Shaping Fine-Tuning of Geospatial Foundation Models: Effects of Label Availability and Temporal Resolution

Key Points

• Increased label availability enhances the model's performance significantly, with a notable reliance on temporal resolution.
• Analysis of geospatial models indicates that temporal resolution plays a crucial role in optimizing fine-tuning outcomes.
• The study examines various factors influencing fine-tuning in models, highlighting the importance of data and timing.
• Findings suggest that improving label availability and adjusting temporal resolution can lead to more effective model calibrations.

Abstract