Phenological models for the beginning of apple blossom in Germany

Five phenological models (M1-M5) were examined with respect to their suitability to calculate the beginning of apple blossom in Germany, the most important fruit crop in Western Europe. Blossoming is the most sensitive period, e. g. to frost, and determines the fruit set of the apple trees. Phenological observations and temperature data from the German Weather Service in the period 1961-2005 were used to fit these five models. For the calculations data from, 5,630 phenological and 523 temperature stations were attributed to a 10 km x 10 km grid using second order universal kriging. Model parameters were optimised on 3,672 grid points for the nationwide approach for Germany and on 148 points for 11 fruit growing regions. Root mean square errors (RMSE) between modelled and observed apple blossom data varied from 4.2 to 5.0 days for the internal and from 4.6 to 5.6 days for the external verification on the basis of phenological records from three fruit growing research centres. The very simple statistical model approach M5 had the advantage of causing the least effort to calculate the bloom date, but it never performed better than any of the best mechanistic models M1-M4. Also, the 'thermal time model' M1 and the sequential model M2 were both easy to handle which makes them a preferable choice for predictions and management decisions in apple orchards. These two models M1 and M2 are also suitable to be implemented in yield models and water budget models in order to replace the use of fixed developmental stages by dynamical calculations. The two combined chilling/forcing models M2 - a sequential model - and M3 - a parallel model - exhibited the lowest average RMSE. Both models (M2 and M3) could preferably be used to project the impact of climate change on the beginning of apple blossom, since these models can compensate a possible lack of chilling by a higher demand for forcing. The present study showed that a) all five models were able to calculate the beginning of blossom for a wider range of apple cultivars, and b) the beginning of apple blossom has moved forward since 1989 due to climate change. The successful modelling of the beginning of apple blossom can be transferred to other fruit crops, which are more sensitive to frost events such as pear, apricot, etc.