Leveraging Exogenous Signals for Hydrology Time Series Forecasting

TL;DR

This paper explores how integrating domain-specific exogenous signals, especially natural periodic data, enhances hydrological rainfall-runoff forecasting models beyond generic foundation models.

Contribution

It demonstrates the importance of including comprehensive exogenous inputs and domain knowledge to improve hydrological time series forecasting accuracy.

Findings

Models with exogenous inputs outperform baseline models.

Incorporating natural annual periodic signals yields significant improvements.

Foundation models are less effective without domain-specific exogenous data.

Abstract

Recent advances in time series research facilitate the development of foundation models. While many state-of-the-art time series foundation models have been introduced, few studies examine their effectiveness in specific downstream applications in physical science. This work investigates the role of integrating domain knowledge into time series models for hydrological rainfall-runoff modeling. Using the CAMELS-US dataset, which includes rainfall and runoff data from 671 locations with six time series streams and 30 static features, we compare baseline and foundation models. Results demonstrate that models incorporating comprehensive known exogenous inputs outperform more limited approaches, including foundation models. Notably, incorporating natural annual periodic time series contribute the most significant improvements.