An Improved Method for Phenology Model Parameterization Using Sequential Optimization

Abstract

Accurate prediction of peak bloom dates (PBD) of flowering cherry trees is critical for organizing local cherry festivals and other associated cultural and economic activities. A two-step phenology model is commonly used for predicting flowering time depending on local temperatures as a result of two consecutive steps followed by chill and heat accumulations. However, an extensive computation requirement for parameter estimation has been a limitation for its practical use. We propose a sequential parameterization method by exploiting previously unused records of development stages. With an extra constraint formed by heat accumulation between two intervening stages, each parameter can then be solved sequentially in much shorter time than the brute-force method. The result was found to be almost identical to the previous solution known for cherry trees (Prunus × yedoensis) in the Tidal Basin, Washington D.C.