Performance of models for the beginning of sweet cherry blossom under current and changed climate conditions

Six phenological models, two simple forcing (F)-models and one sequential chilling/forcing (CF)-model, each with and without day length (DL)-term in the forcing approach were optimised (2001-2010) and validated (2011-2015) on very accurate blossoming data of an experimental sweet cherry orchard at Berlin-Dahlem (cultivar 'Summit'). In parallel, in 3 seasons (2011/2012-2013/2014) climate chamber experiments were performed in order to determine the end of dormancy for 'Summit', which is usually an unknown or uncertain parameter in phenological modelling. Additionally, in the season 2013/2014 an in situ climate change experiment on three trees in the sweet cherry orchard were arranged, which was used to validate the phenological models for distinctly warmer climate conditions at the experimental site. On the basis of our climate chamber experiments we quantified the chilling requirement of 'Summit' trees. Thus, we were able to identify a CF-model for the beginning of sweet cherry blossom which is mostly physiologically based and works well for current and for future climate conditions at the experimental site. This paper also shows how phenological models can fail under warmer climates, if either the model is too simple or the model parameters are wrong. Additional, we confirmed that phenological models with DL-term in the forcing approach clearly surpassed the conventional phenological models without this parameter. The reason for this behaviour is extensively discussed.