LoginSign Up
← Back to all sparks
W

watina

ANALYTICS
Velocity0.0

Querying and Processing Data from the INBO Watina Data Warehouse

github.com/inbo/watina ↗

A groundwater database client that has started doing the domain analysis too

r-packagegroundwaterhydrochemistrydatabase-clientdata-visualizationresearch-tooling
◆Current state
watina is the R interface to the Watina groundwater monitoring database, and its history is mostly about making data retrieval correct: filter depths guessed conservatively when missing, spatial masking, aggregation methods per observation well, and a long series of fixes to keep the lazy database queries working across dbplyr versions. The most recent release moves past retrieval into interpretation, adding ionic ratio calculation and a Van Wirdum diagram to plot chemistry data.
◆Where it's heading
The package is drifting from a database client toward a domain toolkit. Early releases fought the data layer — connection handling moved to inbodb, sorting semantics changed, defunct dbplyr calls worked around. Recent work assumes retrieval is solved and adds hydrochemical analysis on top, along with defensive handling for the physically impossible inputs that analysis exposes, such as zero conductivity in the warehouse.
◆Prediction
Expect further chemistry analysis and plotting helpers rather than new retrieval functions, since that is where the newest release invested and where the accompanying vignette points.

◆Recent moves

  1. 4mo ago

    Ionic ratios and Van Wirdum diagrams for chemistry data

    Adds calculate_ir() for ionic ratios and a Van Wirdum diagram background as a ggplot object users can layer data onto, with a chemistry plotting vignette. The clearest step from retrieving groundwater data to interpreting it.

    View source ↗
  2. 2y ago

    KSgeneral moved to Suggests to survive CRAN removal

    Moves the KSgeneral dependency to Suggests so the package does not break when it leaves CRAN. Packaging insurance with no change to behaviour.

    View source ↗
  3. 5y ago

    Dataframe input restored after a defunct dbplyr call

    Restores get_xg3() and get_chem() for dataframe input by avoiding a defunct dbplyr function. One of several releases spent keeping the lazy query layer working against a moving dependency.

    View source ↗
  4. 5y ago

    dbplyr 2.0 compatibility drops the need for a forked dependency

    The package becomes compatible with CRAN dbplyr 2.0.0, so the forked version is no longer required — a real reduction in installation friction. Lazy results also stop being sorted, with sorting confined to collected tibbles.

    View source ↗
  5. 5y ago

    Spatial clustering of wells and richer location attributes

    Adds cluster_locs() for spatially clustering well locations and expands get_locs() with conservative filter-depth guessing, eight more observation well attributes and four aggregation methods. The release that made retrieval genuinely configurable.

    View source ↗
  6. 6y ago

    Soil surface calculation fixed after a select() dropped the variable

    Corrects a soilsurf_ost calculation that silently dropped the variable by computing it inside select() rather than mutate(). A quiet-failure bug, invisible until someone checked the output.

    View source ↗