Modification of the RothC model for simulations of soil organic C dynamics in dryland regions

The Rothamsted Carbon Model (RothC), models the turnover of soil organic carbon (C) in non-waterlogged topsoils. It requires only easily obtainable inputs, and has been widely and successfully used in many regions of the world. However, it has been shown to have limitations when used to predict soil C dynamics in semiarid areas, where unrealistically high C inputs have been required to obtain good simulations. Soils of semiarid regions, highly vulnerable to soil C depletion, could represent a sink of C and become economically valuable in the C credit market. Hence, we introduced modifications to RothC with the aim of improving the prediction of soil C dynamics for semiarid regions. Modifications made were: i) the minimum value of the rate modifying factor for moisture was changed from 0.2 to 0.15 and 0.10 (indicated with the suffix 20, 15 and 10, respectively, in the name of the model); ii) the maximum soil moisture deficit was increased and the hydrological constants (field capacity, wilting point and hygroscopic water) were calculated using pedotransfer function by van Genuchten (1980) (RothC20, RothC15, and RothC10); and iii) soils that were bare were allowed to dry out more than in the current version of RothC (RothC_N, RothC20_N, RothC15_N, and RothC10_N). The modifications were tested on data from six experiments: five sites from semiarid regions and one from a temperate region. The modifications made in this study, which essentially reduce decomposition rate in soil, improved the model's performance for dry regions, especially when the rotation included a fallow. The model now requires more realistic C inputs to the soil, and one additional input (silt) that is normally measured with clay. These modifications do not impair the performance of the model under temperate conditions and so represent a broadening in the capability of the RothC model. Among all the various models compared, for simulation of SOC turnover we suggest the use of RothC10_N for dry regions as a part of global scale modelling.