Open Access
Table 1
The comparisons of prediction efficiencies of ≥2 MeV electron hourly fluences of different models.
| Model | Our transformer model | Persistence model | LSTM model by Wei et al. (2018) | MLP model by Son et al. (2022) | MLP model by Shin et al. (2016) |
|---|---|---|---|---|---|
| PE | 0.961 | 0.955 | 0.900 | 0.970 | 0.960 |
| RMSE | 0.2136 | 0.2155 | 0.2855 | – | – |
| Model input | Flux + MLT | ≥2 MeV electron hourly fluence 1 h ago | Flux + Kp + R0 | Flux + Vsw + N + T + |B| + Bz + Dst + Kp | A total of 17 input parameters |
| Model output | ≥2 MeV electron hourly fluence | ≥2 MeV electron hourly fluence | ≥2 MeV electron hourly fluence | ≥2 MeV electron hourly fluence | ≥2 MeV electron hourly fluence |
| Satellites | GOES-10 | GOES-10 | GOES-8, GOES-10, GOES-11, GOES-13 | GOES-15, GOES-16 | GOES-15 |
| Test time | 2005 | 2005 | 2008 | October to December from 2011 to 2021 | 2014 |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.