Effects of Cultivar Shifts on Winter Wheat Phenology Under Two Parameterization Methods

Phenology and growth duration of crops have been significantly changed by the combined effects of climate change and cultivar shifts. For the need of accurately evaluating the response of crops phenology to future climate changes, effects of cultivar shift on phenology and its quantitative simulation has become a research hotspot. However, most recent studies are based on the single parameterization method, with less attention paid to effects of different parameterization methods, leading to a certain degree of assessment uncertainty.Winter wheat phenology data and daily meteorological data in 47 agrometeorological observation stations in North China Plain during 1986-2010 are collected. Based on these datasets, a most commonly used phenology model is used to quantize effects of cultivar shifts on phenology, and effects of two parameterization methods on simulated results are also analyzed. The first method uses fixed three cardinal temperatures (Method 1), while in the second method (Method 2) three cardinal temperatures are obtained by minimizing the root mean square error of simulated phenology.Results show that winter wheat critical phenology in North China Plain changes significantly under the frequently change of cultivar during study period. Both two methods perform well in parameterizing the simulation of durations from turning green to heading and from heading to maturity in the winter wheat simulation. The growth duration is prolonged by cultivar shift in the duration from turning green to heading and the duration from heading to maturity, though values given by Method 1 are higher. Both methods indicate effects of cultivar shifts on the duration from heading to maturity is higher than those on the duration from turning green to heading. In addition, the range of simulated trends and their regional distribution are also affected by the different parameterization method used. Among them, the difference of simulation results between two methods in the duration from turning green to heading is higher than the duration from heading to maturity. In the regional distribution, the difference of simulation results between two methods is bigger in the duration from heading to maturity. It verifies that simulation results are potentially affected by parameterization method. Therefore, the selection of parameterization methods and uncertainties introduced by different methods should be carefully considered.