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

In three linesLakeFM is a foundation model pre-trained on large-scale ecological datasets (simulated and observed lakes) to forecast lake dynamics and water quality. It handles irregular multivariate multi-depth time series, generalizes across heterogeneous lakes, and outperforms existing models with physically plausible predictions.

Summary generated by Claude — human-verified