Modeling soil temperature to predict emergence

For annual field crops, the early development phase is crucial in establishing potential crop yield and thus to achieve high yield. Despite its importance, this phase is often neglected in current crop models that focus on later stages and on how plants produce biomass and allocate it to the different organs after emergence. Modeling stand establishment requires assessing accurately enough the conditions of the seedbed, the top 10 cm of the soil profile that serve as growing medium for germinating seeds and plantlets, which drive early development. In this zone, atmospheric conditions generate intense energy and water fluxes, results of biophysical phenomena at the soil-atmosphere interface, and consequently rapid and dramatic changes in temperature and water content that are not yet completely understood. Thus, our goal is to propose a modeling approach to improve the estimation of emergence timing by a simulation of the dynamics of seedbed temperature in response to a few standard meteorological factors that can be routinely acquired: global radiation, air temperature, relative humidity, wind speed and precipitation. The use of these five meteorological factors will allow an easier deployment and use of this model at large scale. We developed a soil model that uses the energy balance approach for the soil surface and a system of coupled partial differential equations for heat and water diffusion in the soil. We made the assumption of a homogeneous soil, without vegetation or residue cover. Our results show that this model is able to capture temporal and spatial variations of the soil temperature and the soil water content. Moreover, this model provides the most critical input data for models predicting the dynamics of emergence. Using data of maize (Zea Mays L.), we evaluate three thermal time models with respectively one, two and three parameters to predict the timing to 50 % of emergence and show that the model with one parameter corresponds to more heat-tolerant varieties and the one with three parameters is more adapted to sensitive cultivars. Finally, we choose the best thermal time model for each cultivar and use the simulated soil temperature to predict emergence on an independent validation dataset. This new emergence model should contribute to the improvement of emergence timing prediction, a key step in reducing the uncertainty in yield prediction.