T-FORS ML model for LSTID forecasting over Europe
Last modified on Apr 4th, 2025, 16:04 UTC
Type
Description
The Machine Learning model, based on CatBoost (a gradient boosting framework) and trained on a human validated LSTID catalogue, uses a diverse set of physical drivers, ranging from geomagnetic indices, and solar wind and activity data, to ionosonde measurements.
Processing Input Parameters
- newell
- Coupling function
- ie
- Auroral electrojet IE
- iu
- Auroral electrojet IU
- Bz
- IMF Bz component measuread at L1
- vx
- Solar radial flux velocity measuread at L1
- rho
- Solar flux density measured at L1
- f_107
- Solar flux at 10.7 cm weavelenght
- hp_30
- Half-hourly geomagnetic index
- ie_variation
- IE time-derivative
- ie_mav_3h
- IE moving average (3 hr)
- ie_mav_12h
- IE moving average (12 hr)
- iu_variation
- IU time-derivative
- iu_mav_3h
- IU moving average (3 hr)
- iu_mav_12h
- IU moving average (12 hr)
- hf
- HF-INT (from TechTIDE)
- hf_mav_2h
- HF-INT moving average (2 hr)
- solar_zenit_angle
- Solar zenit angle
- dst
- Disturbance storm time index
- spectral_contribution
- Spectral energy contribution (from TechTIDE Ionosondes)
- velocity
- Perturbation velocity (from TechTIDE Ionosondes)
- azimuth
- Perturbation azimuth (from TechTIDE Ionosondes)
Data Levels
Resources and Further Information
Resources
Metadata File (XML)More Properties
| Property | Value |
|---|---|
| Algorithm | Not used |
| Software Reference | Not used |
Metadata Information
| Editor | Istituto Nazionale di Geofisica e Vulcanologia |
| Version | 1 |
| Created | Friday 4th April 2025, 16:04 |
| Last Modified | Friday 4th April 2025, 16:04 |