Hydrology, soils and watershed data for the Nanyuki River watershed, Laikipia, Meru, and Nyeri counties, Kenya (2021–2024)

Field observations from the northwestern slopes of Mount Kenya, at the edge of the savanna:

• Streamflow: stage at eight gauges (CA01–CA08) and discharge with 95% uncertainty bounds at seven of them (CA02–CA08), at 5-minute, hourly and daily time steps, July 2021 – January 2024.
• Rainfall: five tipping-bucket gauges (RN01–RN05) at the logging interval, hourly and daily, with lower and upper uncertainty bounds, July/August 2021 – January 2024.
• Soils: physical, hydraulic and chemical properties of 43 fields on 29 farms (smallholder and large farms; tilled, no-till and adjacent bush), with the management survey of each field, January–February 2022.
• Watershed characteristics: a 10 m land cover composite, a 10 m curve number layer, land cover shares and properties of 40 subwatersheds.
• Geospatial data: gauge locations, pour points, subwatersheds, streams and longest flow paths derived from the Copernicus GLO-30 DEM.

How the datasets are organized

The repository has four top-level folders, one per kind of data. Each dataset folder follows the same pattern:

• Data files. For streamflow and rainfall, the data are in a *_Data/ folder with one subfolder per processing level (*_raw, *_L0, *_L1, *_L2; see Processing levels below), and one file per station in each. For soils, the final tables (SoilsData.csv, FieldMgmtData.csv) are at the top of the folder, with the records they were built from in RawData/ and intermediate tables in ProcessedData/. For the watershed and geospatial data, each layer has its own folder.
• QAQC/ holds the quality-control records: logs of every value that was removed, filled or flagged, and short reports of the rules and counts.
• Validation/ holds comparisons with independent data (survey datums, the CETRAD stream gauges, CHIRPS and TAHMO rainfall, regional soil profiles) and the code that produces them. Where a folder has several figures, they are in one PDF with a Figures_PNG/ folder of individual PNGs.
• Code/ holds the scripts that rebuild the processed files from the raw files, with a requirements.txt (the Validation code folders point to the one in the dataset's Code/), an MIT LICENSE and a run order in the folder ReadMe.
• ReadMe.txt in every folder describes the files in that folder; the ReadMe at the top of each dataset folder gives every column, unit, flag and method.

Station files are named by station code and name, variable, processing level and time step, for example CA04_Likii_stage_L1_5min.csv, CA04_Likii_discharge_L2_daily.csv and RN03_Juakali_rainfall_L1_hourly.csv. Station locations are in Geospatial_Data/GaugeLocations.csv.

• Hydrology_Field_Observations/
• Hydrology_DataTags.csv: QA/QC flag codes shared by streamflow and rainfall
• Streamflow/
  • Streamflow_Data/: Streamflow_raw/, Streamflow_L0/, Streamflow_L1/, Streamflow_L2/ (5min_L2/, Hourly_L2/, Daily_L2/), Barologger_raw/, Barologger_L0/, Barologger_L1/, BarologgerDataTags.csv
  • Rating_Curves/: BaRatinAGE rating curves, gaugings, cross-section and elevation surveys
  • QAQC/: gap tables, imputation decisions, build reports
  • Validation/: sensor datum checks; discharge recomputed from the rating equations; comparison with the CETRAD stations (figure, statistics, code)
  • Code/
• Precipitation/
  • Precipitation_Data/: Precipitation_raw/, Precipitation_L1/ (Native_Resolution_L1/, Hourly_L1/, Daily_L1/)
  • QAQC/: service and isolated tips removed, build summary
  • Validation/: comparison with CHIRPS, inter-gauge agreement, systematic losses
  • Code/
• Soils_Field_Observations/
• SoilsData.csv: all soil measurements, long format, with QA/QC flags
• FieldMgmtData.csv: management survey, one row per field
• *_metadata.csv, Soils_DataTags.csv: data dictionaries and flag codes
• RawData/, ProcessedData/, Code/, Validation/
• Watershed_Characteristics/
• SubwatershedProperties.csv: properties of the 40 subwatersheds
• LandCover/: 10 m land cover composite the curve numbers are built from, land cover per subwatershed
• CurveNumber/: curve number layer, class tables and rainfall zones
• inputs_not_redistributed/: how to obtain the inputs that are not included
• QAQC/, Code/
• Geospatial_Data/
• GaugeLocations.csv, PourPoints/, Streams/, WatershedDelineation/, QAQC/, Code/

FILE_MANIFEST.csv at the top of the repository lists every file with its path, size in bytes and SHA-256 checksum, so a download can be checked against it.

Processing levels

Files follow the CUAHSI ODM quality-control levels:

For soils, the level of each measurement is given in SoilsData_metadata.csv.

Conventions

• Time: local time, East Africa Time (UTC+03:00), written YYYY-MM-DD HH:MM:SS. There is no daylight-saving time.
• Missing values: NA in every file. The one exception is CrossSection_build_table.csv, an input to the cross-section code, where an empty cell means "use the booked value".
• Coordinates: WGS 84 (EPSG:4326) unless a file says otherwise; analysis layers in WGS 84 / UTM zone 37N (EPSG:32637).
• Files: CSV, UTF-8, comma-delimited, one header row.
• QA/QC flags: every value carries a flag. The codes are defined in Hydrology_Field_Observations/Hydrology_DataTags.csv (stage, discharge and rainfall), Hydrology_Field_Observations/Streamflow/Streamflow_Data/BarologgerDataTags.csv (barometric pressure) and Soils_Field_Observations/Soils_DataTags.csv (soils).

Privacy

Coordinates of smallholder fields are published at two decimal places (about 1.1 km). Large farms, including the bush sites on them, carry no coordinates, only a location relative to the watershed. No file holds names, telephone numbers or finer positions.

Third-party data

Products that may not be redistributed are not included; the code downloads them, or the folder ReadMe says how to obtain them. The CETRAD Early Warning System and TAHMO station records are not included; only figures or statistics derived from them are. Derived products carry the attribution their sources require (Copernicus GLO-30, FABDEM (comparison only), ERA5-Land, CHIRPS, HYSOGs250m, KENSOTER (ISRIC; summary ranges only), Impact Observatory / Microsoft / Esri land cover, QED cropland included with the permission of The Nature Conservancy); see the folder ReadMe files.

Citation

Please cite the data descriptor that accompanies this repository: Willard, L. L. et al. A multi-domain field dataset to evaluate agricultural impacts at the edge of a Kenyan savanna. Scientific Data (in review).

License

The data are licensed under the Creative Commons Attribution 4.0 International license (CC BY 4.0), except two layers derived from the QED cropland layer (CC BY-SA 4.0, © The Nature Conservancy, 2019). These are licensed under the Creative Commons Attribution-ShareAlike 4.0 International license (CC BY-SA 4.0):

• the land cover composite, Watershed_Characteristics/LandCover/MountKenyaLULC_2022.tif
• the curve number layer, Watershed_Characteristics/CurveNumber/CurveNumber.tif and CN_class.tif

The code in each Code/ folder, and the scripts in Geospatial_Data/QAQC/ and Watershed_Characteristics/QAQC/, are licensed under the MIT license; see the LICENSE file in each of those folders.