Coupled data assimilation (DA) methods are at the forefront of Earth system modeling and prediction. The overarching goal of this proposed project is to conduct a highly collaborative effort to 1) Build an effective land-atmosphere-ocean coupled DA capability for the DOE’s Energy Exascale Earth System Model (E3SM) with NSF NCAR’s Data Assimilation Research Testbed (DART) ensemble-based DA system; 2) Establish coupled DA to ensure well-balanced initial conditions with reduced spin-up for the coupled model system; 3) Examine the influence of the developed coupled DA on improving subseasonal-to-seasonal (S2S) predictability and hindcast simulations with E3SM; and 4) Use the improved hindcast simulations to understand the predictability at the S2S scale, focusing on extreme weather and climate events.

As the foundation of this project, the DART ensemble-based DA system is a community facility for ensemble DA with the ability to assimilate conventional and satellite observations. DART provides a suite of nonlinear and non-Gaussian ensemble assimilation tools that are unique in their ability to provide high-quality Earth system DA.  A prototype DA capability for the E3SM atmosphere model with DART (E3SM-DART) was developed by a project team member recently with promising results. The PI team also has extensive experience in implementing strongly coupled land-atmosphere DA for numerical weather prediction models, proving that strongly coupled land-atmosphere DA outperforms weakly-coupled land-atmosphere DA.

Based on the current work on DART DA with E3SM, we will optimize the system to enhance the efficiency of assimilating atmospheric observations at the standard resolution of E3SM (at 100 km). Then, we will extend the efforts to the high-resolution version of E3SM (at 0.25 degrees), including the regional-refinement capability. Based on the high-resolution model, we will implement strongly coupled land-atmosphere DA with simultaneous assimilation of satellite-derived soil moisture data (e.g., from SMAP) and atmospheric observations. Then, we will implement a capability to assimilate ocean SSTs, soil moisture, and atmosphere observations all together in a strongly coupled DA framework. This will create a framework for future coupled DA in E3SM. We anticipate the developed system to significantly reduce the cost of coupled model spin-up by using strongly coupled DA initial conditions.

With the coupled DA system, we will identify strategies for improving S2S prediction and understanding S2S predictability. The specific study region will be the Maritime Continent and the western Pacific Ocean, with an emphasis on studying the impact of the MJO and ENSO on precipitation anomalies (drought or floods) over the continental United States. We will also study how African Easterly Waves (AEWs) influence the probability of tropical cyclones over the Atlantic Ocean. Specifically, we will develop strategies to generate high-fidelity hindcast simulations and address the new challenges in the development of high-resolution coupled E3SM models. We will also use coupled DA and high-resolution E3SM models to identify the most effective methods for optimizing the use of observations, thus enabling precise and efficient high-resolution model simulations. With improved high-resolution model simulations, we will explore the extent to which coupled DA can enhance S2S predictability (reduce model biases) and how these enhancements are influenced by land-atmosphere and ocean-atmosphere interactions.

The outcomes of this project should lead to a new coupled DA system framework for the E3SM model, speed up initialization for the high-resolution model (~0.25-degree resolution), enhance understanding of the interactions among atmosphere, ocean, and land at S2S timescales, as well as provide effective methods for bias correction (e.g., reducing uncertainties in model simulations) which will improve climate projection for extreme events.