A probabilistic seeing-forecast system for the 10.4m Gran Telescopio Canarias: methodology, validation, and operational integration

Lombardi, G.; Cuevas Ahumada, O.
Bibliographical reference

Adaptive Optics Systems X

Advertised on:
8
2026
Number of authors
2
IAC number of authors
1
Citations
0
Refereed citations
0
Description
We present a probabilistic seeing forecast system developed for the 10.4m Gran Telescopio Canarias (GTC) to support dynamic operations. The model combines numerical weather prediction, empirical calibration of turbulence strength, and a machine-learning post-processing layer (MOS) trained on TNG DIMM and GTC meteorological data. In addition to predicting deterministic values, the system provides probability distributions over discrete seeing intervals with hourly resolution up to 48 hours ahead, enabling direct application to queue scheduling and AO planning. The physical component applies wind-dependent calibrations, while the MOS layer corrects residual systematic errors using local meteorological features. Validation is performed through a back-casting framework on two years of data. The final model achieves a median absolute error of 0.16" and RMSE of 0.33". The system is currently integrated into the GTC operations, while nowcasting features are under development.