Zero-Shot Transfer Capabilities of the Sundial Foundation Model for Leaf Area Index Forecasting

Peining Zhang, Hongchen Qin, Haochen Zhang, Ziqi Guo, Guiling Wang, Jinbo Bi

Keywords: Leaf Area Index (LAI), Remote Sensing Time Series, Vegetation Forecasting, Temporal Forecasting, Foundation Models

TL;DR: Zero-shot time-series foundation model surpasses supervised LSTMs in LAI forecasting when given sufficiently long input windows, proving that foundation models effectively learn unique seasonal dynamics without weight updates.

Abstract: This work investigates the zero-shot forecasting capability of time-series foundation models for Leaf Area Index (LAI) forecasting in agricultural monitoring. Using the HiQ dataset (U.S., 2000-2022), we systematically compare statistical baselines, a fully supervised LSTM, and the Sundial foundation model under multiple evaluation protocols. We find that Sundial, in the zero-shot setting, can outperform a fully trained LSTM provided that the input context window is sufficiently long—specifically, when covering more than one or two full seasonal cycles. We show that a general-purpose foundation model can surpass specialized supervised models on remote-sensing time series prediction without any task-specific tuning. These results highlight the strong potential of pretrained time-series foundation models to serve as effective plug-and-play forecasters in agricultural and environmental applications.

Submission Number: 17