Forecasting Precipitable Water Vapor Using LSTMs

Long-Short-Term-Memory (LSTM) networks have been used extensively for time series forecasting in recent years due to their ability of learning patterns over different periods of time. In this paper, this ability is applied to learning the pattern of Global Positioning System (GPS)-based Precipitable Water Vapor (PWV) measurements over a period of 4 hours. The trained model was evaluated on more than 1500 hours of recorded data. It achieves a root mean square error (RMSE) of 0.098mm for a forecasting interval of 5 minutes in the future, and outperforms the naive approach for a lead-time of up to 40 minutes.

[1]  Lukás Burget,et al.  Extensions of recurrent neural network language model , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[2]  Y. S. Meng,et al.  On the Importance of PWV in Detecting Precipitation , 2018, 2018 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium).

[3]  Thomas Fischer,et al.  Deep learning with long short-term memory networks for financial market predictions , 2017, Eur. J. Oper. Res..

[4]  Stefan Winkler,et al.  Predicting GPS-based PWV Measurements Using Exponential Smoothing , 2019, 2019 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium).

[5]  Frederick R. Forst,et al.  On robust estimation of the location parameter , 1980 .

[6]  Shilpa Manandhar,et al.  A Data-Driven Approach for Accurate Rainfall Prediction , 2019, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Stefan Winkler,et al.  Design of low-cost, compact and weather-proof whole sky imagers for high-dynamic-range captures , 2015, 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).