Open Access

Table 3

Comparison between flow fields vD,t and vref,t for the ARs dataset. RMSE is the root mean squared error (Eq. (1)), Eabs is the mean absolute error (Eq. (2)), Erel is the mean relative error (Eq. (3)), C is the correlation coefficient (Eq. (4)), A is the spatially-averaged normalized dot product (Eq. (5)) and S is the ratio between the integrals of the unsigned shear components of the DeepVel and reference Poynting fluxes (Eq. (9)).

Training set (ARs dataset) Quantity (units) |Bref,z| (G) RMSE Eabs Erel (%) C ∈ [0, 1] A ∈ [−1, 1] S (no units)
STAGGER (τ ≈ 1) vt (km s−1) ≥0 1.297 1.029 102.6 0.653 0.630 0.213
<100 1.264 1.006 78.44 0.692 0.676
≥100 1.604 1.267 353.2 0.166 0.155
MURaM-ARs (τ ≈ 1) vt (km s−1) ≥0 0.719 0.603 70.98 0.905 0.830 1.061
<100 0.701 0.592 56.02 0.912 0.855
≥100 0.878 0.717 226.2 0.845 0.563
MURaM-ARs (τ ≈ 0.1) vt (km s−1) ≥0 0.798 0.663 76.17 0.867 0.785 1.459
<100 0.793 0.662 61.51 0.878 0.824
≥100 0.840 0.678 217.9 0.639 0.410
MURaM-ARs (τ ≈ 0.01) vt (km s−1) ≥0 0.887 0.731 106.5 0.660 0.585 1.038
<100 0.880 0.728 91.03 0.680 0.619
≥100 0.956 0.759 253.9 0.415 0.261
MURaM-ARs (τ ≈ 0.001) vt (km s−1) ≥0 1.114 0.912 127.5 0.331 0.208 0.996
<100 1.104 0.913 118.6 0.281 0.199
≥100 1.214 0.893 225.9 0.580 0.306
MURaM-ARs (d = 1000 km) vt (km s−1) ≥0 0.961 0.851 152.9 0.258 0.164 1.343
<100 0.980 0.876 149.2 0.258 0.159
≥100 0.762 0.619 187.1 0.257 0.215

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.