Assessing the Dynamic Versus Thermodynamic Origin of Climate Model Biases. Issue 16 (18th August 2018)

• Record Type: Journal Article
• Authors: Wehrli, Kathrin; Guillod, Benoit P.; Hauser, Mathias; Leclair, Matthieu; Seneviratne, Sonia I.
• Is Part Of: Geophysical research letters. Volume 45:Issue 16(2018)
• Page Start: 8471
• Page End: 8479
• Publication Date: 2018-08-18
• Subjects: climate model biases -- Earth system models -- atmospheric nudging -- dynamics versus thermodynamics -- global climate models -- systematic biases; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2018GL079220
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Global climate models present systematic biases, among others, a tendency to overestimate hot and dry summers in midlatitude regions. Here we investigate the origin of such biases in the Community Earth System Model. To disentangle the contribution of dynamics and thermodynamics, we perform simulations that include nudging of horizontal wind and compare them to simulations with a free atmosphere. Prescribing the observed large‐scale circulation improves the modeled weather patterns as well as many related fields. However, the larger part of the temperature and precipitation biases of the free atmosphere configuration remains after nudging, in particular, for extremes. Our results suggest that thermodynamical processes, including land‐atmosphere coupling and atmospheric parameterizations, drive the errors present in Community Earth System Model. Our result may apply to other climate models and highlight the importance of distinguishing thermodynamic and dynamic sources of biases in present‐day global climate models. Plain Language Summary: Global climate models have become indispensable tools to simulate past, present, and future climate. However, present‐day models display systematic errors, which are found, for example, in the simulation of surface temperature and precipitation. In this study, we demonstrate that the origin of climate model biases can be more closely isolated, by distinguishing between those due to dynamic processes, that is, related to atmospheric circulation, including wind and pressure systems, and those due to thermodynamic processes, that is, energy exchanges and phase changes, in particular, related to land‐atmosphere interactions and convective processes. To separate these two contributions, we prescribe observed winds into simulations with a global climate model, thereby forcing it toward reproducing the correct atmospheric circulation. Strikingly, the largest part of the temperature and precipitation biases are still found when the winds are corrected. This highlights that thermodynamic processes play a dominant role for biases in the representation of today's climate in the employed model. Because the model has similar biases as other climate models, it is likely that these conclusions would also apply there. Hence, in order to make more precise climate projections, it is important that the representation of thermodynamic processes is further improved by the climate modeling community. Key Points: Thermodynamical versus dynamical sources of biases can be identified using atmospheric nudging of horizontal winds in a climate model Atmospheric nudging improves simulated temperature and precipitation in CESM; however, more than half of the initial bias remains For temperature and precipitation extremes the larger part of the biases remains after correcting for the dynamics