GSSC Now Integrates Camaliot1 Datalabs for Troposphere and Ionosphere Modelling

980 views

In recent years, the GNSS infrastructure has experienced unprecedented growth, both in the space segment and on the ground. Nowadays, millions of Internet-of-Things (IoT) devices, including smartphones, rely on GNSS for accurate positioning. The abundance of these devices presents a remarkable opportunity for GNSS science, offering unprecedented spatiotemporal resolution.

To leverage the immense potential of this data, the European Space Agency (ESA) embarked on the “Application of Machine Learning Technology for GNSS IoT Data Fusion” (CAMALIOT1) project in partnership with RINA Consulting-Centro Sviluppo Material S.p.A, the Politecnico di Milano, Intelligentia srl, and Geomatics Research & Development SRL. Within ESA’s Innovation in Satellite Navigation program (NAVISP), the CAMALIOT1 project was conceived to address the challenges associated with accessing and processing vast volumes of data enhancing the analysis of GNSS data to foster the development of novel scientific applications.

CAMALIOT1 project has highlighted that the combination of IoT, Big Data, and Machine Learning technologies in the GNSS fields is ready to implement new services and products in different industrial sectors. For example, the possibility to realise new high-precision localization services could help the European industry to better manage the logistics operations of complex supply chains and also improve workers’ safety in remote and dangerous areas.

The first phase involved developing an Android app to collect raw GNSS data from smartphones. Over 12,000 users worldwide contributed their raw GNSS data through a crowdsourcing campaign. The second phase was to handle the vast amount of heterogeneous GNSS data generated by CAMALIOT1.

Besides, GSSC Now is an innovative digital platform created by the ESA GNSS Science Support Centre. This infrastructure encompasses data ingestion, processing, and analysis services, incorporating components from the IoT or Machine Learning (ML) models. By utilising this platform, the public can directly access GNSS Science Data from various IoT sources and take advantage of ML capabilities to generate new products.

CAMALIOT1 datalabs are now accessible from GNSS Now platform! They enable the characterisation of the ionosphere and troposphere using raw GNSS data from smartphones. The results of both analyses were validated against data derived from geodetic GNSS receivers and advanced ionospheric maps for Total Electron Content (TEC) in the case of ionospheric characterization.

Figure 1: Camaliot1 datalabs on GSSC Now platform.

Example of “Camaliot Troposphere Modelling” datalab:

This JupyterLab provides users with the ability to train and evaluate ML models for spatiotemporal troposphere prediction in specific regions: the southern part of California and the northern part of the Netherlands. Starting from historical IONEX files and weather data, the data facilitates the preparation of ML datasets and provides notebooks for modelling that can be applied to generate spatial prediction maps and forecast Zenith Wet Delay (ZWD) values one hour in advance. Tropospheric parameter determination supports weather forecasting on Earth.

California_average_1h_0.01_knn3d-At-2021-04-16-00.00.00_Tropo_prediction.png
Figure 2: California Troposphere spatial prediction map.
temporal_prediction.PNG
Figure 3: California spatiotemporal prediction example.

Example of “Camaliot Ionosphere Prediction” datalab:

This JupyterLab allows users to run ML models for spatiotemporal ionosphere prediction in the European region. Using RINEX observations, the datalab provides notebooks for generating spatial prediction maps or forecasting TEC values one day in advance. Characterizing the ionosphere is crucial for monitoring space weather, which is vital for satellite operations and communication.

60T-5T-0.10-KNN-At-2020-03-23-15.00.00.png
Figure 4: VTEC spatial prediction map.
60T-5T-0.10-KNN-CSV.png
Figure 5: VTEC spatial prediction text file. 
ionosphere_TEPM_prediction.PNG
Figure 6: VTEC temporal prediction text file.