arXiv subjects

C. Weinberger

Publications and source records attributed to C. Weinberger.

3 recordsLinked to original sources

arXiv2026

Establishing baseline model performances for optical turbulence forecasting

Accurate forecasting of optical turbulence (OT) is useful to optimise telescope scheduling and maximising observing efficiency, as the performance of adaptive optics systems strongly depends on turbulence conditions. While various numerical weather prediction and machine learning models are developed to predict these complex turbulence properties, properly evaluating their added value requires well-defined baseline reference models. This study establishes baseline performance thresholds for key astroclimatic parameters using two distinct models: one forecasting the parameter average over a specific timescale, and the other forecasting the temporal evolution over the same specific timescale. We apply these frameworks to extensive, multi-year datasets from the sites of the Very Large Telescope (VLT) and the Large Binocular Telescope (LBT) and discuss the possibility of extrapolating reference values of forecast performances for other sites based on their optical turbulence statistical characteristics. Our results show that baseline predictive accuracy varies according to the observing site, the specific parameter, the chosen timescale, and the forecasting approach. Therefore, an OT forecasting model's accuracy cannot be quantified in absolute terms, but relatively to the baseline.

astro-ph.IM↗

arXiv2026

Neural networks applied to forecasting the astroclimatic and meteorological parameters in an operational context

Forecasting optical turbulence and atmospheric parameters is crucial for science operations of present and new generation ground-based telescopes (ELT) as well as for free-space optical communication (FSOC). These forecasts are also useful to optimize the AO use, particularly in the ELT framework too. Recent research has demonstrated that the AR technique currently provides the best performances in predicting atmospheric and astroclimatic parameters on a time scale of 1 hour (e.g., RMSE 0.1" for the seeing), however deep learning offers some intriguing features that include the high adaptability, the correlation that can be found between different parameters treated simultaneously and different architectures can explore the capability for possible improvements. For this reason, it is important to investigate on the possibility to improve forecast performances beyond the score achieved by AR or simply to apply deep learning in those context in which it is not possible to apply the AR. In this contribution we investigate a set of algorithms belonging to the family of the neural networks that in principle are more suitable than the Random Forest (RF) model for this kind of application. We address specific questions related to the forecast performances in operational configurations. Besides that, we also investigate the impacts on gaps on the trained data set due to the lack of measurements of the optical turbulence during the day time. We show evidences that with measurements covering the whole 24-hours forecast performances might improve. Our test has been done on the wind speed in this contribution. This work is done in the framework of the operational forecast system for LBT (ALTA Center) and for the VLT (FATE project) that are leaded by our team.

astro-ph.IM↗

arXiv2026

Establishing baseline model performances for optical turbulence forecasting

Accurate optical turbulence (OT) forecasts help telescopes observe with maximum efficiency, as the performance of adaptive optics (AO) systems strongly depends on turbulence conditions. OT forecasting remains a challenging problem. A wide range of methods has been explored to address this problem, including numerical weather prediction (NWP) models, machine learning algorithms, and hybrid approaches, each varying in complexity, computational cost, and accuracy. Assessing the performance of forecasting methods is essential to benchmark them against simple, well-defined reference models, known as baselines, which establish reference thresholds. This paper aims to calculate these thresholds for the astroclimatic and atmospheric parameters most relevant for ground-based astronomy related to two of these baseline methods that are suitable for use in two different types of forecast, which are among the most relevant for application to the ground-based astronomy: the forecast of the average of a parameter on a defined timescale and the forecast of the temporal evolution of a parameter on a defined timescale. We apply these baseline methods to a dataset related to a rich statistical sample of many years encompassing key astroclimatic and atmospheric parameters above the sites of the Very Large Telescope and the Large Binocular Telescope, and we discuss the implications of these results and the possibility of extrapolating such values for general rules. We demonstrate that the predictive performance of each reference method depends on the observing sites, the parameter that is forecasted, the forecast timescale, and the type of forecast used. That means that it is meaningless to quantify the performance of OT forecasts in absolute terms, since they depend on the context.

astro-ph.IM↗