On GOES-16 data, a deep ensemble plus MC dropout yields the best calibrated probabilistic convective initiation nowcasts among five Bayesian deep learning methods.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.ao-ph 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Bayesian Deep Learning for Convective Initiation Nowcasting Uncertainty Estimation
On GOES-16 data, a deep ensemble plus MC dropout yields the best calibrated probabilistic convective initiation nowcasts among five Bayesian deep learning methods.