How Effective are Large Time Series Models in Hydrology? A Study on Water Level Forecasting in Everglades

2025-05-02

AI Summary

Traditional physics-based and statistical models for water-level forecasting in the Florida Everglades suffer from high computational costs and poor generalizability. Method: This study systematically evaluates 12 task-specific models and 5 time-series foundation models (e.g., Chronos, TimesNet, PatchTST) on real-world wetland hydrological data. Contribution/Results: It provides the first empirical validation of time-series foundation models for complex, nonstationary wetland hydrology forecasting, revealing that Chronos significantly outperforms all baselines (23.6% lower MAE) and uncovering critical architecture–hydrodynamics alignment principles. Most foundation models underperform, while task-specific models exhibit strong architecture-dependent performance. We propose a multiscale error evaluation and attribution framework, establishing the first empirically grounded benchmark for environmental AI model selection tailored to wetland hydrology.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web search models and ranking

Abstract

The Everglades play a crucial role in flood and drought regulation, water resource planning, and ecosystem management in the surrounding regions. However, traditional physics-based and statistical methods for predicting water levels often face significant challenges, including high computational costs and limited adaptability to diverse or unforeseen conditions. Recent advancements in large time series models have demonstrated the potential to address these limitations, with state-of-the-art deep learning and foundation models achieving remarkable success in time series forecasting across various domains. Despite this progress, their application to critical environmental systems, such as the Everglades, remains underexplored. In this study, we fill the gap by investigating twelve task-specific models and five time series foundation models across six categories for a real-world application focused on water level prediction in the Everglades. Our primary results show that the foundation model, Chronos, significantly outperforms all other models while the remaining foundation models exhibit relatively poor performance. Moreover, the performance of task-specific models varies with the model architectures. Lastly, we discuss the possible reasons for the varying performance of models.

Problem

• Evaluating large time series models for Everglades water level forecasting
• Addressing limitations of traditional hydrological prediction methods
• Comparing performance of task-specific and foundation models in hydrology

Innovation

• Uses large time series models for hydrology
• Compares twelve task-specific forecasting models
• Evaluates five time series foundation models