Code for: A Continuous Implicit Neural Representation Framework with Gradient Regularization for Sea Surface Height Reconstruction From Satellite Altimetry

This archive contains the source code used for the SIREN-TV implicit neural representation experiments in the manuscript. The code includes data-conversion notebooks for the OSE and OSSE benchmark configurations, training scripts for the INR-based SSH reconstruction with spatial-gradient TV regularization, and figure-generation scripts for the manuscript results. The OSE experiment follows the SSH MapMed Ocean Data Challenges benchmark using real multi-mission along-track altimetry observations, with CryoSat-2 new-orbit observations withheld for evaluation. The OSSE experiment follows the NATL60 Ocean Data Challenges benchmark, using sparse pseudo-altimetric observations as input and the full NATL60 SSH field as the gridded reference for evaluation.