Mississippi River discharge forecasts

Skip to content

Initializing search

Abstract

Forecasting the discharge of large rivers is a critical challenge for water resource management,<br>navigation, flood prevention, and the operation of hydraulic infrastructure. Yet, building a robust<br>forecasting pipeline from real-world data requires far more than simply training a machine learning model:<br>you first need to identify relevant data sources, clean and harmonize them, and engineer appropriate<br>explanatory variables.

This tutorial offers a comprehensive approach to forecasting daily river discharge for the Mississippi<br>River at St. Louis, Missouri, with a one-day ahead horizon, using open data from the USGS and Copernicus<br>ERA5.

The progression is deliberately incremental in order to measure the contribution of each step, each of<br>which constitutes a section of this tutorial:

Construction of the reference hydrological time series from USGS data;

Initial univariate experiments using only the local discharge history;

Integration of upstream hydrological stations located on the Mississippi and Missouri rivers;

Multivariate forecasts leveraging these different stations simultaneously;

Construction of a meteorological dataset from ERA5 data and spatial aggregation of precipitation;

Integration of meteorological variables into the forecasting pipeline.

Beyond the performance achieved, this tutorial emphasizes methodology: respecting data chronology,<br>preventing time leakage, building sliding windows, feature engineering, and comparing several modeling<br>approaches.

It is aimed at students, data scientists, and hydrologists who wish to build step by step a reproducible<br>forecasting pipeline applied to a real-world time series problem.

© 2026 Eric Duhamel