Error analysis of soil temperature simulations using measured and estimated hourly weather data with 2DSOIL

Many crop simulation models use 1-h time steps for atmospheric, soil and plant processes but often meteorological data are only available as daily summaries. The objective of this study was to investigate how errors in estimation of hourly values of solar radiation and air temperature affect errors in simulation of soil temperature using the model 2DSOIL. 2DSOIL is a two-dimensional finite element model that simulates water flow, chemical, water and solute uptake by plant roots, chemical equilibria processes, and gas and heat transport in soil. The standard deviations of the differences between hourly estimated air temperatures were about 2degreesC and 85 W m(-2) for solar radiation. The mean difference in simulated and measured soil temperatures using measured hourly weather data for all depths at both sites was less than 1degreesC and standard deviations were about 1-3degreesC indicating low bias. The range of errors was highest in the surface soils when the soil was wetted after rainfall. Relative to simulated soil temperatures using measured hourly data, simulated soil temperatures using estimated data were, on average over all depths, 2degreesC lower and standard deviations ranged from 2 to 3degreesC. The errors were similar over all depths. Use of estimated hourly air temperature and radiation generally resulted in underpredictions of soil temperature by 2-3degreesC and increased error. Also maximum daily soil temperatures were underestimated. Published by Elsevier Science Ltd.