A dynamic stage–discharge rating for a migrating tropical river: Kalman-filtered rating shifts and their uncertainty at Bweramule, Semliki River, Uganda

Study region: The Semliki River at Bweramule (6,634 km²), on the Uganda–Democratic Republic of Congo border between Lake Edward and Lake Albert: the forecast point of an operational flood forecasting system, on a meander bend that migrates up to 11 m a year. Study focus: A fixed stage–discharge rating is exact on the day of its last gauging and silently wrong afterwards if the control moves. We replace it with a dynamic rating – the installed curve plus a stage shift that follows a random walk – estimated from 176 current-meter gaugings (1950–1975) and 11 ADCP gaugings (2024–2025) with a Kalman filter and smoother, the drift rate and gauging error fitted by maximum likelihood. Its 90 % band widens every day without a gauging. We test it against a fixed rating by one-step-ahead prediction, and measure the loop from unsteady flow with the Jones formula against an unsteady HEC-RAS hindcast. New hydrological insights for the region: The control at Bweramule drifts at 6.8 cm per square-root year. The dynamic rating raises the log-likelihood of the 1950–1975 gaugings from 153 to 253 (AIC −304 to −503), halves the gauging scatter (9.5 to 4.2 %) and predicts each next gauging with a 90 % band 42 % narrower (±0.10 against ±0.17 m) that is equally reliable. The control stood about 0.17 m low in 1950–1956 and 0.17 m high by 1975. In October 2026, 375 days after the last gauging, the stage for a given flow is known to ±0.13 m (about ±6 % in flow); a gauging every six months would hold it at ±0.10 m. Hysteresis is small: at the 99th-percentile rate of stage change the Jones correction is 0.6 %, and HEC-RAS agrees. Contents. The preprint (PDF), prepared to the author guidelines of Journal of Hydrology: Regional Studies; one notebook, reproduce_paper.ipynb, that reproduces the paper's results (also runs in Google Colab); semliki_dynrating.py – Kalman filter, RTS smoother, maximum-likelihood drift rate, rating with band, Jones correction; rating_R2026-10_table.csv – the installed rating R2026-10 at Bweramule (stage on the ADCP staff gauge, m; discharge, m³/s). The Bweramule gaugings, telemetry and HEC-RAS model are held by the Ministry of Water and Environment, Uganda, and are available on request; with them, the last cell of the notebook reproduces Table 3 and Figs. 2–3.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24367492
Citations
1
Primary Topic
Hydrology and Watershed Management Studies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

A dynamic stage–discharge rating for a migrating tropical river: Kalman-filtered rating shifts and their uncertainty at Bweramule, Semliki River, Uganda

Monica NAFULA, Benon Zaake, Charles KAKOOZA, Faizal Waiswa et al.
1 citations
Zenodo (CERN European Organization for Nuclear Research)
Hydrology and Watershed Management Studies
preprint

A dynamic stage–discharge rating for a migrating tropical river: Kalman-filtered rating shifts and their uncertainty at Bweramule, Semliki River, Uganda

Monica NAFULA, Benon Zaake, Charles KAKOOZA, Faizal Waiswa, Geoffry Kasibante
preprint en
1 citations

Abstract

Study region: The Semliki River at Bweramule (6,634 km²), on the Uganda–Democratic Republic of Congo border between Lake Edward and Lake Albert: the forecast point of an operational flood forecasting system, on a meander bend that migrates up to 11 m a year. Study focus: A fixed stage–discharge rating is exact on the day of its last gauging and silently wrong afterwards if the control moves. We replace it with a dynamic rating – the installed curve plus a stage shift that follows a random walk – estimated from 176 current-meter gaugings (1950–1975) and 11 ADCP gaugings (2024–2025) with a Kalman filter and smoother, the drift rate and gauging error fitted by maximum likelihood. Its 90 % band widens every day without a gauging. We test it against a fixed rating by one-step-ahead prediction, and measure the loop from unsteady flow with the Jones formula against an unsteady HEC-RAS hindcast. New hydrological insights for the region: The control at Bweramule drifts at 6.8 cm per square-root year. The dynamic rating raises the log-likelihood of the 1950–1975 gaugings from 153 to 253 (AIC −304 to −503), halves the gauging scatter (9.5 to 4.2 %) and predicts each next gauging with a 90 % band 42 % narrower (±0.10 against ±0.17 m) that is equally reliable. The control stood about 0.17 m low in 1950–1956 and 0.17 m high by 1975. In October 2026, 375 days after the last gauging, the stage for a given flow is known to ±0.13 m (about ±6 % in flow); a gauging every six months would hold it at ±0.10 m. Hysteresis is small: at the 99th-percentile rate of stage change the Jones correction is 0.6 %, and HEC-RAS agrees. Contents. The preprint (PDF), prepared to the author guidelines of Journal of Hydrology: Regional Studies; one notebook, reproduce_paper.ipynb, that reproduces the paper's results (also runs in Google Colab); semliki_dynrating.py – Kalman filter, RTS smoother, maximum-likelihood drift rate, rating with band, Jones correction; rating_R2026-10_table.csv – the installed rating R2026-10 at Bweramule (stage on the ADCP staff gauge, m; discharge, m³/s). The Bweramule gaugings, telemetry and HEC-RAS model are held by the Ministry of Water and Environment, Uganda, and are available on request; with them, the last cell of the notebook reproduces Table 3 and Figs. 2–3.

Zenodo (CERN European Organization for Nuclear Research)
Ministry of Water and Environment (UG)
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