0.2042,22.549229166666667
+2022-08-01,0.21291500000000002,22.541854166666667
+2022-08-02,0.19954324324324324,23.395166666666668
+2022-08-03,0.18916666666666668,23.48164583333333
+2022-08-04,0.1803590909090909,23.96858333333333
+2022-08-05,0.23629111111111112,23.639375
+2022-08-06,0.24016444444444443,23.495041666666665
+2022-08-07,0.22315454545454544,24.242166666666666
+2022-08-08,0.21674347826086957,24.791645833333334
+2022-08-09,0.20997727272727273,25.011520833333336
+2022-08-10,0.21056046511627907,24.72191666666667
+2022-08-11,0.23880208333333333,23.27870833333333
+2022-08-12,0.2233175,22.59420833333333
+2022-08-13,0.2133741935483871,20.73122916666667
+2022-08-14,0.2059608695652174,20.594958333333334
+2022-08-15,0.2089388888888889,20.847395833333334
+2022-08-16,0.19738947368421053,20.752083333333335
+2022-08-17,0.19272666666666669,20.3776875
+2022-08-18,0.19363939393939394,20.1254375
+2022-08-19,0.18237142857142857,21.325666666666667
+2022-08-20,0.16993055555555556,21.994937500000002
+2022-08-21,0.1592025,21.982666666666663
+2022-08-22,0.19892826086956522,22.472354166666666
+2022-08-23,0.1965311111111111,22.527833333333334
+2022-08-24,0.1706142857142857,22.014717391304348
+2022-08-25,0.15557272727272725,22.09675
+2022-08-26,0.14133846153846152,22.9908125
+2022-08-27,0.1290711111111111,23.132
+2022-08-28,0.11599024390243903,22.9248125
+2022-08-29,0.10556341463414634,23.543229166666666
+2022-08-30,0.0936,23.732041666666664
+2022-08-31,0.08508666666666667,22.527666666666665
+2022-09-01,0.07501428571428571,21.150625
+2022-09-02,0.06653,21.064208333333333
+2022-09-03,0.060950000000000004,21.813166666666664
+2022-09-04,0.05633333333333333,22.137541666666667
+2022-09-05,0.05296551724137931,22.373229166666665
+2022-09-06,0.07322000000000001,22.584416666666666
+2022-09-07,0.08732727272727273,21.677125
+2022-09-08,0.0936974358974359,21.042020833333332
+2022-09-09,0.08527105263157894,20.222666666666665
+2022-09-10,0.0749030303030303,20.186375
+2022-09-11,0.0904,21.42441666666667
+2022-09-12,0.12316551724137931,22.462645833333337
+2022-09-13,0.14648372093023257,21.982083333333335
+2022-09-14,0.1325575,19.840708333333335
+2022-09-15,0.123368,19.788104166666667
+2022-09-16,0.11618947368421054,18.406166666666667
+2022-09-17,0.10711538461538463,18.900791666666667
+2022-09-18,0.09890344827586206,19.607291666666665
+2022-09-19,0.08972413793103448,20.519041666666666
+2022-09-20,0.08081515151515152,20.734437500000002
+2022-09-21,0.07209642857142858,19.866520833333333
+2022-09-22,0.06630833333333334,20.534145833333334
+2022-09-23,0.059455999999999995,17.029
+2022-09-24,0.05469999999999999,15.130562500000002
+2022-09-25,0.055856,16.575541666666666
+2022-09-26,0.07150975609756098,17.782770833333334
+2022-09-27,0.0680264705882353,16.915354166666667
+2022-09-28,0.06204166666666666,15.2645
+2022-09-29,0.05712105263157894,15.639770833333335
+2022-09-30,0.05382826086956522,15.216270833333333
+2022-10-01,0.10548636363636364,15.289208333333335
+2022-10-02,0.15381956521739132,15.835104166666667
+2022-10-03,0.16625,12.966687499999999
+2022-10-04,0.17792916666666667,11.767916666666666
+2022-10-05,0.18882888888888888,12.862326086956523
+2022-10-06,0.17082999999999998,14.531229166666668
+2022-10-07,0.164025,15.283333333333333
+2022-10-08,0.15978421052631578,14.507333333333333
+2022-10-09,0.1570888888888889,11.772874999999999
+2022-10-10,0.15511999999999998,12.0531875
+2022-10-11,0.15257435897435898,12.964041666666667
+2022-10-12,0.15081463414634147,13.544166666666667
+2022-10-13,0.16409347826086956,15.826333333333332
+2022-10-14,0.16785,15.002375
+2022-10-15,0.16081904761904764,13.479750000000001
+2022-10-16,0.16002857142857144,14.384062499999999
+2022-10-17,0.15695,15.132916666666667
+2022-10-18,0.1635,12.7211875
+2022-10-19,0.15678333333333333,10.392625
+2022-10-20,0.1561095238095238,10.336645833333334
+2022-10-21,0.15275,10.790833333333333
+2022-10-22,0.15118085106382978,11.334354166666666
+2022-10-23,0.15220425531914894,12.071979166666667
+2022-10-24,0.18089545454545455,13.650687499999998
+2022-10-25,0.17014583333333333,14.396791666666667
+2022-10-26,0.17680217391304345,14.871875000000001
+2022-10-27,0.17273095238095237,14.055041666666668
+2022-10-28,0.16870285714285715,12.420791666666666
+2022-10-29,0.16732142857142857,13.050333333333334
+2022-10-30,0.1656325581395349,11.412645833333334
+2022-10-31,0.167575,12.9329375
+2022-11-01,0.1831918918918919,14.941833333333333
+2022-11-02,0.17624285714285715,14.541291666666666
+2022-11-03,0.17603333333333335,13.6700625
+2022-11-05,0.17337142857142857,15.544829787234042
+2022-11-06,0.17351176470588237,16.6714375
+2022-11-07,0.1736111111111111,17.391604166666667
+2022-11-08,0.1724384615384615,14.37868085106383
+2022-11-09,0.16972777777777778,12.216666666666667
+2022-11-10,0.1690047619047619,12.538854166666667
+2022-11-11,0.2114125,14.834104166666668
+2022-11-12,0.20083720930232557,16.644145833333333
+2022-11-13,0.19390869565217392,13.388145833333333
+2022-11-14,0.18660465116279068,9.942333333333334
+2022-11-15,0.18782500000000002,8.750645833333333
+2022-11-16,0.21428695652173915,9.716666666666667
+2022-11-17,0.19396666666666668,8.901291666666667
+2022-11-18,0.19002000000000002,7.379145833333333
+2022-11-19,0.18739285714285714,7.507229166666666
+2022-11-20,0.18595208333333332,6.145291666666666
+2022-11-21,0.18741818181818182,5.006854166666667
+2022-11-22,0.18557692307692308,6.4035
+2022-11-23,0.1804,7.116770833333334
+2022-11-27,0.20780625,8.912770833333333
+2022-11-28,0.20118125,10.143166666666668
+2022-11-29,0.19205,8.442291666666668
+2022-11-30,0.20725833333333332,9.5213125
+2022-12-01,0.19960645161290322,7.345229166666667
+2022-12-02,0.19391724137931032,6.000395833333333
+2022-12-03,0.1942,9.268083333333333
+2022-12-04,0.199,7.820083333333333
+2022-12-05,0.19855,5.540717391304348
+2022-12-07,0.2262,9.842075
+2022-12-14,0.1946466666666667,4.254770833333334
+2022-12-15,0.22555833333333333,5.398375000000001
+2022-12-16,0.23270416666666668,7.551895833333333
+2022-12-17,0.22089142857142857,5.647125
+2022-12-18,0.21474444444444446,5.597729166666666
+2022-12-19,0.21080000000000002,3.9066041666666664
+2022-12-20,0.2094,3.533804347826087
+2022-12-21,0.214275,3.3049999999999997
+2022-12-23,0.24728666666666665,6.480208333333334
+2022-12-29,0.22453076923076923,3.118229166666667
+2022-12-30,0.2230181818181818,2.5568695652173914

data/PAH-ENV-001_soil_moisture_30min.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e6de57684030c9ff5d25dc0dbd799bb53b477f5096284dbc5d75b47f4293a09
+size 38680892

data/PAH-ENV-001_soil_temperature_30min.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4ab39bb4cf7fb8e67795aa855d635f1de3fb0fe27725732903836852c715d4e
+size 14658001

docs/ATTRIBUTION.md

@@ -0,0 +1,108 @@
+# Attribution and Source Acknowledgement
+
+This dataset pack, **Pahzuzu Fortunate Finds: SERC Paired Soil Moisture and Temperature (2016–2022)** (`PAH-ENV-001`), is a derivative work created by Pahzuzu Data from publicly available NEON source data. The cleaning, schema normalisation, depth code translation, gap documentation, and metadata enrichment are the contribution of Pahzuzu Data. All credit for the underlying physical measurements, instrumentation, calibration, and original data collection belongs to NEON and its operating partners.
+
+---
+
+## Source data
+
+**Original publisher**
+
+The National Ecological Observatory Network (NEON) — a continental-scale ecological observatory operated by **Battelle** with primary funding from the **US National Science Foundation** (NSF) under Cooperative Agreement DBI-1638695.
+
+**Source products**
+
+| Product | Title | URL |
+|---|---|---|
+| DP1.00094.001 | Soil water content and water salinity | https://data.neonscience.org/data-products/DP1.00094.001 |
+| DP1.00041.001 | Soil temperature | https://data.neonscience.org/data-products/DP1.00041.001 |
+
+**Source site**
+
+Smithsonian Environmental Research Center (SERC), Edgewater, Maryland, USA. NEON Domain D02 (Mid-Atlantic). Site coordinates: approximately 38.890° N, 76.560° W.
+
+**Collection period**
+
+April 2016 through December 2022 inclusive (the pre-2023 portion of the SERC soil sensor records as of the acquisition date).
+
+**Original licence**
+
+CC0 1.0 Universal (Public Domain Dedication) — https://creativecommons.org/publicdomain/zero/1.0/
+
+NEON deliberately publishes its data under the CC0 dedication, and that dedication — not any automatic statutory rule — is the operative basis for reuse. Federal-funding status is a supporting consideration rather than a guaranteed trigger on its own: 17 USC §105 covers works of the US federal government itself, and NEON's data is produced by a federally funded observatory operated by Battelle, a private contractor. The CC0 dedication removes any ambiguity about derivative works, commercial use, or attribution requirements for reuse under its terms.
+
+**Licence transition note (2026-06-30):** NEON's usage policy moved from CC0 to CC BY 4.0 on 30 June 2026 for newly released data. The source files in this pack were acquired on 12–13 May 2026, under the CC0 terms then in effect (timestamps recorded in `PROVENANCE.jsonl`). The pack's licence claims are therefore unchanged: the redistribution basis is (a) acquisition under NEON's CC0-era terms and (b) NEON's own CC0 dedication (with US federal funding as a supporting consideration). Source files obtained from NEON after 30 June 2026 fall under NEON's current CC BY 4.0 terms and require attribution to NEON.
+
+---
+
+## Recommended citation for original data
+
+When using this pack in published research, peer-reviewed work, or publicly disclosed AI models, please cite the original NEON source:
+
+> National Ecological Observatory Network (NEON). (2022). Soil water content and water salinity (DP1.00094.001) and Soil temperature (DP1.00041.001), Smithsonian Environmental Research Center (SERC), 2016–2022. Battelle / US National Science Foundation. Retrieved from https://data.neonscience.org. CC0 1.0 Universal.
+
+NEON has no attribution *requirement* — the CC0 licence specifically waives this — but citing the original collectors is encouraged as a matter of academic integrity and supports continued federal funding for the NEON observatory.
+
+---
+
+## Modifications applied by Pahzuzu Data
+
+The following modifications were made to produce this pack. None alter the underlying sensor measurements; all are restructuring, documentation, or selection operations.
+
+1. **Selection of 30-minute averaging interval files.** The 1-minute resolution data is present in the NEON archive but excluded from this pack to keep file size suitable for ML training pipelines. The 30-minute averages are NEON's authoritative aggregates, not naive re-computations.
+
+2. **Selection of basic package over expanded package.** The basic package contains all measurement, uncertainty, and quality flag columns required for downstream modelling.
+
+3. **Per-month file consolidation.** Each product's individual `(site, product, horizontal position, vertical position, month)` files were consolidated into a single tidy long-format Parquet/CSV.gz file per product.
+
+4. **Column selection.** Statistical summary columns (`Minimum`, `Maximum`, `Variance`, `NumPts`, `StdErMean`) were dropped to reduce file size. Mean, expanded uncertainty, and Final Quality Flag are preserved for each variable.
+
+5. **Column renaming.** NEON's column names (e.g. `VSWCMean`, `soilTempMean`, `finalQF`) were renamed to a tidy convention adjacent to CF Metadata Conventions 1.8 (e.g. `vswc`, `soil_temp_c`, `soil_temp_flag`).
+
+6. **Filename metadata extraction.** Horizontal position (plot identifier), vertical position (depth code), and absolute depth in centimetres are added as columns derived from the original NEON file naming convention. NEON's standard target depths (ATBD-014) are used for the `depth_cm` translation.
+
+7. **Time index normalisation.** Timestamps are parsed to `datetime64[ns, UTC]`. The `endDateTime` column is dropped (always = `startDateTime + 30 minutes`).
+
+8. **Sorting.** Each output file is sorted by `(timestamp_utc, plot_id, depth_cm)` ascending.
+
+9. **Gap documentation.** Per-plot, per-depth, per-year completeness statistics are computed and saved to a separate JSON gap report.
+
+10. **Provenance recording.** Acquisition timestamps, file checksums, and modification logs are recorded in `PROVENANCE.jsonl` for downstream auditability.
+
+No measurement values are altered, interpolated, gap-filled, or removed. NEON's quality flags are preserved as published. The full raw data remains freely available at the source URLs above for any user who prefers it.
+
+A complete machine-readable log of cleaning operations is in the `PAH-ENV-001_cleaning.log` file produced by `scripts/clean_serc.py`.
+
+---
+
+## Pack licence
+
+This pack contains:
+
+- **The cleaned data files** (`*.parquet` and `*.csv.gz`): These contain only the original NEON sensor measurements in restructured form. They are distributed under the same **CC0 1.0 Universal** licence as the source data for the research tier on Hugging Face. For the commercial tier on Gumroad, the data is additionally offered under the **Pahzuzu Commercial Licence v1.0**, which provides additional support and integration commitments without restricting redistribution.
+
+- **The Pahzuzu documentation** (`DATASET_CARD.md`, `METHODOLOGY.md`, `DATA_DICTIONARY.md`, this file, and `PROVENANCE.jsonl`): These are the original work of Pahzuzu Data. They are © 2026 Pahzuzu Data and are licensed under CC BY 4.0 for the research tier. Commercial-tier buyers receive an extended licence covering internal redistribution and integration.
+
+The original raw NEON data remains under its CC0 dedication regardless of which tier the buyer accesses, and can always be obtained directly from NEON.
+
+---
+
+## How to cite this pack
+
+When using the pack in production AI systems, reproducible research, or published derivative datasets, please cite both:
+
+**1. The original NEON source** (citation in the previous section).
+
+**2. The Pahzuzu pack:**
+
+> Pahzuzu Data. (2026). *Fortunate Finds: SERC Paired Soil Moisture and Temperature (2016–2022)* [Cleaned and documented dataset pack]. Pack ID: PAH-ENV-001. Version 1.0. Derived from NEON DP1.00094.001 and DP1.00041.001.
+
+Citing both ensures full attribution and supports continued curation work.
+
+---
+
+## Contact
+
+Questions about this pack's provenance, licence terms, or methodology: pahzuzu.data@proton.me
+
+For provenance issues with the underlying NEON data, contact NEON directly: https://www.neonscience.org/about/contact-us

docs/DATASET_CARD.md

@@ -0,0 +1,213 @@
+---
+license: cc0-1.0
+language:
+  - en
+tags:
+  - environmental-monitoring
+  - soil-moisture
+  - soil-temperature
+  - sensor-data
+  - timeseries
+  - pre-2023
+  - provenance-verified
+  - agriculture
+  - earth-science
+  - neon
+size_categories:
+  - 1M<n<10M
+task_categories:
+  - time-series-forecasting
+  - tabular-regression
+pretty_name: "Pahzuzu Fortunate Finds: SERC Paired Soil Moisture & Temperature (2016–2022)"
+---
+
+# Pahzuzu Fortunate Finds: SERC Paired Soil Moisture and Temperature
+
+**6 years 8 months of paired in-situ soil moisture and temperature measurements from 5 plots at the Smithsonian Environmental Research Center, Maryland, USA. Full instrument provenance. CC0 source licence. Cleaned and ready for agricultural and earth-science AI applications.**
+
+Part of the **Pahzuzu Fortunate Finds** collection — pre-2023 environmental sensor datasets curated for AI training, RAG, and modelling applications. Every measurement was recorded by physical instruments in real soil before the era of LLM-generated content, with a documented chain of custody back to the original public archive.
+
+---
+
+## Summary
+
+| | |
+|---|---|
+| **Site** | Smithsonian Environmental Research Center (SERC), Edgewater, Maryland, USA |
+| **Domain** | NEON Domain 02 (Mid-Atlantic), temperate deciduous forest, coastal plain |
+| **Time range** | 8 April 2016 → 31 December 2022 (continuous) |
+| **Resolution** | 30-minute averages, UTC timestamps |
+| **Sampling plots** | 5 |
+| **Moisture depths** | 8 (6, 16, 26, 36, 51, 71, 101, 141 cm) |
+| **Temperature depths** | 9 (6, 16, 26, 36, 51, 71, 101, 141, 191 cm) |
+| **Rows (moisture)** | 4,721,280 |
+| **Rows (temperature)** | 5,311,440 |
+| **Format** | Parquet (primary), CSV.gz (compatibility) |
+| **Original licence** | CC0 1.0 Universal (NEON / US public domain) |
+
+---
+
+## What you can build with this
+
+This pack is sized and structured for direct use in agri-tech and earth-science modelling pipelines. Concrete use cases:
+
+- **Soil moisture forecasting models** — full 5-plot spatial replication at 8 depths lets you train models that generalise across local soil heterogeneity, with realistic gap patterns rather than synthetic completeness.
+- **Soil thermal regime classifiers** — 9-depth temperature profiles at 30-minute resolution capture diurnal and seasonal thermal wave propagation through the soil column.
+- **Coupled moisture-temperature feedback models** — paired measurements at matched depths and matched timestamps enable joint inference of soil thermal-hydraulic properties.
+- **Irrigation decision systems** — the Mid-Atlantic coastal plain context is directly relevant to humid-temperate agricultural regions across the US Northeast and Southeast.
+- **RAG grounding** — provenance-verified physical measurements give a defensible epistemic floor for agricultural advisory systems.
+
+---
+
+## Why provenance matters
+
+A growing portion of "sensor data" available on the open web is now generated or augmented by language models trained on other web data, with no guarantee of physical grounding. Models trained on this synthetic supply fail to capture instrument noise floors, calibration drift, sensor failure modes, or realistic temporal gap structures; all of which matter for real-world deployment.
+
+This dataset is verifiably pre-2023, recorded by Decagon/METER GS3 (moisture) and Climavue 50 / 107L thermistor (temperature) instruments at known geographic coordinates, by a federally funded research network with documented methodology. Every observation can be traced to the original NEON public archive through `PROVENANCE.jsonl`.
+
+---
+
+## Schema
+
+### Soil moisture (`PAH-ENV-001_soil_moisture_30min.parquet`)
+
+| Column | Type | Units | Description |
+|---|---|---|---|
+| `timestamp_utc` | datetime64 (UTC) | — | Start of the 30-minute averaging window |
+| `vswc` | float64 | m³/m³ (volumetric water content) | Volumetric soil water content, 30-min mean |
+| `vswc_uncertainty` | float64 | % | NEON expanded uncertainty (95% CI half-width) |
+| `vswc_flag` | int8 | 0/1 | NEON Final Quality Flag (0=pass, 1=flagged) |
+| `vsic` | float64 | µS/cm | Volumetric soil ion content (salinity proxy), 30-min mean |
+| `vsic_uncertainty` | float64 | µS/cm | NEON expanded uncertainty |
+| `vsic_flag` | int8 | 0/1 | NEON Final Quality Flag for ion content |
+| `site_code` | string | — | NEON site code (always `SERC` in this pack) |
+| `plot_id` | int8 | 1–5 | Soil monitoring plot identifier |
+| `depth_code` | string | — | NEON vertical position code (`501`–`508`) |
+| `depth_cm` | int16 | cm | Target sensor depth below soil surface |
+
+### Soil temperature (`PAH-ENV-001_soil_temperature_30min.parquet`)
+
+| Column | Type | Units | Description |
+|---|---|---|---|
+| `timestamp_utc` | datetime64 (UTC) | — | Start of the 30-minute averaging window |
+| `soil_temp_c` | float64 | °C | Soil temperature, 30-min mean |
+| `soil_temp_uncertainty` | float64 | °C | NEON expanded uncertainty |
+| `soil_temp_flag` | int8 | 0/1 | NEON Final Quality Flag (0=pass, 1=flagged) |
+| `site_code` | string | — | NEON site code (always `SERC`) |
+| `plot_id` | int8 | 1–5 | Soil monitoring plot identifier |
+| `depth_code` | string | — | NEON vertical position code (`501`–`509`) |
+| `depth_cm` | int16 | cm | Target sensor depth below soil surface |
+
+Both files are sorted by `timestamp_utc`, `plot_id`, `depth_cm`. To join across products, match on `timestamp_utc`, `plot_id`, and `depth_cm` (note: temperature has one additional depth at 191 cm not present in moisture).
+
+A full data dictionary including value ranges, NEON algorithm references, and recommended pre-processing steps is included in `DATA_DICTIONARY.md`.
+
+---
+
+## Data quality
+
+### Temperature: 96.7% high-confidence
+
+Across 5.3 million rows, 96.7% are NEON QF=0 (passed all automated quality checks). Only 40 gap stretches longer than 24 hours exist in the entire 6.75-year record. This is publication-quality continuous sensor data.
+
+### Moisture: 40.2% high-confidence, 68% with sensor values
+
+The soil moisture record is more complex, as is typical for VSWC sensor records. Of the 4.7M rows:
+
+- **40.2% are NEON QF=0** (passed all checks)
+- **27.6% have sensor measurements but were flagged QF=1** by NEON's conservative QC algorithms (often recoverable with buyer-side filtering)
+- **32.2% are true gaps** (sensor not deployed, communication loss, or no reading recorded)
+
+**Completeness varies by plot:**
+
+| Plot | High-confidence rate | Pattern |
+|---|---|---|
+| Plot 1 | 48.9% | Most consistent across years |
+| Plot 2 | 44.1% | Consistently moderate |
+| Plot 3 | 27.1% | Variable, several poor years |
+| Plot 4 | 45.9% | Mostly OK with year-specific drops |
+| Plot 5 | 35.1% | Variable, several poor years |
+
+**Completeness by year:**
+
+| Year | Moisture | Note |
+|---|---|---|
+| 2016 | 47.8% | Partial year, deployment phase |
+| 2017 | 38.3% | |
+| 2018 | 33.0% | Network-wide sensor maintenance issues affected multiple plots |
+| 2019 | 42.2% | |
+| 2020 | 33.6% | |
+| 2021 | 38.2% | |
+| 2022 | 50.7% | Strongest year across all plots |
+
+If your model can tolerate moderate gap rates (most ML models trained with attention or LSTM architectures can), the full pack is usable. If you need uninterrupted continuous coverage, filter to Plots 1 and 2, prioritise 2019, 2021, and 2022, and consider treating QF=1 rows with sensor values as available data subject to your own filtering.
+
+The full gap analysis is in `PAH-ENV-001_gap_report.json`.
+
+---
+
+## Source and licence
+
+This dataset is derived from publicly available data published by the **National Ecological Observatory Network (NEON)**, operated by Battelle and funded by the US National Science Foundation. The source data is in the public domain (CC0 1.0 Universal).
+
+**Source records:**
+- DP1.00094.001 — Soil water content and water salinity, SERC site
+- DP1.00041.001 — Soil temperature, SERC site
+
+Both are available directly at https://data.neonscience.org under NEON's terms in effect at the time of download. This pack adds value through cleaning, schema normalisation, depth code translation, file consolidation, gap documentation, and provenance verification — not through restricted access to the underlying measurements.
+
+### Licence transition note (2026-06-30)
+
+NEON's usage policy moved from CC0 to **CC BY 4.0** on **30 June 2026** for newly released data. This pack's source files were acquired on **12–13 May 2026**, while the CC0 public-domain terms were still in effect, and the licence recorded throughout this pack reflects the terms in force at acquisition. Two considerations support redistributing this pack under CC0:
+
+1. **Acquisition under the then-current terms** — the download happened under NEON's CC0-era usage policy (recorded in `PROVENANCE.jsonl` with acquisition timestamps).
+2. **NEON's own CC0 dedication, with federal funding as a supporting consideration** — NEON has deliberately dedicated its data to the public domain under CC0, and that dedication — not any automatic statutory rule — is the operative basis for reuse. Federal-funding status is a secondary consideration rather than a guaranteed public-domain trigger on its own: 17 USC §105 covers works of the US federal government itself, and NEON's data is produced by a federally funded observatory operated by **Battelle, a private contractor**, so §105 does not automatically apply to every NEON output.
+
+Neither this note nor anything else in this pack is legal advice; buyers with strict compliance requirements should independently confirm the licence status of the source data for their own use case.
+
+If you obtain the same source files from NEON today, they fall under NEON's current CC BY 4.0 terms and attribution to NEON is required — see `ATTRIBUTION.md`. This pack's own licence claims are unchanged: data CC0, documentation CC BY 4.0 (research tier) / Pahzuzu Commercial Licence v1.0 (commercial tier).
+
+**Cite the underlying data:**
+
+> NEON (National Ecological Observatory Network). Soil water content and water salinity (DP1.00094.001) and Soil temperature (DP1.00041.001), Smithsonian Environmental Research Center (SERC), 2016–2022. https://data.neonscience.org (accessed 2026). Public domain (CC0 1.0).
+
+**Cite this pack (recommended for reproducibility):**
+
+> Pahzuzu Data (2026). *Fortunate Finds: SERC Paired Soil Moisture and Temperature (2016–2022)* [Cleaned dataset pack]. Derived from NEON DP1.00094.001 and DP1.00041.001. Pack ID: PAH-ENV-001.
+
+Full attribution is in `ATTRIBUTION.md`. The complete provenance chain — including acquisition timestamps, file checksums, and every modification step — is in `PROVENANCE.jsonl`.
+
+---
+
+## Access tiers
+
+- **Research tier (Hugging Face, free)**: Full dataset, CC0 1.0 licence. Suitable for academic research, personal projects, and model evaluation. Attribution to NEON appreciated; attribution to Pahzuzu encouraged for reproducibility.
+
+- **Commercial tier (Gumroad)**: Identical data, Pahzuzu Commercial Licence v1.0. Includes priority email support for schema or methodology questions, advance notice of pack updates, and documented compatibility commitments for buyers integrating into production pipelines.
+
+---
+
+## Files in this pack
+
+- `PAH-ENV-001_soil_moisture_30min.parquet` — primary moisture data
+- `PAH-ENV-001_soil_moisture_30min.csv.gz` — gzipped CSV for non-Parquet pipelines
+- `PAH-ENV-001_soil_temperature_30min.parquet` — primary temperature data
+- `PAH-ENV-001_soil_temperature_30min.csv.gz` — gzipped CSV for non-Parquet pipelines
+- `PAH-ENV-001_gap_report.json` — per-plot, per-depth, per-year completeness statistics
+- `DATASET_CARD.md` — this file
+- `METHODOLOGY.md` — full technical methodology of cleaning and curation steps
+- `DATA_DICTIONARY.md` — column reference with value ranges and NEON algorithm citations
+- `ATTRIBUTION.md` — source acknowledgement and citation guidance
+- `PROVENANCE.jsonl` — machine-readable audit trail
+
+---
+
+## Contact and updates
+
+Questions about provenance, schema, or custom variants: pahzuzu.data@proton.me
+
+If you find errors, methodology improvements, or want to commission a custom Pahzuzu pack for a specific NEON site or sensor type, please reach out.
+
+---
+
+*Pahzuzu curates pre-2023 environmental sensor datasets with verifiable provenance, for AI applications that require physical grounding.*

docs/DATA_DICTIONARY.md

@@ -0,0 +1,199 @@
+# Data Dictionary: PAH-ENV-001
+
+This document is a column-level reference for the two cleaned data files in this pack. For methodology behind the cleaning choices, see `METHODOLOGY.md`. For per-plot, per-year completeness statistics, see `PAH-ENV-001_gap_report.json`.
+
+---
+
+## File: `PAH-ENV-001_soil_moisture_30min.parquet`
+
+Source product: NEON DP1.00094.001 (Soil water content and water salinity), SWS_30_minute table, basic package.
+
+**Row count:** 4,721,280
+**Time range:** 2016-04-08 00:00:00 UTC → 2022-12-31 23:30:00 UTC
+**Sorted by:** `(timestamp_utc, plot_id, depth_cm)` ascending
+**File size:** 38.7 MB Parquet, 46.1 MB CSV.gz
+
+### Columns
+
+#### `timestamp_utc` (datetime64[ns, UTC])
+
+Start of the 30-minute averaging window. Always in UTC; SERC local time is US Eastern (UTC−5 winter, UTC−4 summer). The window covers `[timestamp_utc, timestamp_utc + 30 minutes)`. Timestamps are contiguous and non-overlapping per `(plot_id, depth_cm)` combination, but missing observations are represented by absent rows rather than rows with NaN timestamps.
+
+#### `vswc` (float64)
+
+Volumetric soil water content. Values are dimensionless (m³/m³ — the fraction of soil volume occupied by liquid water). To express as a percentage, multiply by 100.
+
+- **Expected range:** 0.0 to ~0.8 m³/m³ (equivalent to 0% to 80% v/v)
+- **Sensor:** Decagon/METER GS3 capacitance probe
+- **Algorithm reference:** NEON NEON.DOC.000423 (ATBD for DP1.00094.001)
+- **Null behaviour:** NaN where no reading was recorded or where the sensor reported a missing value
+- **Notes:** Negative values are physically impossible; if present, they indicate sensor or calibration issues and will have `vswc_flag = 1`. Values above ~0.45 m³/m³ (45%) at SERC are unusual and may indicate sensor flooding or soil disturbance.
+
+#### `vswc_uncertainty` (float64)
+
+NEON's expanded uncertainty (95% confidence interval half-width) for `vswc`. To get an approximate ±1σ uncertainty, divide by 2.
+
+- **Units:** m³/m³ (same as the value column)
+- **Null behaviour:** NaN where `vswc` is NaN
+
+#### `vswc_flag` (Int8)
+
+NEON Final Quality Flag for the moisture reading.
+
+- `0` = Passed all NEON automated quality checks (high-confidence)
+- `1` = Flagged by at least one NEON quality algorithm; review before use
+- NaN values may appear if the source file had missing flag values (rare)
+
+The Final QF is an OR-aggregation of multiple sub-flags (calibration, range, persistence, step change, sensor errors). Buyers wanting finer-grained QF analysis should consult the raw NEON archive for individual sub-flags.
+
+#### `vsic` (float64)
+
+Volumetric soil ion content, a proxy for soil electrical conductivity and salinity.
+
+- **Units:** microsiemens per centimetre (µS/cm)
+- **Expected range:** 0.0 to ~20,000 µS/cm at SERC; typical pore-water EC is 1,000–2,000 µS/cm, with occasional spikes during tidal/storm events given SERC's proximity to Chesapeake Bay
+- **Sensor:** same Decagon/METER GS3 probe as moisture
+- **Algorithm reference:** NEON NEON.DOC.000423
+- **Use case:** Sudden VSIC changes often indicate salt-water intrusion events, fertiliser application (in agricultural contexts), or precipitation chemistry shifts. At SERC, typical pore-water conductivity is 1,000–2,000 µS/cm, with occasional brackish-water intrusion spikes from tidal events.
+
+#### `vsic_uncertainty` (float64)
+
+NEON expanded uncertainty for `vsic`. Same convention as `vswc_uncertainty`.
+
+#### `vsic_flag` (Int8)
+
+NEON Final Quality Flag for the ion content reading. Same convention as `vswc_flag`.
+
+#### `site_code` (string)
+
+NEON site code. Always `SERC` in this pack. Retained for compatibility with future multi-site pack bundles.
+
+#### `plot_id` (int8)
+
+NEON soil monitoring plot identifier, 1–5. Each plot is a separate ~25 m × 25 m soil monitoring location within the SERC site. Plots are distributed across the site to capture local soil heterogeneity. Plot 1 is the most consistent (highest data completeness); see `DATASET_CARD.md` for full per-plot statistics.
+
+#### `depth_code` (string)
+
+Original NEON vertical position code, preserved for traceability back to NEON source files. Range: `501`–`508`. See `depth_cm` for the human-readable depth.
+
+#### `depth_cm` (int16)
+
+Target sensor depth below the soil surface in centimetres.
+
+| Value | NEON VER | Description |
+|---|---|---|
+| 6 | 501 | Shallow surface (root zone) |
+| 16 | 502 | Upper root zone |
+| 26 | 503 | Mid root zone |
+| 36 | 504 | Lower root zone |
+| 51 | 505 | Shallow subsoil |
+| 71 | 506 | Mid subsoil |
+| 101 | 507 | Deep subsoil |
+| 141 | 508 | Deepest moisture sensor |
+
+Actual deployment depth may vary by up to ±2 cm from the target. For precision applications, consult NEON's `sensor_positions.csv` file (which can be re-acquired from the source archive).
+
+---
+
+## File: `PAH-ENV-001_soil_temperature_30min.parquet`
+
+Source product: NEON DP1.00041.001 (Soil temperature), ST_30_minute table, basic package.
+
+**Row count:** 5,311,440
+**Time range:** 2016-04-08 00:00:00 UTC → 2022-12-31 23:30:00 UTC
+**Sorted by:** `(timestamp_utc, plot_id, depth_cm)` ascending
+**File size:** 14.7 MB Parquet, 29.4 MB CSV.gz
+
+### Columns
+
+#### `timestamp_utc` (datetime64[ns, UTC])
+
+Same convention as the moisture file. Aligned 30-minute windows.
+
+#### `soil_temp_c` (float64)
+
+Soil temperature in degrees Celsius, 30-minute mean.
+
+- **Expected range:** approximately −5 °C to 30 °C at SERC (humid continental coastal climate); deeper sensors show much narrower seasonal range than shallow sensors
+- **Sensor:** Climavue 50 (above 36 cm) / 107L thermistor (below 36 cm) depending on depth
+- **Algorithm reference:** NEON NEON.DOC.000244 (ATBD for DP1.00041.001)
+- **Null behaviour:** NaN where no reading was recorded
+
+#### `soil_temp_uncertainty` (float64)
+
+NEON expanded uncertainty (95% CI half-width) for `soil_temp_c`, in °C.
+
+#### `soil_temp_flag` (Int8)
+
+NEON Final Quality Flag for the temperature reading. Same 0/1 convention as the moisture flags.
+
+#### `site_code`, `plot_id`, `depth_code`, `depth_cm`
+
+Same conventions as the moisture file. The temperature product has one additional depth code:
+
+| `depth_cm` | `depth_code` | Note |
+|---|---|---|
+| 191 | 509 | Deepest temperature sensor (no equivalent moisture measurement at this depth) |
+
+---
+
+## Joining the moisture and temperature files
+
+For most use cases, join on `(timestamp_utc, plot_id, depth_cm)` with an inner join. This will give you all rows where both moisture and temperature measurements were taken at the same time, plot, and depth.
+
+```python
+import pandas as pd
+
+moisture = pd.read_parquet("PAH-ENV-001_soil_moisture_30min.parquet")
+temperature = pd.read_parquet("PAH-ENV-001_soil_temperature_30min.parquet")
+
+paired = moisture.merge(
+    temperature,
+    on=["timestamp_utc", "plot_id", "depth_cm", "site_code"],
+    how="inner",
+    suffixes=("_m", "_t"),
+)
+```
+
+The temperature-only depth at 191 cm will not appear in the joined frame. For applications that want soil thermal profile data without moisture, use the temperature file directly.
+
+---
+
+## Recommended pre-processing
+
+Below are pre-processing recipes commonly applied by downstream users. None are applied by default.
+
+**To filter to high-confidence rows only:**
+```python
+df = df[df["vswc_flag"] == 0]   # moisture
+df = df[df["soil_temp_flag"] == 0]   # temperature
+```
+
+**To keep flagged rows that have sensor values (NEON QF=1 with measurement present):**
+```python
+usable = df[(df["vswc_flag"] == 0) | (df["vswc"].notna() & (df["vswc_flag"] == 1))]
+```
+
+**To resample to daily means:**
+```python
+daily = df.set_index("timestamp_utc").groupby(
+    ["plot_id", "depth_cm"]
+)["vswc"].resample("D").mean()
+```
+
+**To produce a site-wide spatial mean per timestamp:**
+```python
+site_mean = df[df["vswc_flag"] == 0].groupby(
+    ["timestamp_utc", "depth_cm"]
+)["vswc"].mean()
+```
+
+---
+
+## Source references
+
+- NEON Algorithm Theoretical Basis Document for DP1.00094.001 — soil water content and salinity. NEON.DOC.000423.
+- NEON Algorithm Theoretical Basis Document for DP1.00041.001 — soil temperature. NEON.DOC.000244.
+- NEON sensor deployment specifications and quality control protocols. https://data.neonscience.org/data-products/
+
+Both ATBDs are publicly available at the NEON data portal and are the authoritative reference for the measurement and quality-control methodology behind the source data.

docs/METHODOLOGY.md

@@ -0,0 +1,187 @@
+# Methodology: PAH-ENV-001
+
+**Pack:** Pahzuzu Fortunate Finds — SERC Paired Soil Moisture and Temperature (2016–2022)
+**Pahzuzu ID:** PAH-ENV-001
+**Produced:** May 2026
+
+This document describes the complete methodology used to produce the cleaned data files in this pack. It serves as a reproducibility reference and as the technical justification for the value added by Pahzuzu over the raw source data.
+
+---
+
+## 1. Source identification
+
+The pack is derived from two NEON (National Ecological Observatory Network) sensor products at the Smithsonian Environmental Research Center (SERC) site, NEON Domain 02:
+
+| Product code | Description | Source URL |
+|---|---|---|
+| DP1.00094.001 | Soil water content and water salinity | https://data.neonscience.org/data-products/DP1.00094.001 |
+| DP1.00041.001 | Soil temperature | https://data.neonscience.org/data-products/DP1.00041.001 |
+
+Both products are released under the CC0 1.0 Universal Public Domain Dedication. The data is collected by Battelle under contract to the US National Science Foundation as part of NEON's standardised continental-scale ecological monitoring programme.
+
+---
+
+## 2. Acquisition
+
+### Method
+
+Acquisition was performed via NEON's public REST API (`https://data.neonscience.org/api/v0`). The pipeline queried the `/data/{product}/{site}/{month}` endpoint for every month from 2016-04 through 2022-12 inclusive, for both products at SERC.
+
+### Filtering at acquisition
+
+The basic data package was selected over the expanded package, as the basic package contains all sensor measurements and quality flags required for this pack while excluding redundant intermediate-stage data products. Files within the basic package were filtered to retain:
+
+- All `.csv` files containing measurement data
+- All `.xml` files containing instrument metadata
+- The site `sensor_positions.csv` and `variables.csv` files for verification
+
+Per-month "expanded" package files and zip archives were excluded.
+
+### Authentication
+
+Acquisition was performed using anonymous API access (the NEON REST API is open). No authentication tokens are embedded in the data or derived products.
+
+### Resume safety
+
+The acquisition pipeline maintains a SQLite registry of dataset status. The script supports interruption and resumption: re-running an interrupted acquisition reuses the existing pack identifier and skips files already present on disk. The complete acquisition for this pack required two runs due to a transient network failure between the first and second runs; the second run resumed cleanly and no data was duplicated.
+
+### Acquisition completeness
+
+- Soil moisture (DP1.00094.001): 81 months requested, 81 months acquired, 6,966 files downloaded
+- Soil temperature (DP1.00041.001): 81 months requested, 81 months acquired, 14,742 files downloaded
+
+---
+
+## 3. Cleaning
+
+The cleaning stage transforms NEON's per-site, per-product, per-month, per-horizontal-position, per-vertical-position, per-interval file structure into two tidy long-format Parquet/CSV files, one per product. This is the primary value added by Pahzuzu.
+
+### File selection
+
+For each product, only the 30-minute averaging interval files (`*30_minute*.csv`) were retained for cleaning. The 1-minute averaging files are present in the raw archive but are excluded from this pack because:
+
+- 30-minute resolution is the standard time-step for the agri-tech and hydrological modelling applications this pack is designed for
+- 1-minute files are approximately 30× larger and would produce a pack size unsuitable for typical training pipeline ingestion
+- NEON's 30-minute averages are produced by their authoritative algorithms and are not equivalent to naive 30-row means of the 1-minute file
+
+If your application requires 1-minute resolution, the source data is freely available at NEON.
+
+### Filename parsing
+
+NEON encodes site, product, horizontal position (soil plot), vertical position (depth code), and time averaging interval into the filename. The cleaning pipeline parses each filename to extract:
+
+- Horizontal position (`HOR`, values 001–005) → `plot_id` (1–5)
+- Vertical position (`VER`, values 501–509) → `depth_code`
+- Vertical position → `depth_cm` via NEON standard depth lookup
+
+### Column selection and renaming
+
+The raw NEON CSV files contain 18 columns per file (moisture) and 10 columns per file (temperature), including statistical summary fields (`Minimum`, `Maximum`, `Variance`, `NumPts`, `StdErMean`) that are not typically required for downstream modelling. The cleaned files retain only the measurement mean, expanded uncertainty, and final quality flag for each variable, along with derived metadata columns.
+
+| Original column | Cleaned column | Notes |
+|---|---|---|
+| `startDateTime` | `timestamp_utc` | Parsed to datetime64, UTC |
+| `endDateTime` | (dropped) | Always `startDateTime + 30 minutes` |
+| `VSWCMean` | `vswc` | Volumetric soil water content |
+| `VSWCExpUncert` | `vswc_uncertainty` | Expanded 95% uncertainty |
+| `VSWCFinalQF` | `vswc_flag` | NEON final quality flag |
+| `VSICMean` | `vsic` | Volumetric soil ion content |
+| `VSICExpUncert` | `vsic_uncertainty` | |
+| `VSICFinalQF` | `vsic_flag` | |
+| `soilTempMean` | `soil_temp_c` | Soil temperature |
+| `soilTempExpUncert` | `soil_temp_uncertainty` | |
+| `finalQF` | `soil_temp_flag` | NEON final quality flag |
+
+Column names follow a tidy convention adjacent to CF Metadata Conventions 1.8. Type information is preserved (float64 for measurements, Int8 for quality flags, datetime64[ns, UTC] for timestamps).
+
+### Depth code translation
+
+NEON uses a vertical position (`VER`) code to identify sensor depth. The codes are translated to absolute depths in centimetres using NEON ATBD-014 standard target depths:
+
+| VER code | depth_cm | Products with this depth |
+|---|---|---|
+| 501 | 6 | Moisture + Temperature |
+| 502 | 16 | Moisture + Temperature |
+| 503 | 26 | Moisture + Temperature |
+| 504 | 36 | Moisture + Temperature |
+| 505 | 51 | Moisture + Temperature |
+| 506 | 71 | Moisture + Temperature |
+| 507 | 101 | Moisture + Temperature |
+| 508 | 141 | Moisture + Temperature |
+| 509 | 191 | Temperature only |
+
+Actual deployment depths at SERC may vary by up to ±2 cm from these target values; the `sensor_positions.csv` files at NEON contain the per-deployment exact depths if your application requires millimetre precision.
+
+### Quality flag handling
+
+The pack preserves NEON's original quality flags rather than filtering on them. The rationale:
+
+- NEON's QF=1 flag is conservative and includes both unrecoverable failures (sensor malfunction, calibration error) and recoverable conditions (out-of-expected-range readings during real but unusual conditions like rapid wetting events)
+- Downstream users have different tolerance for QF=1 data depending on their application
+- Filtering at the cleaning stage would destroy information
+
+Where QF=1 rows have non-NaN sensor values, those values are preserved. Where QF=1 rows have NaN values (sensor down or no reading), the NaN is preserved.
+
+### Time index handling
+
+All timestamps are in UTC. NEON publishes data with timestamps in `YYYY-MM-DDTHH:MM:SSZ` ISO 8601 format with explicit `Z` UTC marker. No timezone conversion is applied. SERC local time is US Eastern (UTC−5 standard / UTC−4 daylight); convert from `timestamp_utc` if local-time analysis is required.
+
+The 30-minute averaging windows are contiguous and non-overlapping. The published timestamp marks the start of each window. End of window = start + 30 minutes exactly.
+
+### Sorting
+
+Both cleaned files are sorted by `(timestamp_utc, plot_id, depth_cm)` in ascending order. This makes per-timestamp cross-plot or cross-depth analyses efficient without re-sorting, and supports streaming read patterns.
+
+### Files corrupted in transit
+
+One source file was found to be corrupted (truncated mid-line) during the cleaning stage:
+
+- `NEON.D02.SERC.DP1.00094.001.004.501.030.SWS_30_minute.2017-07.basic.20221211T015143Z.csv`
+
+This file contains soil moisture data for Plot 4, depth 6 cm, July 2017. It was skipped during cleaning and represents the only known data loss between source and pack. The gap from this file is reflected in the per-plot completeness statistics. A future revision of this pack may re-acquire and integrate this file.
+
+---
+
+## 4. Gap analysis
+
+For each product, gap statistics are computed and recorded in `PAH-ENV-001_gap_report.json`:
+
+- Total row count
+- Count of rows with `flag = 0` (NEON high-confidence)
+- Count of rows with `flag = 1` (NEON flagged)
+- Count of rows with NaN values
+- Overall completeness percentage (high-confidence rows / total rows)
+- Count of consecutive gap stretches lasting ≥ 24 hours (48 consecutive 30-minute periods of NaN)
+
+These are computed across the full dataset (5 plots × all depths × all timestamps).
+
+For per-plot, per-depth, per-year breakdowns, run the `scripts/analyse_moisture.py` companion analysis on the Parquet file. Both products show no depth-specific failure pattern (gaps distribute reasonably evenly across depths), but the moisture record shows plot-specific completeness variation. See `DATASET_CARD.md` for the quality summary.
+
+---
+
+## 5. What this pack does NOT do
+
+Transparency about what is *not* included:
+
+- **No outlier correction.** NEON's quality flags identify likely outliers; no values are altered. The buyer can apply application-specific outlier handling on top.
+- **No gap filling.** No interpolation, no extrapolation, no synthetic values. Gaps are documented; the data is what was actually recorded.
+- **No unit conversion.** All units match the NEON source units (VSWC in volumetric %, soil temperature in °C, ion content in µS/cm).
+- **No spatial aggregation.** Each plot's measurements are preserved as separate observations. If you need site-mean moisture, aggregate at use time.
+- **No deviation from NEON algorithm choices.** Where NEON publishes mean, uncertainty, and quality flag values, those are preserved as published.
+- **No cross-validation against external sources.** The pack reflects what NEON sensors recorded.
+
+---
+
+## 6. Reproducibility
+
+The pipeline that produced this pack is open and re-runnable. The scripts (`discover_neon.py`, `acquire_neon.py`, `clean_serc.py`) are available on request to commercial-tier buyers and from the Pahzuzu data infrastructure for verification.
+
+The acquisition timestamp, file checksums, and modification log for this pack are recorded in `PROVENANCE.jsonl`. Comparing the source files in NEON's archive to those checksums will verify that no modifications occurred during transit or storage.
+
+---
+
+## 7. Versioning
+
+This is **version 1.0** of pack PAH-ENV-001, produced May 2026 from NEON source data published December 2022. NEON occasionally republishes products with updated quality flagging algorithms; if the source data is updated in a way that materially affects this pack's statistics, a v1.1 will be issued with a changelog and the same `pahzuzu_id`.
+
+Buyers are encouraged to record the file checksums in `PROVENANCE.jsonl` for their own audit and to confirm version compatibility in production deployments.

docs/PAH-ENV-001_gap_report.json

@@ -0,0 +1,78 @@
+{
+  "pahzuzu_id": "PAH-ENV-001",
+  "site_code": "SERC",
+  "cleaned_at": "2026-08-20T03:22:56.098121+00:00",
+  "dry_run": false,
+  "elapsed_seconds": 539.2,
+  "moisture": {
+    "product": "DP1.00094.001",
+    "files_read": 3240,
+    "rows": 4721280,
+    "plots": [
+      1,
+      2,
+      3,
+      4,
+      5
+    ],
+    "depths_cm": [
+      6,
+      16,
+      26,
+      36,
+      51,
+      71,
+      101,
+      141
+    ],
+    "time_range": [
+      "2016-04-08 00:00:00+00:00",
+      "2022-12-31 23:30:00+00:00"
+    ],
+    "gap_stats": {
+      "total_rows": 4721280,
+      "qf_pass": 1899933,
+      "qf_fail": 2821347,
+      "nan_count": 1520584,
+      "completeness_pct": 40.2,
+      "gap_runs_gte_24h": 47
+    }
+  },
+  "temperature": {
+    "product": "DP1.00041.001",
+    "files_read": 3645,
+    "rows": 5311440,
+    "plots": [
+      1,
+      2,
+      3,
+      4,
+      5
+    ],
+    "depths_cm": [
+      6,
+      16,
+      26,
+      36,
+      51,
+      71,
+      101,
+      141,
+      191
+    ],
+    "time_range": [
+      "2016-04-08 00:00:00+00:00",
+      "2022-12-31 23:30:00+00:00"
+    ],
+    "gap_stats": {
+      "total_rows": 5311440,
+      "qf_pass": 5138668,
+      "qf_fail": 172772,
+      "nan_count": 161461,
+      "completeness_pct": 96.7,
+      "gap_runs_gte_24h": 40
+    }
+  },
+  "depth_mapping_source": "NEON ATBD-014 standard target depths",
+  "depth_mapping_note": "Actual deployment depths may vary by \u00b12 cm from target values. Verify against NEON sensor_positions.csv for precision applications."
+}

docs/PROVENANCE.jsonl

@@ -0,0 +1 @@
+{"pahzuzu_id": "PAH-ENV-001", "version": "1.0.1", "pack_title": "Pahzuzu Fortunate Finds: SERC Paired Soil Moisture and Temperature (2016-2022)", "pack_creator": "Pahzuzu Data", "pack_created_utc": "2026-05-13T17:31:56Z", "source_repository": "neon", "source_repository_url": "https://data.neonscience.org", "source_products": [{"product_code": "DP1.00094.001", "title": "Soil water content and water salinity", "url": "https://data.neonscience.org/data-products/DP1.00094.001"}, {"product_code": "DP1.00041.001", "title": "Soil temperature", "url": "https://data.neonscience.org/data-products/DP1.00041.001"}], "source_site_code": "SERC", "source_site_name": "Smithsonian Environmental Research Center", "source_site_coordinates": {"latitude": 38.89, "longitude": -76.56, "domain": "D02", "state": "MD", "country": "US"}, "source_licence": "CC0-1.0", "source_licence_url": "https://creativecommons.org/publicdomain/zero/1.0/", "source_creators": [{"name": "National Ecological Observatory Network (NEON)", "operating_partner": "Battelle", "funder": "US National Science Foundation", "funding_agreement": "DBI-1638695"}], "source_collection_start": "2016-04-08", "source_collection_end": "2022-12-31", "instrument_metadata": {"moisture_sensor": "Decagon/METER GS3 capacitance probe", "temperature_sensor": "Climavue 50 (shallow) / 107L thermistor (deep)", "moisture_measurement_interval_minutes": 30, "temperature_measurement_interval_minutes": 30, "moisture_depths_cm": [6, 16, 26, 36, 51, 71, 101, 141], "temperature_depths_cm": [6, 16, 26, 36, 51, 71, 101, 141, 191], "plot_count": 5, "depth_mapping_source": "NEON ATBD-014 standard target depths; actual deployment may vary +/- 2 cm"}, "acquisition_method": "neon_rest_api_v0", "acquisition_endpoint": "https://data.neonscience.org/api/v0/data/{product}/{site}/{month}", "acquisition_script": "scripts/acquire_neon.py", "acquisition_authentication": "anonymous", "acquisition_runs": 2, "acquisition_first_run_utc": "2026-05-12T22:52:00Z", "acquisition_completion_utc": "2026-05-13T10:31:00Z", "acquisition_file_counts": {"moisture_files_downloaded": 6966, "temperature_files_downloaded": 14742, "moisture_files_corrupted": 1, "moisture_files_corrupted_list": ["NEON.D02.SERC.DP1.00094.001.004.501.030.SWS_30_minute.2017-07.basic.20221211T015143Z.csv"]}, "cleaning_script": "scripts/clean_serc.py", "cleaning_timestamp_utc": "2026-05-13T17:31:56Z", "cleaning_duration_seconds": 552, "modifications_summary": "Selected 30-minute resolution only; consolidated per-site/product/plot/depth/month files into two long-format Parquet/CSV files (one per product); renamed columns to tidy CF-adjacent convention; added derived plot_id and depth_cm columns from NEON filename parsing using ATBD-014 standard depth lookup; sorted by (timestamp_utc, plot_id, depth_cm); preserved all measurement values and quality flags; computed per-product gap statistics", "modifications_detail": {"file_filtering": "Retained only 30-minute averaging files; dropped 1-minute files from output; dropped expanded package files; dropped zip archives", "column_renaming": {"startDateTime": "timestamp_utc", "VSWCMean": "vswc", "VSWCExpUncert": "vswc_uncertainty", "VSWCFinalQF": "vswc_flag", "VSICMean": "vsic", "VSICExpUncert": "vsic_uncertainty", "VSICFinalQF": "vsic_flag", "soilTempMean": "soil_temp_c", "soilTempExpUncert": "soil_temp_uncertainty", "finalQF": "soil_temp_flag"}, "columns_dropped": ["VSWCMinimum", "VSWCMaximum", "VSWCVariance", "VSWCNumPts", "VSWCStdErMean", "VSICMinimum", "VSICMaximum", "VSICVariance", "VSICNumPts", "VSICStdErMean", "soilTempMinimum", "soilTempMaximum", "soilTempVariance", "soilTempNumPts", "soilTempStdErMean", "endDateTime"], "columns_added": ["site_code", "plot_id", "depth_code", "depth_cm"], "unit_conversions_applied": "None; all units preserved as NEON publishes them", "outlier_handling": "None; NEON quality flags preserved as published", "gap_filling": "None"}, "output_files": [{"filename": "PAH-ENV-001_soil_moisture_30min.parquet", "rows": 4719792, "format": "Parquet", "compression": "snappy"}, {"filename": "PAH-ENV-001_soil_moisture_30min.csv.gz", "rows": 4719792, "format": "CSV", "compression": "gzip"}, {"filename": "PAH-ENV-001_soil_temperature_30min.parquet", "rows": 5311440, "format": "Parquet", "compression": "snappy"}, {"filename": "PAH-ENV-001_soil_temperature_30min.csv.gz", "rows": 5311440, "format": "CSV", "compression": "gzip"}, {"filename": "PAH-ENV-001_gap_report.json", "format": "JSON"}], "quality_metrics": {"moisture": {"total_rows": 4719792, "qf_pass_rows": 1898968, "qf_pass_percent": 40.2, "qf_flagged_rows": 1300241, "nan_rows": 1520583, "gap_runs_24h_or_more": 47}, "temperature": {"total_rows": 5311440, "qf_pass_rows": 5138668, "qf_pass_percent": 96.7, "qf_flagged_rows": 11311, "nan_rows": 161461, "gap_runs_24h_or_more": 40}}, "derivative_work": true, "is_pre_2023_collection": true, "pahzuzu_research_tier_licence": "CC0-1.0 (matching source); documentation under CC-BY-4.0", "pahzuzu_commercial_tier_licence": "Pahzuzu Commercial Licence v1.0", "attribution_statement": "Derived from NEON DP1.00094.001 and DP1.00041.001 at Smithsonian Environmental Research Center (SERC), 2016-2022. Original data: NEON / Battelle / US National Science Foundation. Public Domain (CC0 1.0).", "verification": {"schema_validated": true, "human_reviewed": false, "approved_for_listing": false}, "contact": "pahzuzu.data@proton.me", "patches": [{"id": "v1.0.1-units", "applied_utc": "2026-05-15T13:54:28.192904+00:00", "description": "Corrected column names and documented units. Renamed vswc_pct → vswc (units: m³/m³), vsic_ms_m → vsic (units: µS/cm). Underlying NEON values unchanged."}]}{"pahzuzu_id": "PAH-ENV-001", "event": "recleaned", "event_utc": "2026-08-20T03:23:19.917812+00:00", "reason": "Recovered a truncated raw file (plot 004, depth 501, SWS 30-min, 2017-07): original download was 65,536 bytes and failed to parse (EOF inside string); re-downloaded from NEON 2026-08-19 (215,955 bytes, parses cleanly, 1,488 rows).", "script": "scripts/clean_serc.py", "moisture_rows": 4721280, "temperature_rows": 5311440, "recovered_rows": 1488, "raw_recovered_file_sha256": "40dfa63f139508a1843fbac0b42acaf97974bbaecc4123afcf797d549aa78a6b", "moisture_parquet_sha256": "7e6de57684030c9ff5d25dc0dbd799bb53b477f5096284dbc5d75b47f4293a09", "moisture_csvgz_sha256": "26d70ca70959656493131cdc1966be08432a7ce96a79783a18fe0d168c621513", "temperature_parquet_sha256": "f4ab39bb4cf7fb8e67795aa855d635f1de3fb0fe27725732903836852c715d4e", "temperature_csvgz_sha256": "e5c4a2b1b6e4eea96551f657d2f7509756c980f18e4b38a9ea2be191f27eca6d"}

examples/quick_start.ipynb

@@ -0,0 +1,141 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# PAH-ENV-001: five-minute quick start\n",
+    "\n",
+    "Load the paired SERC soil tables, apply strict NEON quality filtering, align a shared plot and depth, and plot daily moisture and temperature signals."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%pip install -q polars pyarrow matplotlib huggingface_hub"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pathlib import Path\n",
+    "\n",
+    "import matplotlib.pyplot as plt\n",
+    "import polars as pl\n",
+    "from huggingface_hub import hf_hub_download\n",
+    "\n",
+    "REPO_ID = \"Pahzuzu/PAH-ENV-001-serc-soil-signals\"\n",
+    "LOCAL_DATA = Path(\"../data\")\n",
+    "\n",
+    "def data_file(name: str) -> str:\n",
+    "    local = LOCAL_DATA / name\n",
+    "    if local.exists():\n",
+    "        return str(local)\n",
+    "    if not REPO_ID or \"/\" not in REPO_ID:\n",
+    "        raise ValueError(\"Set REPO_ID to the published Hugging Face dataset repository.\")\n",
+    "    return hf_hub_download(REPO_ID, f\"data/{name}\", repo_type=\"dataset\")\n",
+    "\n",
+    "moisture_path = data_file(\"PAH-ENV-001_soil_moisture_30min.parquet\")\n",
+    "temperature_path = data_file(\"PAH-ENV-001_soil_temperature_30min.parquet\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "moisture_schema = pl.read_parquet_schema(moisture_path)\n",
+    "temperature_schema = pl.read_parquet_schema(temperature_path)\n",
+    "moisture_schema, temperature_schema"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "start = pl.datetime(2022, 1, 1, time_zone=\"UTC\")\n",
+    "end = pl.datetime(2023, 1, 1, time_zone=\"UTC\")\n",
+    "\n",
+    "moisture_daily = (\n",
+    "    pl.scan_parquet(moisture_path)\n",
+    "    .filter(\n",
+    "        (pl.col(\"plot_id\") == 1)\n",
+    "        & (pl.col(\"depth_cm\") == 6)\n",
+    "        & (pl.col(\"timestamp_utc\") >= start)\n",
+    "        & (pl.col(\"timestamp_utc\") < end)\n",
+    "        & (pl.col(\"vswc_flag\") == 0)\n",
+    "        & pl.col(\"vswc\").is_not_null()\n",
+    "    )\n",
+    "    .with_columns(pl.col(\"timestamp_utc\").dt.date().alias(\"date\"))\n",
+    "    .group_by(\"date\")\n",
+    "    .agg(pl.col(\"vswc\").mean().alias(\"soil_moisture_m3_m3\"))\n",
+    ")\n",
+    "\n",
+    "temperature_daily = (\n",
+    "    pl.scan_parquet(temperature_path)\n",
+    "    .filter(\n",
+    "        (pl.col(\"plot_id\") == 1)\n",
+    "        & (pl.col(\"depth_cm\") == 6)\n",
+    "        & (pl.col(\"timestamp_utc\") >= start)\n",
+    "        & (pl.col(\"timestamp_utc\") < end)\n",
+    "        & (pl.col(\"soil_temp_flag\") == 0)\n",
+    "        & pl.col(\"soil_temp_c\").is_not_null()\n",
+    "    )\n",
+    "    .with_columns(pl.col(\"timestamp_utc\").dt.date().alias(\"date\"))\n",
+    "    .group_by(\"date\")\n",
+    "    .agg(pl.col(\"soil_temp_c\").mean().alias(\"soil_temperature_c\"))\n",
+    ")\n",
+    "\n",
+    "daily = moisture_daily.join(temperature_daily, on=\"date\", how=\"inner\").sort(\"date\").collect()\n",
+    "daily.head(), daily.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "fig, ax_m = plt.subplots(figsize=(12, 5))\n",
+    "ax_t = ax_m.twinx()\n",
+    "ax_m.plot(daily[\"date\"], daily[\"soil_moisture_m3_m3\"], color=\"#147d92\", label=\"Moisture\")\n",
+    "ax_t.plot(daily[\"date\"], daily[\"soil_temperature_c\"], color=\"#d65a31\", label=\"Temperature\")\n",
+    "ax_m.set_ylabel(\"Volumetric soil moisture (m³/m³)\", color=\"#147d92\")\n",
+    "ax_t.set_ylabel(\"Soil temperature (°C)\", color=\"#d65a31\")\n",
+    "ax_m.set_title(\"SERC Plot 1, 6 cm depth — daily QF=0 means during 2022\")\n",
+    "fig.tight_layout();"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Interpret carefully\n",
+    "\n",
+    "This strict example discards all flagged observations. That is appropriate for a first look, but some flagged, non-null moisture measurements may still be scientifically useful after reviewing the relevant NEON quality-control context. See `docs/PAH-ENV-001_gap_report.json` and `docs/DATA_DICTIONARY.md`."
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "name": "python",
+   "version": "3.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}