Table 1 A summary of variable definitions used in this paper.
From: FuXi: a cascade machine learning forecasting system for 15-day global weather forecast
Variables | Definitions |
|---|---|
C, H, W | Channel dimensions, and spatial dimensions in latitude and longitude directions, respectively. |
D | A set containing all the forecast initialization times in the testing dataset. |
c, i, j, t0, τ | Indices for variables, latitude coordinates, and longitude coordinates, as well as forecast initialization time and forecast lead time steps added to t0, respectively. |
\({{{\mbox{X}}}}^{t},{\hat{{{\mbox{X}}}}}^{t},{{\mbox{M}}}\,\) | Ground truth and model predicted weather parameters at time step t, and climatological mean computed using ERA reanalysis data between 1993 and 2016. |
Z, R, T, U, V, TP, MSL | They represent geopotential, relative humidity, temperature, u component of wind, v component of wind, 6-hourly total precipitation, and mean sea-level pressure, respectively. |
