blending/steps.py hardcodes compute_noise_stddev_adjs(conditional=True), whereas the standalone nowcasts/steps.py uses conditional=self.__config.conditional. Passing conditional=False to steps.forecast() means that the noise coefficients are sized over the rain area (because conditional is always true), while the field statistics use the whole domain.
Same conditional=True for Kalman filter implementation but Object of type EnKFCombinationConfig has no attribute conditional
blending/steps.pyhardcodescompute_noise_stddev_adjs(conditional=True), whereas the standalonenowcasts/steps.pyusesconditional=self.__config.conditional. Passingconditional=Falsetosteps.forecast()means that the noise coefficients are sized over the rain area (because conditional is always true), while the field statistics use the whole domain.Same conditional=True for Kalman filter implementation but
Object of type EnKFCombinationConfig has no attribute conditional