• A Kalman filter expanded fusion (KFEF) method integrates GRNN and Kriging interpolation for high-accuracy spatiotemporal temperature prediction.
• KFEF achieves 61.54% RMSE reduction compared to RBF, and 34.21% and 32.43% reductions relative to STK and GP, respectively.
• The method enables dynamic temperature field reconstruction along ranging paths with limited sensors, critical for atmospheric refraction correction.
• The framework is validated through simulation, indoor, and kilometer-scale outdoor experiments, showing strong potential for long-distance high-precision ranging and other applications.