A Deep State Space Model for Rainfall-Runoff Simulations

TL;DR

This paper introduces the S4D-FT state space model for rainfall-runoff simulations, demonstrating it outperforms LSTM and physically-based models across numerous watersheds, offering a new deep learning approach for hydrological modeling.

Contribution

The paper presents the novel S4D-FT state space model for rainfall-runoff prediction, outperforming LSTM and traditional models in diverse hydrological regions.

Findings

S4D-FT outperforms LSTM in rainfall-runoff simulations across 531 watersheds.

S4D-FT surpasses physically-based Sacramento Soil Moisture model.

The model expands the toolkit of deep learning methods in hydrology.

Abstract

The classical way of studying the rainfall-runoff processes in the water cycle relies on conceptual or physically-based hydrologic models. Deep learning (DL) has recently emerged as an alternative and blossomed in hydrology community for rainfall-runoff simulations. However, the decades-old Long Short-Term Memory (LSTM) network remains the benchmark for this task, outperforming newer architectures like Transformers. In this work, we propose a State Space Model (SSM), specifically the Frequency Tuned Diagonal State Space Sequence (S4D-FT) model, for rainfall-runoff simulations. The proposed S4D-FT is benchmarked against the established LSTM and a physically-based Sacramento Soil Moisture Accounting model across 531 watersheds in the contiguous United States (CONUS). Results show that S4D-FT is able to outperform the LSTM model across diverse regions. Our pioneering introduction of the S4D-FT for rainfall-runoff simulations challenges the dominance of LSTM in the hydrology community and expands the arsenal of DL tools available for hydrological modeling.