An Empirical Model for Atmospheric Temperature Estimation Using a Geographic Grid Neural Network Framework

Atmospheric temperature (T) is a critical parameter in Global Navigation Satellite System (GNSS) tropospheric tomography, and facilitates the conversion of wet refractivity (N w) to water vapor density (ρ v). This study presents an empirical model for T estimation based on a Geographic Grid Neural Network (GGNN) framework. The proposed GGNNTemp model has neural networks with a 1°×1° geographic grid. Each geographic grid node employs a multilayer feedforward neural network with inputs of month and altitude and output of temperature. Comparative evaluations against the GPT2w model demonstrate the superior performance of the GGNNTemp model, achieving a BIAS value of 0.01 K and an RMSE of 1.60 K—significantly outperforming GPT2w’s 12.37 K RMSE. The model exhibits consistent accuracy across varying latitudes, altitudes (up to 10 km), and seasons, with ρ v conversion errors maintained below 5‰. This work highlights the efficacy of GGNN technology in modeling atmospheric parameters and advances its potential for broader applications in GNSS meteorology.