The GLACE-Hydrology Experiment: Advancing Understanding of Land Processes in Coupled Climate –Hydrological System

Soil moisture is a key variable that affects regional hydroclimate through land-atmosphere interactions and feedback processes. The Global Land Atmosphere Coupling experiments (GLACE) have highlighted an important role of soil moisture memory process in sub-seasonal to seasonal climate predictability. However, how land-atmosphere interactions affect the soil moisture memory is largely unknown. Here, we describe the GLACE-Hydrology experiment - a new set of climate model ensemble that is specifically aimed at studying the effects of land-atmosphere coupling on soil moisture memory and predictability problems. The GLACE-Hydrology experiment consists of: (1) land-coupled ensemble in which land surface is interactively coupled to the atmosphere using a standard AMIP type run, (2) land-uncoupled ensemble in which land-atmosphere coupling is removed/tampered using a AMIP type run with specified soil moisture climatology from the coupled simulation. Effects of land-atmosphere coupling on soil moisture memory and predictability are studied using offline land model simulations those are run with three hourly atmospheric forcing data from the above two experiments. We implemented the GLACE-Hydrology experiment in the Community Earth System Modeling framework (CESM), and performed 5-member ensemble runs from 1971 to 2015 using observed sea surface temperature data. The soil moisture memory is determined as the e-folding decay time in soil moisture lag-autocorrelation structure. We are employing various predictability measures, e.g. signal to noise ratio and hindcasts anomaly correlations with the observations to study the effects of land-atmosphere coupling on soil moisture predictability. An initial analysis of model ensemble suggests significant changes in soil moisture memory due to land-atmosphere coupling. The soil moisture memory decreases by 2 months in the southwestern United States, and it increases by 1.5 months in the upper Great Plains region. We find an inverse relationship between soil moisture memory and predictability changes in these two regions. A decreased soil moisture memory in the land-coupled ensemble contributes to an improved soil moisture monthly anomaly correlation hindcast skill for the 1979 to 2013 period in the southwestern Unites states, and vice-versa for the upper Great Plains region. Diagnoses of underlying climate processes including feedbacks of land-atmosphere interactions on precipitation, evapotranspiration, and soil moisture memory will be discussed. We also compare soil moisture memory estimates from other large ensemble climate model experiments, e.g. CESM-large ensemble, and ECHAM5-VIC AMIP large ensemble.