Modelling the course of frost tolerance in winter wheat: I. Model development

A Canadian model that simulates the course of frost tolerance in winter wheat under continental climatic conditions was adopted and further developed for use in an oceanic climate. Experiments with two cultivars were conducted during two winters in Central Norway. All plants were hardened at the same location. After hardening, in mid November, they were distributed to three locations with contrasting winter climates. Plants were sampled several times during autumn and winter and tested for frost tolerance, expressed as LT50 (the temperature at which 50% of the plants were killed). Results from the experiment were used in parameterization and cross validation of the new model, called FROSTOL, which simulates LT50 on a daily basis from sowing onwards. Frost tolerance increases by hardening and decreases by dehardening and stress, the latter caused by either low temperatures, or by conditions where the soil is largely unfrozen and simultaneously covered with snow. The functional relationships of the model are all driven by soil temperature at 2 cm depth. One of them is in addition affected by snow cover depth, and two of them are conditioned by stage of vernalization. Altogether five coefficients allotted to four of the functional relationships produced a good agreement (R2 = 0.84) between measured and modelled values of LT50. A cross validation of the model indicated that the parameters were satisfactorily insensitive to variation in winter weather.