Abstract:Accurate prediction of atmospheric optical turbulence in localized environments is essential for estimating the performance of free-space optical systems. Macro-meteorological models developed to predict turbulent effects in one environment may fail when applied in new environments. However, existing macro-meteorological models are expected to offer some predictive power. Building a new model from locally-measured macro-meteorology and scintillometer readings can require significant time and resources, as well as a large number of observations. These challenges motivate the development of a machine-learning informed hybrid model framework. By combining some baseline macro-meteorological model with local observations, hybrid models were trained to improve upon the predictive power of each baseline model. Comparisons between the performance of the hybrid models, the selected baseline macro-meteorological models, and machine-learning models trained only on local observations highlight potential use cases for the hybrid model framework when local data is expensive to collect. Both the hybrid and data-only models were trained using the Gradient Boosted Decision Tree (GBDT) architecture with a variable number of in-situ meteorological observations. The hybrid and data-only models were found to outperform three baseline macro-meteorological models, even for low numbers of observations, in some cases as little as one day. For the first baseline macro-meteorological model investigated, the hybrid model achieves an estimated 29% reduction in mean absolute error (MAE) using only one days-equivalent of observation, growing to 41% after only two days, and 68% after 180 days-equivalent training data. The number of days-equivalent training data required is potentially indicative of the seasonal variation in the local microclimate and its propagation environment.

Application of Machine Learning Models for Convective Meteorological Events

New mean tropospheric temperature models based on machine learning algorithms for Brazil

Short-range forecasting system for meteorological convective events in Rio de Janeiro using remote sensing of atmospheric discharges

A new classification approach for detecting severe weather patterns

CME Arrival Modeling with Machine Learning

Exploring the Use of Machine Learning Weather Models in Data Assimilation

Predicting the Geoeffectiveness of CMEs Using Machine Learning

Machine Learning Investigation of Downburst Prone Environments in Canada

Approximation of a Convective-Event-Monitoring System Using GOES-R Data and Ensemble ML Models

Hybrid Optical Turbulence Models Using Machine Learning and Local Measurements

Comparative Analysis of Machine Learning Models for Tropical Cyclone Intensity Estimation

Lightning-Fast Convective Outlooks: Predicting Severe Convective Environments with Global AI-based Weather Models

Evaluation of a High Resolution WRF Model for Southeast Brazilian Coast: The Importance of Physical Parameterization to Wind Representation

Storm gust prediction with the integration of machine learning algorithms and WRF model variables for the Northeast United States

Utilizing Machine Learning Models with Limited Meteorological Data as Alternatives for the FAO-56PM Model in Estimating Reference Evapotranspiration

Statistical post-processing of multiple meteorological elements using the multimodel integration embedded method

AVIATION ACCIDENT AND INCIDENT FORECASTING COMBINING OCCURRENCE INVESTIGATION AND METEOROLOGICAL DATA USING MACHINE LEARNING

Assessment of Wind and Wave High-Resolution Forecasts During High-Energy Weather Events in the Brazilian Coast

Establishment and assessment of urban meteorological disaster emergency response capability based on modeling methods

Do AI models produce better weather forecasts than physics-based models? A quantitative evaluation case study of Storm Ciarán