NASS QuickStats: current through 2024 · FieldPulse v1.0 · April 2026

High Plains Aquifer · 8 states · 227 counties on footprint

The Ogallala is draining.

A county-level accountability map of the aquifer that sustains $35 billion in annual agricultural production across the High Plains. Thickness + decline from 208 measurement-backed counties, plus 19 more sampled from USGS McGuire's published rasters (SIR 2012-5177 / 5291). NB02 CatBoost forecaster with conformal 80% bands on 89 counties. Every number traces to a source.

Counties on footprint227/606
Pumping estimate9.5M AF/yr
NB02 spatial-CV R²0.93
Simulation horizon1950—2100
Scenario
Visualization
Sat. thickness (m)
<5
10
25
55
100
150+
measured wellsUSGS rasteroff-aquifer
§ 01 Live view
2024
SCENARIO: STATUS QUO
loading 606 HPA counties…
§ 01b County drill-down
Pick a county
Click any county on the map — or a row in the leaderboard below — to see its thickness trajectory, crop-mix attribution, and data provenance.
Try Sheridan KS or Dallam TX
§ 05 Methodology
Every number on screen traces to a row in docs/data_sources.md.
22
Data sources
22 upstream feeds → 35 parquets. USGS HPA boundary + McGuire rasters (SIR 2012-5177 thickness 2009, SIR 2012-5291 wlc predev→2011, SIR 2023-5143 wlc predev→2019 + 2017–19 biennium), USGS NWIS wells & water-level history, NGWMN, KGS WIZARD / WIMAS / HPA bedrock, TX TWDB Groundwater Database, NE DNR Statewide Well Registry + 2021 Groundwater Management Summary (23-NRD allocation rules), USDA NASS Census 2022, USDA NASS IWMS 2018 (Table 28 method mix), USDA ERS cost + returns, EPA eGRID 2022, EIA State Electricity Profiles (industrial retail prices), NOAA NCEI nClimDiv per-county precipitation (1895–present), Deines et al. 2019 AIM-HPA (annual Landsat irrigation classification, 1984–2017), US Census TIGER/Line 2022.
4
Model tiers
T1 Physics baseline — linear back-projection from 2024 thickness + annualized decline. T2 Scenario deltas on T1 forward-sim (regional policy counterfactuals). T3 NB02 CatBoost forecaster (R² ≈ 0.93 on spatial-CV across 90 TWDB + KGS WIZARD counties) for next-year thickness, wrapped in Romano et al. 2019 conformalized quantile regression for calibrated 80% bands. T4 Per-crop pumping attribution (pure arithmetic: acres × IWMS rate × method efficiency).
0.52×
USGS calibration
Our inferred pumping is ~half of USGS 2015 reported groundwater withdrawal. Percentage deltas unaffected; absolute CO₂ numbers ~2× understated. Surfaced in the UI as a conservative estimate — the honesty is the credibility feature. Both the inferred and the USGS-reported totals ship in the GeoJSON for per-county audit.
References · primary data
click to expand ▾
McGuire, V.L., Lund, K.D., Densmore, B.K., 2012
Saturated Thickness and Water in Storage in the High Plains Aquifer, 2009, and Water-Level Changes and Changes in Water in Storage in the High Plains Aquifer, 1980–2009. USGS Scientific Investigations Report 2012–5177.
McGuire, V.L., 2013
Water-Level and Storage Changes in the High Plains Aquifer, Predevelopment to 2011 and 2009–11. USGS Scientific Investigations Report 2012–5291. (Legacy — kept for audit.)
McGuire, V.L., & Strauch, K.R., 2024
Water-Level and Recoverable Water in Storage Changes, High Plains Aquifer, Predevelopment to 2019 and 2017 to 2019. USGS Scientific Investigations Report 2023–5143. Primary source for our per-county decline rate (69-year horizon).
NOAA NCEI, 2024
nClimDiv Per-County Monthly Precipitation Dataset (1895–present). Annual 1991–2020 normals plus 2019–2023 recent window anomaly percentage per county — the climate context that frames the aquifer story.
U.S. EIA, 2024
State Electricity Profiles — industrial-sector retail price in cents/kWh by state (annual 2024). Combined with per-county kWh/AF pumping intensity to derive a real-dollar pumping cost per acre-foot.
Nebraska DNR, 2021
Groundwater Management Summary (annual report). Source of our manually-curated 8-NRD allocation-rule table: base allocations, carryforward caps, moratoria, Compact-Call hard caps, NCORPE streamflow-augmentation co-ownership.
Deines, J.M., et al., 2019
Annual Irrigation Maps — High Plains Aquifer (AIM-HPA). 30 m binary Landsat classification, 1984–2017, 34 annual rasters. RSE 233: 111400. Per-county zonal sums drive the 'Irrigated acres' trajectory in every drill-down.
USGS NGWMN
National Ground-Water Monitoring Network — federated water-level + well time series from state + federal partners. Accessed 2026-04 via Water Data OGC.
KGS WIZARD
Kansas Geological Survey Water Information Storage and Retrieval Database — 26k wells, 611k measurements. Primary source for all KS thickness + decline.
TWDB GWDB
Texas Water Development Board Groundwater Database — precomputed saturated thickness per well. Primary source for TX Panhandle counties.
NE DNR
Nebraska Department of Natural Resources Statewide Well Registry — 93-county summary of active irrigation wells + median total depth.
USDA NASS Census of Agriculture 2022
County-level irrigated acres per crop (corn, soybeans, sorghum, wheat, cotton, alfalfa). Baseline spine for all 606 HPA-state counties.
USDA NASS IWMS 2023
Irrigation and Water Management Survey — per-state × per-crop acre-feet-per-acre application rates. Feeds the pumping-inference model.
USDA ERS Commodity Costs + Returns
National gross value of production, latest year (2024). Used for per-acre revenue and ag-value deltas in scenarios.
EPA eGRID 2022
Per-state electric grid CO₂ intensity (kg CO₂ / kWh). Combined with pumping energy intensity (kWh / AF) to compute embodied emissions.
US Census TIGER/Line 2022
cb_2022_us_county_500k cartographic boundaries. County polygons used for zonal statistics over USGS rasters and for frontend choropleth geometry.
References · modeling
click to expand ▾
Deines, J.M., et al., 2019
Annual irrigation dynamics in the U.S. Northern High Plains derived from Landsat satellite data. 9 m saturated-thickness threshold for center-pivot economic viability.
Romano, Y., Patterson, E., Candès, E.J., 2019
Conformalized Quantile Regression. The split-conformal wrapper NB02 uses to inflate LightGBM quantile bands from miscalibrated 0.35 coverage to a finite-sample-valid 0.80.
Basso, B., et al., 2025
Sheridan-6 Local Enhanced Management Area (LEMA) rules — ~27% pumping reduction, crop mix shift corn→sorghum/wheat — the source for the 'Kansas LEMA, aquifer-wide' scenario.
Prokhorenkova, L., et al., 2018
CatBoost: unbiased boosting with categorical features. Winning learner for NB02 (depth 5, 700 iterations, l2_leaf_reg 3.0) — beats per-county OLS, LightGBM, and XGBoost on 5-fold spatial CV.
baseline manifest (live)scenarios index (live)docs/methodology.mddocs/limitations.mdbaseline.parquet · 606 rowsGitHub · Vigneshwarr3/FusionHack_2026