Precipitable water vapor (PWV) is crucial for weather forecasting and climate change research. However, accurate PWV estimation is challenging, especially in the absence of measured meteorological data. In this study, we develop a conversion model based on the random forest algorithm, called RPWV, which can directly derive PWV from zenith total delay without measured meteorological data. The results indicate that the RPWV model demonstrates high accuracy and reliability for PWV retrieval across North America. Specifically, using global navigation satellite system PWV as a reference, the bias and root mean square (RMS) values for RPWV are 0.01 mm and 1.87 mm, respectively. Moreover, compared with the conventional linear model and backpropagation neural network model, the accuracy of the RPWV is improved by 81.3% and 13.4%, respectively. In contrast to radiosonde PWV, the bias and RMS values of the RPWV are 1.85 mm and 3.40 mm, respectively. This model provides a straightforward and efficient method for estimating PWV and has potential for weather forecasting and climate research in environments where meteorological data is scarce.
25 November 2025 in Measurement Science and Technology
As the primary contributor to the error budget in microwave-dependent geodetic techniques, tropospheric delay has emerged as a bottleneck in achieving precise positioning of global navigation satellite systems (GNSSs). Due to the dynamic nature of water vapor, effective global zenith total delay (ZTD) modeling is essential yet challenging, underscoring the need for robust solutions. Artificial intelligence methods provide an effective approach for capturing the spatiotemporal characteristics of ZTD for modeling. In this contribution, we employ stacking ensemble learning to develop a machine-learning-based global ZTD (MGZTD) model, which integrates four decision tree models and artificial neural networks, leveraging the global forecast system to implement real-time nonlinear compensation for the empirical ZTD model. Validation against ERA5-derived ZTD shows that the model’s optimization achieves a 73.2% accuracy improvement over the baseline without correction. When benchmarked against GNSS post-processed ZTD, the stacking algorithm outperforms base learner schemes, particularly in tropical coastal environments. Furthermore, the MGZTD model achieves a global mean root mean square (RMS) error of 18.9 mm, yielding improvements of 52.5%, 51.3%, and 51.6% over the state-of-the-art GPT3, PVoxel, and IGPZWD models, respectively. The proposed model can effectively simulate the subtle random time-varying and spatially complex distribution of ZTD, delivering accurate and stable prediction information for users. These indicate the promising potential of the MGZTD model in enhancing real-time GNSS wide-area positioning services and water vapor remote sensing.
1 January 2026 in IEEE Transactions on Geoscience and Remote Sensing