Colorado River study finds the snow-fed upper basin was vulnerable to groundwater loss

A Colorado River Basin study estimates greater groundwater losses upstream than downstream during water years 2020–2022.

These are regional totals calculated from satellite and hydrological data, not evidence that every upstream well fell further.

Greater recharge did not guarantee resilience

The study appeared in Water Resources Research on October 13; Live Science revisited it on October 22 with an author email interview. Researchers combined satellite gravity and hydrological data for water years 2004–2023. The upper basin’s snowfall and precipitation did not prevent substantial losses during the severe drought. A water year runs from October 1 of the preceding calendar year through September 30.

Read the uncertainty alongside the comparison

Estimated groundwater losses during water years 2020–2022 were 17.5±3.5 km³ upstream and 9.4±6.5 km³ downstream. The central estimates differ by about 1.9 times, but their uncertainties matter. These are total volumes for differently sized basins, not losses per unit area or rankings of individual communities. Satellites do not isolate groundwater directly; other storage components must be estimated and separated.

Why 58.3 km³ is not the groundwater total

The paper reports a 58.3±4.5 km³ decline in total water storage over water years 2004–2023. Groundwater loss for that period is separately reported as 25.5±6.0 km³. Live Science’s description of 58.3 km³ as groundwater does not match that distinction and is not adopted here. Combining soil, snow, surface water and groundwater into the wrong category changes the story.

What remains unresolved

Surface-water shortages and groundwater losses were more closely linked upstream; parts of the lower basin showed weaker coupling. The author interview points to both snowmelt recharge and sustained extraction. Statistical links do not determine each cause’s contribution or predict the same ratio in the next drought. Further satellite, well and pumping records are needed. The interview is separate reporting, not an independent replication experiment.

Primary sources and independent checks

Original paper: satellite/hydrological data, water years, estimates and uncertainty; web-tool text preserved after direct HTTP 403: Oyedele et al. · Water Resources Research · 2026-10-13

Independent reporting: author email on recharge and limits, not replication; the 58.3 km³ category is corrected against the paper: Live Science · 2026-10-22

Read in other languages

한국어 · English · Español · 日本語 · Tiếng Việt