Abstract
Phenology algorithms in crop growth models have inevitable systematic errors and uncertainties. In this study, the phenology simulation algorithms in APSIM classical (APSIM 7.9) and APSIM next generation (APSIM-NG) were compared for spring barley models at high latitudes. Phenological data of twelve spring barley varieties were used for the 2014–2018 cropping seasons from northern Sweden and Finland. A factorial-based calibration approach provided within APSIM-NG was performed to calibrate both models. The models have different mechanisms to simulate days to anthesis. The calibration was performed separately for days to anthesis and physiological maturity, and evaluations for the calibrations were done with independent datasets. The calibration performance for both growth stages of APSIM-NG was better compared to APSIM 7.9. However, in the evaluation, APSIM-NG showed an inclination to overestimate days to physiological maturity. The differences between the models are possibly due to slower thermal time accumulation mechanism, with higher cardinal temperatures in APSIM-NG. For a robust phenology prediction at high latitudes with APSIM-NG, more research on the conception of thermal time computation and implementation is suggested.
Highlights
Process-based crop models simulate dynamic and complex interactions between environment, genotype, and management factors
At Röbäcksdalen in 2017, the variation in days to anthesis among the varieties was less than the variation in days to physiological maturity
The soil moisture and phenology data at Röbäcksdalen suggest that barley development was rapid as the warm and dry conditions intensified 40 days after sowing in 2018
Summary
Process-based crop models simulate dynamic and complex interactions between environment, genotype, and management factors. Various algorithms and parameters in crop models simulate different plant and soil processes on several interactions and linkages. The processes related to phenology, dry matter accumulation and partitioning, and soil hydrology and chemistry are simulated with the various algorithms and parameters. Model algorithms and parameters are still simplifications of real systems [1], which makes crop models contain unavoidable systematic errors. At the user level, the quality of input data [2] and the choice of model parameterization approach [3] create further uncertainties. Since crop models are being extensively applied to a wide range of agricultural research questions and hypothesis testing, such as assessments of climate change effects [4], decision making and planning [5], 4.0/)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
