Textual Environmental Context and Spatial Graphs for LLM-Based Regional SST Forecasting

Key points

• We study how these heterogeneous conditions can be presented to a large language model (LLM) for regional multi-step forecasting without serializing the full SST grid as text.1
• We formulate forecasting as conditional numerical generation: historical SST and anomaly sequences, date-aligned environmental records, and static ocean knowledge form a textual context, while regional spatial state is supplied through continuous graph-derived prefixes.1
• A static graph encodes persistent geographic--climatological relations, and a dynamic graph encodes recent SST correlations and localized tropical-cyclone influence.1
• On SST forecasting in the South China Sea, the complete configuration achieves the best MAE and $\Rtwo$ among the compared methods over ten forecast steps.1

Sources (1)

• [1]Textual Environmental Context and Spatial Graphs for LLM-Based Regional SST Forecasting
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 07:42 AM

  Sea surface temperature (SST) forecasting depends on local temporal persistence, regional spatial dependence, and environmental conditions that evolve with the forecast date.

  We study how these heterogeneous conditions can be presented to a large language model (LLM) for regional multi-step forecasting without serializing the full SST grid as text.

Extractive summary: sentences quoted from the sources.