Sensitivity Analysis of the RothC Model Using Two Climatic Datasets: A Case Study of Arable Soils in Rostov Oblast

The Rothamsted Carbon (RothC) model was developed to simulate soil organic carbon dynamics under different types of land use, different climate conditions, and different agricultural technologies. The model has been effectively applied across different scales and in different environmental zones; however, it is sensitive to key input variables. This study estimates the sensitivity of soil carbon storage projections in response to variations in climate inputs using the RothC model. Two climate data sets were used for the sensitivity analysis: the Climatic Research Unit (CRU) dataset (CRU TS v4.05, 1901–2020), which has a spatial resolution of about 50 × 50 km² and is derived from interpolated data from ground-based weather stations, and TerraClimate database, which is built based on CRU data and enhances the resolution to around 4 × 4 km² through additional modeling and data inputs. Sequestration estimates for Rostov oblast have been derived using the CRU database. The accuracy of estimates for Rostov oblast is of particular importance, since the proportion of Chernozems in the total land structure is almost 65%. The sensitivity analysis has revealed that the replacement of the CRU data with TerraClimate meteorological data significantly underestimates the sequestration rate. On average, this underestimation was about 100 kg C ha^–1 year^–1 across all scenarios. For application in Rostov oblast, the TerraClimate database should be adjusted, particularly in the range of minimum precipitation values, using data from local meteorological stations or other models.