LakeFM: Toward a Foundation Model for Aquatic Ecosystems Using Irregular Multivariate Multi-depth Time Series Data

2026-06-09

AI Summary

Existing lake ecological monitoring data are commonly characterized by irregular sampling in both time and depth, and exhibit substantial heterogeneity across lakes in terms of variables, structure, and observation protocols, hindering the generalizability of conventional methods. To address this challenge, this work proposes LakeFM—the first foundation model tailored for aquatic ecosystems that handles irregular, multivariate, multi-depth time series. LakeFM leverages large-scale simulated and observational data for pretraining to learn universal cross-lake representations. By integrating physical constraints and overcoming the limitations of regular sampling assumptions, the model enables effective transfer learning and dynamic forecasting. It achieves performance on par with or superior to existing approaches across multiple tasks while generating physically consistent predictions aligned with real-world lake dynamics.

Abstract

Understanding and forecasting lake dynamics is critical for monitoring water quality and ecosystem health across lakes and reservoirs. While machine learning methods have been recently applied to ecological time-series data, existing works assume regular sampling in time and depth, and struggle to generalize across lakes with heterogeneous variables, depths, and observation patterns. To address these limitations, we introduce \textsc{LakeFM}, a foundation model for aquatic systems, pre-trained on large-scale ecological datasets comprising both simulated and observed lakes. Through extensive empirical evaluation, we show that \textsc{LakeFM} learns meaningful representations spanning broader lake-level characteristics, and achieves competitive or often superior-forecasting performance compared to existing time-series foundation and non-foundation models, while producing physically plausible predictions consistent with real-world lake dynamics.

Problem

• aquatic ecosystems
• irregular time series
• multivariate data
• multi-depth observations
• lake dynamics

Innovation

• foundation model
• irregular time series
• multivariate multi-depth data
• aquatic ecosystems
• lake dynamics forecasting