Soil carbon and nitrogen stock of the Japanese agricultural land estimated by the national soil monitoring database (2015–2018)

Abstract

ABSTRACT Estimation of the soil organic carbon (SOC) stock based on the monitoring data is important to validate the carbon model estimates and provide the initial values of SOC stock to the model. The Japanese National Greenhouse Gas Inventory Report (NIR) has applied the RothC model since 2015 to report changes in SOC in agricultural land. This study estimated soil carbon and nitrogen stocks and C:N ratio of top 30 cm arable land in Japan using the 2015–2018 database of the national soil monitoring project. Estimates based on the stationary monitoring data from the top 30 cm of the soil layer arethe first public release. We modified categorization of the soil temperature regime and soil-type classification for estimation, expecting to mitigate the underestimation and overestimation of SOC stock in the areas of soil temperature 15–17°C and more than 17°C, respectively, and to optimize the soil classification to 16 fewer types to allow an appropriate estimation even in the current situation of decreasing number of survey sites. Statistical analysis to see the effects of land use and soil temperature on SOC stock, TN stock, and C:N ratio showed that soil types other than Andosols had an inverse temperature-dependent trend for all variables, while Andosols had an unclear trend in soil temperature, supporting existing studies. Also, when Lowland soils (Fluvisols or Anthrosols in the World Reference Base) were surrounded by Andosols, redeposition of volcanic ash could affect the temperature trend in SOC stock for Lowland soils. The distribution map of the SOC stock and C:N ratio across Japan visually showed a high value in the Andosols distribution areas and a temperature-dependent trend at the distribution areas of the other soil types. The total SOC stock estimated in this study was 176±12.8 TgC in paddy field, 100±8.6 TgC in upland field, 25±1.7 TgC in orchard, 78±8.9 TgC in grassland, and 379±17.9 TgC in total, the standard deviations of which were much smaller than those of the mean values for each land use calculated by the simple average method.