Work

Decoding LSTM to Reveal Baseflow Contributions in Fractured and Sedimentary Mountain Basins: A Case Study in the Sangre de Cristo Mountains, Southwestern United States

Year: 2026

Type: article

Abstract: This study investigates how a Long Short-Term Memory (LSTM) model inter-nally represents baseflow contributions in snowmelt-driven, semi-arid mountain basins with heterogeneous geologic characteristics. Five basins in the Sangre de Cristo Mountains of northern New Mexico, spanning fractured Precambrian bedrock and sedimen-tary-volcanic terrain, were used to evaluate both model performance and interpretability. Baseflow dynamics were inferred post hoc using the Baseflow Index (BFI) and a two-reservoir HEC-HMS (Hydrologic Engineering Center’s Hydrologic Modeling System) model. Although baseflow was not explicitly included in model training, internal cell state activations showed strong correlations with both shallow and deep baseflow com-ponents derived from the HEC-HMS model. To better understand how these relationships may change under climatic stress, BFI-based baseflow patterns were further analyzed un-der pre-drought and drought conditions. Results indicate that the LSTM learned to inter-nally distinguish between short- and long-residence flowpaths, encoding physically meaningful hydrologic behavior. This work demonstrates the potential for LSTM models to offer valuable insights into baseflow generation and groundwater–surface water inter-actions, particularly critical in water-scarce regions facing increasing drought frequency.

Source: Hydrology

DOI: 10.3390/hydrology13020051

Authors: Michael Rosati, Yeo H. Lim, Katie Zemlick, Kamran Syed

Institution: University of North Dakota

Language: English

Topic: Hydrological Forecasting Using AI

Subfield: Environmental Engineering

Field: Environmental Science

Domain: Physical Sciences

SDG: Climate action

Open Access status: gold

Estimated APC paid: $2,172