资源环境科技发展态势分析平台

To estimate the robustness of hydrologic models under projected future climate change, researchers test transferability between climatically contrasting observed periods. This approach can only assess the performance changes induced by altered precipitation and related environmental dynamics (e.g., greening under wet conditions), since the instrumental record does not contain temperatures or carbon dioxide levels that are similar to future climate change projections. Additionally, there is an inherent assumption that long-term persistence of changes in precipitation will not further impact catchment response. In this study, we undertake a series of virtual catchment experiments using an ecohydrologic model that simulates dynamic vegetation growth, nutrient cycling, and subsurface hydrology. These experiments explore a number of climate change scenarios. We compare simulations based on persistent altered climate states against simulations designed to represent historical periods with the same precipitation but limited time for ecohydrologic adaptation. We find that persistence of precipitation changes as well as increased temperature and elevated carbon dioxide levels can all substantially impact streamflow under drier future conditions. For wetter future scenarios, simulated differences in the flow regime were smaller, but there was still notable divergence in modeled low flows and other hydrologic variables. The results suggest that historical periods with equivalent precipitation statistics cannot necessarily be used as proxies for future climate change when examining catchment runoff response and/or model performance. The current literature likely underestimates the potential for nonstationarity in hydrologic assessments, especially for drier future scenarios.

Plain Language Summary Climate change is expected to impact catchment processes in the future, altering hydrologic response. This raises the issue that models and methods developed using past data may not be reliable for simulating future conditions. To understand the extent of this problem, researchers often test models under climatically contrasting past periods. For example, if a model calibrated under wet conditions performs well simulating a dry period with 10% less annual rainfall, we assume that the same model could reliably simulate a future climate with a 10% reduction in rainfall. However, future climate change scenarios also have higher temperatures and carbon dioxide (CO2) levels, and the changes are expected to persist over many decades. Therefore, simulating contrasting past periods may cause less degradation in model performance than simulating future climate change. This study uses a detailed ecohydrologic model to run experiments analogous to a catchment under persistent climate change versus historical variability. We find that persistence of altered conditions, higher temperatures, and increased CO2 levels all influence catchment response substantially, meaning that climatically distinct past periods are not accurate proxies for future climate change. This suggests that testing model performance under past variability will underestimate the risk that climate change poses to hydrologic assessments.