Output gallery • potentiomap

Synthetic monitoring wells distributed inside an irregular area of interest, with point fill progressing from blue lower heads to red higher heads.

Monitoring network

Method: released synthetic points. Units: UTM metres; synthetic head units. Network geometry conditions every interpolation. Preparation code.

Synthetic well points labeled with standardized observation names inside the area of interest.

Standardized potentiometric points

Method: ps_make_points(). Units: UTM metres; synthetic head units in Z. Preparation pathways.

Thin-plate spline surface shaded blue at lower synthetic heads and red at higher heads, with monitoring wells.

TPS surface

Method: thin-plate spline. Units: synthetic head units; 75-metre cells. Method comparison.

Inverse-distance-weighted surface shaded blue at lower synthetic heads and red at higher heads, with monitoring wells.

IDW surface

Method: inverse distance weighting. Units: synthetic head units; 75-metre cells. IDW parameters.

Ordinary-kriging surface

Method: ordinary kriging with the released variogram workflow. Units: synthetic head units; 75-metre cells. Warnings and comparison.

Universal-kriging surface

Method: universal kriging with quadratic drift. Units: synthetic head units; 75-metre cells. Warnings and comparison.

Four panels compare TPS, IDW, ordinary kriging, and universal kriging using one head scale, the same grid, contours, and wells.

Four-method comparison

Methods: TPS, IDW, OK, UK. Units: synthetic head units; identical 75-metre cells. No method is universally preferred. Full comparison.

One-unit TPS contours drawn as blue lines over the area of interest with monitoring wells.

Contour-only product

Method: ps_contours() from TPS. Units: one synthetic head unit. Contour settings.

TPS modeled surface with dark one-unit contour lines and white monitoring wells.

Surface plus contours

Method: TPS and regular contours. Units: synthetic head units. Quick start.

IDW modeled surface and contours with each synthetic monitoring well labeled by identifier.

Surface plus labeled wells

Method: IDW with point labels. Units: synthetic head units. Labels connect the map to reviewable observations. Observation fields.

Three panels compare the original TPS surface, one-pass mean smoothing, and two-pass median smoothing on one head scale.

Unsmoothed and smoothed surfaces

Methods: original TPS, focal mean, focal median. Units: synthetic head units. Smoothing changes the model. Smoothing cautions.

Original and mean-smoothed TPS surfaces with one-unit contours, displayed side by side.

Unsmoothed and smoothed contours

Method: contours before and after focal mean smoothing. Units: synthetic head units. Retain the original.

Hydraulic-gradient magnitude raster colored from light yellow lower magnitude to dark red higher magnitude, with modeled-head contours.

Hydraulic-gradient raster

Method: terrain derivative of the TPS surface. Units: head change per horizontal map unit. Gradient products.

TPS surface and contours with a sparse set of black arrowheads pointing toward decreasing modeled head.

Sparse arrow layout

Method: ps_flow_arrows(), larger res_factor. Units: direction from modeled head; length is cartographic. Arrow controls.

TPS surface and contours with a denser set of black arrowheads pointing toward decreasing modeled head.

Dense arrow layout

Method: ps_flow_arrows(), smaller res_factor. Units: direction from modeled head; density is not monitoring density. Arrow controls.

Sparse arrow lines over a TPS surface with orange circular symbols marking the final vertex of each line.

Arrow-tip points

Method: ps_arrow_vertices(which = “last”). Units: UTM metres. Tips support the downgradient check. Direction verification.

Sparse arrow lines over a TPS surface with blue square symbols marking the first vertex of each line.

Arrow-base points

Method: ps_arrow_vertices(which = “first”). Units: UTM metres. Direction verification.

Native potentiomap quicklook PNG of the synthetic TPS surface with blue and white contours and labeled wells.

Exported native quicklook

Method: ps_export_surfaces() and ps_quicklook(). Units: synthetic head units. Export and read back.

Three synthetic monitoring-event TPS surfaces shown on the same grid and head scale for January, April, and July 2026.

Repeated monitoring events

Method: external R iteration around TPS on a common template. Units: synthetic head units. Event workflow.

Thirty-six public USGS Stanley Shale wells shown inside a buffered analysis polygon near Hot Springs, Arkansas.

Public monitoring network

Source: USGS 2017 Stanley Shale subset. Units: NAD83 UTM zone 15N metres. Source and filtering.

TPS surface from 36 USGS Stanley Shale wells, with twenty-foot contours and well points, shaded blue at lower and red at higher groundwater-level altitude.

Public-data software example

Method: TPS with 20-foot contours. Units: feet above NAVD 88; 75-metre cells. This is not the official interpreted surface. Complete example.

Two panels summarize retained, removed, warning, and error counts from synthetic observation QA and show the selected measurement time for each spring-event well around a noon target.

Observation QA and event selection

Methods: ps_check_observations() and ps_select_event(). Settings: nearest reading in a three-hour window around noon UTC. A time window does not guarantee hydraulic synchroneity. QA and event workflow.

Observed versus predicted synthetic heads for IDW and TPS beside a bar chart of three-fold validation RMSE, with a dashed one-to-one line.

Held-out-well validation

Methods: IDW and TPS; three-fold direct prediction with seed 21. Units: synthetic head units. The errors describe this held-out-well task, not continuous area-wide accuracy. Validation designs.

Prediction support

Method: support retained by ps_interpolate(). Settings: 300-metre grid and declared geometric criteria. Classes describe support, not confidence or correctness. Support guidance.

Raster of the absolute modeled-head range between IDW and TPS, colored pale yellow at lower disagreement and dark red at higher disagreement.

Method disagreement

Method: cellwise IDW–TPS range from ps_method_disagreement(). Units: synthetic head units. Method spread is descriptive and is not statistical uncertainty. Ensembles and disagreement.

Ordinary-kriging prediction standard error raster colored pale yellow at lower model-conditional error and dark red at higher error, with wells overlaid.

Model-conditional kriging uncertainty

Method: ordinary-kriging prediction standard error. Units: synthetic head units. It is conditional on the fitted covariance model and is not a universal error bound. Uncertainty scope.

Modeled autumn-minus-spring head-change raster with blue negative changes, red positive changes, and paired synthetic event wells.

Monitoring-event head change

Method: paired wells and common-grid IDW surfaces. Units: synthetic head units; autumn minus spring. Modeled head change is not storage change. Event comparison.

Bar chart ranking seven synthetic wells by the RMSE difference between a full IDW surface and each leave-one-well surface.

Leave-one-well influence

Method: ps_well_influence() with IDW and 400-metre cells. Units: synthetic head units. Ranking depends on the method, grid, network, and comparison metric. Network sensitivity.

Existing synthetic monitoring wells shown as blue circles, feasible candidates as yellow triangles, and two sequentially selected spatial-coverage candidates as red triangles.

Candidate monitoring locations

Method: sequential spatial-coverage scoring with two selections. Units: projected metres. Scores do not replace access, screen design, feasibility, budget, or permitting review. Candidate constraints.

Depth surface and transect section

Methods: ps_depth_to_water_surface() and ps_cross_section(). Units: synthetic metres; section uses declared 3× vertical exaggeration. The section samples surfaces and does not invent hydrostratigraphy. Profiles and sections.