Characterize and mitigate long-lead coastal precipitation instabilities

Determine whether the unrealistically high, grid-point-scale precipitation patterns that emerge near coastlines in AICON forecasts beyond approximately 72 hours are confined to precipitation or also affect other forecast variables, and develop measures to alleviate these instabilities.

Background

AICON develops spurious, highly localized precipitation patterns near coastlines at long forecast lead times, particularly after approximately 72 hours. The affected grid points repeatedly produce excessive precipitation, creating the reported “rain-pox” behavior. Similar instabilities were observed in an earlier version of ECMWF’s AIFS and were mitigated there by reducing the training weight assigned to soil moisture.

The paper states that investigations into alleviating the AICON instabilities are ongoing and explicitly leaves unresolved whether their effects are limited to precipitation. Resolving the scope and cause of this instability is important for improving the reliability of long-range operational precipitation forecasts and for determining whether other forecast fields are also affected.