Recent hindcast experiments involving the use of apply_bias_correction have highlighted a bug where the CLI will use one of the forecast_error_ (bias) cubes as the forecast, and process and return this as the forecast cube. Proper handling of this situation should recognise forecast cube is missing.
Acceptance criteria:
replicate the behaviour via appropriate acceptance tests
add functionality to separate out forecast cube from forecast_error (bias) cubes based on the variable names
Recent hindcast experiments involving the use of
apply_bias_correction
have highlighted a bug where the CLI will use one of theforecast_error_
(bias) cubes as the forecast, and process and return this as the forecast cube. Proper handling of this situation should recognise forecast cube is missing.Acceptance criteria: