Evaluation of soil carbon dynamics after forest cover change in CMIP6 land models using chronosequences

Abstract Land surface models are used to provide global estimates of soil organic carbon (SOC) changes after past and future change land use (LUC), in particular re-/deforestation. To evaluate how well the capture decadal-scale SOC LUC, we first consistent comparison simulated time series LUC by six all which participated coupled model intercomparison project phase 6 (CMIP6) with chronosequences (SCCs). For this measurements adjacent plots at four high-quality data sites temperate tropical regions. We find that initial stocks differ among due different approaches represent SOC. Models generally meet direction reforestation cropland but amplitude rate vary strongly them. The normalized root mean square errors multi range from 0.5 0.8 across 0.1–0.7 when excluding outliers. Further, simulate losses deforestation for crop or grassland too slow lack harvest impacts an overestimation recovery on grassland. representation management, especially nitrogen levels is important drops abandonment forest regrowth. Crop fire management match dynamics more difficult quantify as SCC rarely report these events. Based our findings, identify strengths propose potential improvements applied simulating LUC.