Enhanced Neural Network Model for Worldwide Estimation of Weighted Mean Temperature

Precise modeling of weighted mean temperature (Tm) is critical for realizing real-time conversion from zenith wet delay (ZWD) to precipitation water vapor (PWV) in Global Navigation Satellite System (GNSS) meteorology applications. The empirical Tm models developed by neural network techniques have been proved better performances on the global scale; they also fewer model parameters and are thus easy operate. This paper aims further deepen research with network, expand application scope provide users more solutions acquisition Tm. An enhanced (ENNTm) has radiosonde data distributed globally. Compared other models, ENNTm some advanced features both design performance, Firstly, cover whole troposphere rather than just near Earth’s surface; secondly, ensemble learning was employed weaken impact sample disturbance performance elaborate preprocessing, including up-sampling down-sampling, which adopted achieve furthermore, designed meet requirements three different conditions providing sets parameters, i.e., estimating without measured meteorological elements, only pressure. validation work carried out using distribution, results show that compared competing perspectives under same conditions, proposed expanded estimation provided choices applications GNSS-PWV retrival.