A Network-Constrained Quality Control Framework for Identifying Reliability Concerns in River Discharge Records: A Case Study in the Northwest Territories

Abstract Published river discharge is usually treated as deterministic, yet rating-curve construction, temporary shifts and overrides introduce uncertainties missed by conventional quality control (QC); only 69% of Water Survey of Canada (WSC) records can be reproduced from stage, rating curves and shifts, falling to 41% in January–February. We introduce Network-Constrained Quality Control (NCQC), which couples a stochastic error model with hydrometric redundancy to detect, attribute and assess inconsistencies that escape station-level QC. NCQC combines seven diagnostics: metadata-independent provenance detection, physical admissibility, satellite vegetation-index regime classification, regime-conditioned mass-balance closure, temporal–spatial attribution, cumulative-sum persistence detection and a network-independence test. Applied to 319 monthly observations (2000–2026) from three gauges at the Mackenzie–Liard confluence, Canada, the node closes without detectable bias at a robust relative scale of 1.67%, far tighter than published discharge uncertainty. NCQC flagged 54 months (16.9%) in 22 episodes, including a 14-month anomaly in 2015–2016, a persistent + 4.79% open-water offset in 2018–2019 and provisional winter deviations reaching − 38.9% in 2025–2026. In 4327 synthetic error-injection trials at an 8.0% background flag rate, detection probabilities for 5% and 10% sustained biases were 0.72 and 0.89 (0.66 and 0.87 for newly raised flags), against fewer than 6% for a station-level baseline. Correct attribution reached 0.74–0.80 upstream but only 0.30 downstream. Reported significant-digit precision independently identified the approved-to-provisional transition, and the ice-season closure scale fell from 1.46% (2000–2019) to 0.20% (2020–2024), consistent with possible dependence among estimates. NCQC thus identifies discharge inconsistencies and evaluates when mass-balance agreement stops being independent evidence of accuracy.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11563761/v1
Primary Topic
Hydrology and Watershed Management Studies
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

A Network-Constrained Quality Control Framework for Identifying Reliability Concerns in River Discharge Records: A Case Study in the Northwest Territories

Hooman Razavi, Hamidreza Shirkhani, Hossein Bonakdari
2 links
Research Square
Hydrology and Watershed Management Studies
preprint

A Network-Constrained Quality Control Framework for Identifying Reliability Concerns in River Discharge Records: A Case Study in the Northwest Territories

Hooman Razavi, Hamidreza Shirkhani, Hossein Bonakdari
preprint en

Abstract

Abstract Published river discharge is usually treated as deterministic, yet rating-curve construction, temporary shifts and overrides introduce uncertainties missed by conventional quality control (QC); only 69% of Water Survey of Canada (WSC) records can be reproduced from stage, rating curves and shifts, falling to 41% in January–February. We introduce Network-Constrained Quality Control (NCQC), which couples a stochastic error model with hydrometric redundancy to detect, attribute and assess inconsistencies that escape station-level QC. NCQC combines seven diagnostics: metadata-independent provenance detection, physical admissibility, satellite vegetation-index regime classification, regime-conditioned mass-balance closure, temporal–spatial attribution, cumulative-sum persistence detection and a network-independence test. Applied to 319 monthly observations (2000–2026) from three gauges at the Mackenzie–Liard confluence, Canada, the node closes without detectable bias at a robust relative scale of 1.67%, far tighter than published discharge uncertainty. NCQC flagged 54 months (16.9%) in 22 episodes, including a 14-month anomaly in 2015–2016, a persistent + 4.79% open-water offset in 2018–2019 and provisional winter deviations reaching − 38.9% in 2025–2026. In 4327 synthetic error-injection trials at an 8.0% background flag rate, detection probabilities for 5% and 10% sustained biases were 0.72 and 0.89 (0.66 and 0.87 for newly raised flags), against fewer than 6% for a station-level baseline. Correct attribution reached 0.74–0.80 upstream but only 0.30 downstream. Reported significant-digit precision independently identified the approved-to-provisional transition, and the ice-season closure scale fell from 1.46% (2000–2019) to 0.20% (2020–2024), consistent with possible dependence among estimates. NCQC thus identifies discharge inconsistencies and evaluates when mass-balance agreement stops being independent evidence of accuracy.

Research Square
University of Ottawa (CA), National Research Council Canada (CA), Ottawa University (US)
Natural Sciences and Engineering Research Council of Canada
Clean water and sanitation
Hydrology and Watershed Management Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon