Storm-Level Independent Verification of Operational Cloud Seeding: Seven Seasons of Utah's Program, 2016-2025
Abstract
Operational winter cloud seeding programs claim precipitation increases of 3-15%, but the evaluations supporting these claims are produced by program operators using target/control methods that cannot separate seeding effects from operator storm selection: storms are seeded precisely when atmospheric conditions favor the target areas. This study performs, to the author's knowledge, the first independent storm-level evaluation of an operational program built entirely from public records. We assemble a machine-readable record of Utah's operations across seven winters (2016-17 and 2019-20 through 2024-25): 181 generator sites and 525 seeded storm periods totaling 37,773 generator-hours, transcribed from the operator's seasonal reports (the 2024-25 set published online, six earlier seasons provided by the Division of Water Resources), together with per-storm 700 hPa winds from the HRRR archive and daily precipitation from 546 SNOTEL stations in seven states. Three research designs of increasing selection-robustness are applied. A naive seeded-versus-unseeded storm comparison yields an implausible +335% for the largest program, quantifying the selection artifact. Same-sector near-miss controls reduce the pooled estimate to +24% [+4, +50]. The most robust design, a wind-rotation test in which plume exposure is recomputed from each storm's own wind field per generator, yields, pooled over seven seasons (937 program-events, 143 seeded), +5.8% [-0.0, +12.1] in-plume during seeded storms against a placebo of +3.0% [+0.1, +6.2] in-plume during unseeded storms. The placebo is now resolvably positive, revealing a wind-aligned orographic gradient that the smooth control does not fully absorb; the placebo-corrected seeding effect is +2.7% [-2.1, +8.8] per seeded storm. Seven seasons therefore remain consistent with a small positive operational effect and with zero, and exclude per-storm effects above ~9%, ruling out the upper half of the 3-15% claims. Carried through to water at the State's own baseline, the estimate implies an expected ~95,000 acre-feet per year (38% of the 249,600 claimed) at ~$52 per acre-foot, with a 17% probability that the program removes water from its target areas. All data and code are released.
Related
- Towards operational data-driven regional weather forecasting: extreme-event skill and near-real-time verification with HRRRCast
- Independent Verification of the European Climate Law's Effect on Nitrogen Dioxide Pollution: A Satellite-Derived, Difference-in-Differences Analysis
- Snow cover variability in the Western Beskids (Poland) during the 1951–2023 winter seasons
- Variability of suspended sediment transport dynamics between seasons and flow pathways in a small agricultural catchment
- Satellite Quantification of Methane Emissions from Individual Countries in Southeast Asia: Roles of Dry and Wet Seasons
BibTeX
@misc{judy2026stormlevel,
title = {{Storm-Level Independent Verification of Operational Cloud Seeding: Seven Seasons of Utah's Program, 2016-2025}},
author = {Bryce Judy},
howpublished = {EarthArXiv},
year = {2026},
month = oct,
doi = {10.31223/x5d23z},
url = {https://eartharxiv.org/repository/view/15605/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Storm-Level Independent Verification of Operational Cloud Seeding: Seven Seasons of Utah's Program, 2016-2025
AU - Bryce Judy
PY - 2026
DA - 2026/10/20
PB - EarthArXiv
DO - 10.31223/x5d23z
UR - https://eartharxiv.org/repository/view/15605/
AB - Operational winter cloud seeding programs claim precipitation increases of 3-15%, but the evaluations supporting these claims are produced by program operators using target/control methods that cannot separate seeding effects from operator storm selection: storms are seeded precisely when atmospheric conditions favor the target areas. This study performs, to the author's knowledge, the first independent storm-level evaluation of an operational program built entirely from public records. We assemble a machine-readable record of Utah's operations across seven winters (2016-17 and 2019-20 through 2024-25): 181 generator sites and 525 seeded storm periods totaling 37,773 generator-hours, transcribed from the operator's seasonal reports (the 2024-25 set published online, six earlier seasons provided by the Division of Water Resources), together with per-storm 700 hPa winds from the HRRR archive and daily precipitation from 546 SNOTEL stations in seven states. Three research designs of increasing selection-robustness are applied. A naive seeded-versus-unseeded storm comparison yields an implausible +335% for the largest program, quantifying the selection artifact. Same-sector near-miss controls reduce the pooled estimate to +24% [+4, +50]. The most robust design, a wind-rotation test in which plume exposure is recomputed from each storm's own wind field per generator, yields, pooled over seven seasons (937 program-events, 143 seeded), +5.8% [-0.0, +12.1] in-plume during seeded storms against a placebo of +3.0% [+0.1, +6.2] in-plume during unseeded storms. The placebo is now resolvably positive, revealing a wind-aligned orographic gradient that the smooth control does not fully absorb; the placebo-corrected seeding effect is +2.7% [-2.1, +8.8] per seeded storm. Seven seasons therefore remain consistent with a small positive operational effect and with zero, and exclude per-storm effects above ~9%, ruling out the upper half of the 3-15% claims. Carried through to water at the State's own baseline, the estimate implies an expected ~95,000 acre-feet per year (38% of the 249,600 claimed) at ~$52 per acre-foot, with a 17% probability that the program removes water from its target areas. All data and code are released.
ER -