A space-time observation system for soil organic carbon

In this paper we present a framework for a space-time observation system for soil organic carbon (STOS-SOC). We propose that the RothC model be embedded within the STOS-SOC which is driven by satellite-derived inputs, and readily available geospatial inputs, such as digital soil maps. In particular, advances in remote sensing have enabled the development of satellite products which represent key inputs into soil carbon models, examples being evapotranspiration and biomass inputs to soil which characterise space-time variations in management and land use. Starting from an initial calibrated base for prediction, as new observations are acquired data assimilation techniques could be used to optimise calibration algorithms and predicted model outputs. Here we present initial results obtained from the implementation of the proposed STOS-SOC approach to the 1445 km2 Cox Creek catchment located in northern New South Wales, Australia. Our results showed that use of satellite derived biomass inputs using MODIS satellite product (MOD17A3) improved the accuracy of simulations by 16 % compared to C inputs derived through other methods normally adopted in the spatialization of the RothC model. We further discuss the possibility of improving the capabilities of the STOS-SOC for future applications.