Improved JRA-3Q Soil Temperature in Permafrost Regions

Abstract Reanalysis is a valuable potential data source for permafrost studies. The latest-generation reanalysis of the Japanese Reanalysis for three quarters of a century (JRA-3Q) benefits from improved snow and soil schemes and demonstrates encouraging performance for soil temperature in permafrost regions compared to its predecessor, JRA-55, and other state-of-the-art reanalyses. We find JRA-3Q to have an overall mean annual air temperature bias of −0.17°C, with −0.55°C in permafrost regions. The snow depth was underestimated by −5.5 cm. In permafrost regions, the mean annual ground temperature bias was about −0.09°C. The estimated permafrost area from JRA-3Q is between 10.8 and 15.8 × 10 6 km 2 . The active layer thickness is substantially overestimated by about 0.65 m. The JRA-3Q soil temperature exhibits a pronounced warm bias in Alaska, which is very likely due to the overestimated snow insulation and simplified soil organic content. The decoupled energy conservation parameterization (DECP) method employed in the JRA-3Q soil scheme restricts its suitability for the interpretation of detailed permafrost phenomena, such as zero-curtain effects. This DECP method is used in many state-of-the-art land surface models; our results demonstrate the need for additional contributions to improve the representation of permafrost-specific processes. Significance Statement Permafrost acts as a critical tipping element in the Earth system due to the positive feedback between permafrost carbon and climate warming. Despite this importance, the availability of in situ observations is limited by harsh, remote environments. Reanalysis provides a valuable alternative to observations. JRA-3Q, the latest generation reanalysis with improved snow and soil schemes, performs better on soil temperature in permafrost regions compared to its predecessor (JRA-55) and other state-of-the-art reanalyses. Permafrost areas estimated from JRA-3Q air and soil temperatures are within the range of other estimates from the literature, suggesting that JRA-3Q is suitable for broad-scale permafrost applications. However, the application of more detailed permafrost phenomenon requires additional effort because of limitations inherent in how the model simulates soil freezing behavior.