PWVFnet: A Short-Time Troposphere PWV Forecast Model Combined Empirical Models and Spatiotemporal ConvLSTM Network

Precipitable water vapor (PWV) plays a crucial role in extreme weather forecasting, early warning, and correcting signal delays for satellite earth observations. To obtain future PWV information, a short-time troposphere PWV forecast model combined empirical model and spatiotemporal convolutional long short-term memory network (ConvLSTM) network is developed based on ERA5 PWV data from 2019 to 2022 to forecast PWV in the next several hours. First, the Fourier transform is introduced to construct an empirical model for the periodic variations in the PWV time series, enhancing the interpretability of the forecast model. The periodic component is removed from the ERA5 PWV to obtain the residual term without obvious patterns. A ConvLSTM deep learning network is employed to learn its spatiotemporal variations and build a residual forecast model. Finally, the empirical and ConvLSTM models are combined to obtain the short-term tropospheric PWV forecast model named PWVFnet. The forecasted PWV is validated using 2023 ERA5 PWV data. The numerical results show that the root-mean-square error, standard deviation, and mean absolute error values of the 1-step forecasted PWV achieve 2.54, 2.53, and 1.82 mm, respectively. Similar results are presented to those of GNSS PWV. The model is further validated during quiet and active tropospheric periods, with forecast accuracies of 1.88 and 3.58 mm, respectively. The model’s accuracy is better than 3 mm when the forecast horizon does not exceed 6 h. Interestingly, it is found that intense tropospheric changes hinder the precise forecasting of ConvLSTM, but reducing the forecast region can mitigate the adverse effects caused by complex tropospheric variations.

• Publication year

  2026

• Venue

  IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

• Publication date

  Unknown publication date

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1109/JSTARS.2025.3648114
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Research on the PWV prediction model based on the ERA5-PWV calibration and WOA-RNN-BiLSTM-multihead-attention fusion algorithms

  2025cited by this paper
• GNSS Precipitable Water Vapor Prediction for Hong Kong Based on ICEEMDAN-SE-LSTM-ARIMA Hybrid Model

  2025cited by this paper
• Water vapor content prediction based on neural network model selection and optimal fusion

  2024cited by this paper
• Reconstruction of global ionospheric TEC maps from IRI-2020 model based on deep learning method

  2024influential reference
• Joint prediction of the capacity and temperature of Li-ion batteries by using ConvLSTM Network

  2024cited by this paper
• Multi-input Unet model based on the integrated block and the aggregation connection for MRI brain tumor segmentation

  2023influential reference
• Radar precipitation nowcasting based on ConvLSTM model in a small watershed in north China

  2023cited by this paper
• A Combined Model to Predict GNSS Precipitable Water Vapor Based on Deep Learning

  2023influential reference
• Investigating the Inter-Relationships among Multiple Atmospheric Variables and Their Responses to Precipitation

  2023cited by this paper
• A storm-time ionospheric TEC model with multichannel features by the spatiotemporal ConvLSTM network

  2023cited by this paper
• Global forecasting of ionospheric vertical total electron contents via ConvLSTM with spectrum analysis

  2022cited by this paper
• Estimation of diurnal-provided potential evapotranspiration using GNSS and meteorological products

  2022cited by this paper
• Improvement of AI forecast of gridded PM2.5 forecast in China through ConvLSTM and Attention

  2022cited by this paper
• Evaluation and Calibration of MODIS Near-Infrared Precipitable Water Vapor over China Using GNSS Observations and ERA-5 Reanalysis Dataset

  2021cited by this paper
• A new zenith hydrostatic delay model for real-time retrievals of GNSS-PWV

  2021cited by this paper
• Detecting heavy rainfall using anomaly-based percentile thresholds of predictors derived from GNSS-PWV

  2021cited by this paper
• A Global Assessment of Precipitable Water Vapor Derived From GNSS Zenith Tropospheric Delays With ERA5, NCEP FNL, and NCEP GFS Products

  2021cited by this paper
• Exploring the Relationships of Atmospheric Water Vapor Contents and Different Land Surfaces in a Complex Terrain Area by Using Doppler Radar

  2021cited by this paper
• Forecasting Precipitable Water Vapor Using LSTMs

  2020influential reference
• Hourly PWV Dataset Derived from GNSS Observations in China

  2019cited by this paper
• A new global grid model for the determination of atmospheric weighted mean temperature in GPS precipitable water vapor

  2019influential reference
• Predicting precipitable water vapor by using ANN from GPS ZTD data at Antarctic Zhongshan Station

  2019cited by this paper
• Retrieving Precipitable Water Vapor From Shipborne Multi‐GNSS Observations

  2019cited by this paper
• PyTorch: An Imperative Style, High-Performance Deep Learning Library

  2019cited by this paper
• On the suitability of ERA5 in hourly GPS precipitable water vapor retrieval over China

  2019cited by this paper
• Consistency Evaluation of Precipitable Water Vapor Derived From ERA5, ERA‐Interim, GNSS, and Radiosondes Over China

  2019influential reference
• Assessments of GMI-Derived Precipitable Water Vapor Products over the South and East China Seas Using Radiosonde and GNSS

  2018cited by this paper
• Studies of precipitable water vapour characteristics on a global scale

  2018cited by this paper
• On the Statistical Significance of Climatic Trends Estimated From GPS Tropospheric Time Series

  2018cited by this paper
• Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

  2015cited by this paper
• Investigation of the 16-year and 18-year ZTD Time Series Derived from GPS Data Processing

  2015cited by this paper
• A hybrid LS-HE and LS-SVM model to predict time series of precipitable water vapor derived from GPS measurements

  2015cited by this paper
• Empirical Evaluation of Rectified Activations in Convolutional Network

  2015influential reference
• Improved one/multi-parameter models that consider seasonal and geographic variations for estimating weighted mean temperature in ground-based GPS meteorology

  2014influential reference
• Adam: A Method for Stochastic Optimization

  2014cited by this paper
• Generating Sequences With Recurrent Neural Networks

  2013cited by this paper
• Global cooling after the eruption of Mount Pinatubo: a test of climate feedback by water vapor.

  2002cited by this paper
• GPS Meteorology: Mapping Zenith Wet Delays onto Precipitable Water

  1994influential reference

• No citing papers are available for this paper.