Understanding Atmosphere-Ocean-Land-Ice Interactions in the Earth System

Abstract

As numerical models of the Earth system become more sophisticated – in terms of number of component models and the complexity of physical processes simulated – it becomes more difficult to understand their biases. This is especially true for near-surface quantities such as 2-meter temperature and wind speed that are influenced by interface processes. This dissertation consists of four studies that address this difficult problem. All span multiple components of the Earth system and are global in scope, making use of global observational data sets and Earth system models (ESMs). Earth system models parameterize ocean surface fluxes of heat, moisture and momentum with empirical bulk flux algorithms, which introduce biases and uncertainties into simulations. We compare, for the first time, the effects of three different algorithms in both atmosphere and ocean model simulations using E3SM. Flux differences between algorithms are larger in atmosphere simulations (where wind speeds can vary) than ocean simulations (where wind speeds are fixed by forcing data). Surface flux changes lead to global scale changes in the energy and water cycles, notably including ocean heat uptake and global mean precipitation rates. Compared to the control algorithm, both the Coupled Ocean-Atmosphere Response Experiment (COARE) and University of Arizona (UA) algorithms reduce global mean precipitation and top of atmosphere radiative biases. Ocean barrier layers (BLs) separate the mixed layer from the top of the thermocline and are able to insulate the mixed layer from entrainment of cold thermocline water. Here, we provide the first global BL assessment in three ESMs. Compared to observations, models reproduce the global distributions as semipermanent features in some tropical regions and seasonal features elsewhere. However, model BLs are generally too thin in tropical regions and too thick in higher latitudes. BL thickness biases are related to atmosphere biases in the tropics, but at higher latitudes biases are dominated by ocean circulation errors. Global and regional water cycle is a crucial component of the Earth system, and numerous studies have addressed the individual components (e.g., precipitation). Here we assess, for the first time, if remote sensing and reanalysis data sets can accurately and self consistently portray the Amazon water cycle. This is further assisted with satellite ocean salinity measurements near the mouth of the Amazon River. Ensemble means, which are widely used for individual components, are found to produce large biases in water cycle closure. Closure is achieved with only a small subset of data combinations, which rules out the lower precipitation and higher evaporation estimates. The common approach of using the Obidos stream gauge (located hundreds of kilometres from the river mouth) to represent the entire Amazon discharge is found to misrepresent the seasonal cycle, and this can affect the apparent influence of Amazon discharge on tropical Atlantic salinity. Near-surface air temperature (SAT) over Greenland has important effects on mass balance of the ice sheet. Here, extensive in situ SAT measurements (~1400 station-years) are used to assess monthly mean SAT from sixteen global and regional products. Ice sheet-average annual mean SAT from different data sets are highly correlated in recent decades, but their long term means and trends differ enough to affect results of ESM evaluations. Compared with the best observational estimate, thirty-one ESM historical runs from the CMIP5 archive reach ~5 degrees Celsius for 1901–2000 average bias and have opposite trends for a number of sub-periods. In the course of the above studies, my other research contributions have included model development and evaluation activities for the DOE E3SM Coupled Model Version 1, analysis of subtropical cloud errors in the E3SMv1 Atmosphere Model, and analysis of Greenland ice sheet surface interface processes in the Regional Arctic System Model.