Machine-Learned Mapping Functions for the Asymmetric Troposphere

Accurate empirical tropospheric mapping functions (MFs) are essential for high-precision geodetic applications such as GNSS and VLBI. However, traditional MFs assume azimuthal symmetry or explicitly model gradients on regular grids, limiting their ability to represent real atmospheric condition. To address the limitation, this study designs a novel physics-inspired machine learning-based framework implemented with Multilayer Perceptron and develops global empirical asymmetric MFs on five years of ERA5-based ray-tracing data. Compared to the widely used GPT3 model, our models reduce global RMSE by up to 12.0% for hydrostatic and 2.8% for wet components and remain competitive or superior when GPT3 is used with gradient corrections. The improvements are particularly evident at low elevation angles, high-altitude sites, and over continental regions. To our knowledge, this work marks the first global machine learning-based empirical MF models that directly capture atmospheric asymmetry, offering a flexible and physically consistent alternative to traditional models in space geodesy.