Comparing two models of the diurnal temperature cycle for simulations of crop and pest phenology under climate change

Summary

Air temperature is the most important determinant of crop and pest phenology. Developmental rates of organisms are thus usually modelled as a function of growing degree-days. These can be evaluated based on either daily or hourly temperatures. However, as climate model do not provide reliable information at high temporal resolution, statistical or empirical models are needed to generate hourly temperature data for assessing crop and pest phenology under climate change.
We present the comparison of two approaches for generating hourly temperatures from daily values of the minimum (Tmin) and maximum (Tmax) temperature. The statistical downscaling approach proposed by Hirschi et al. (2012) uses a nearest neighbour re-sampling procedure to prescribe realistic diurnal cycles constrained on the observed or simulated Tmin and Tmax. The model of Parton and Logan (1981) assumes a truncated sine wave for daytime and an exponential function for night-time temperatures. It was modified here to reduce the overestimation of daytime growing degree-days sums and obtain smooth transitions between two consecutive days.
The performance of the two approaches was tested with observed data and climate scenarios for Switzerland. To illustrate the potential for application, we discuss the the prediction of phenological stages of the codling moth (Stoeckli et al., 2012), a common orchard pest in Switzerland.