ALBERT

Description: Data assimilation combines observational information with numerical models taking into account the errors in both the observations and the model. In ensemble data assimilation the errors in the model state are dynamically estimated using an ensemble of model states. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. The coupled models simulate different compartments of the Earth system as well as their interactions. For example coupled atmosphere-ocean models like the AWI Climate Model (AWI-CM), simulate the physics in both compartments and fluxes in between then. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. Ensemble data assimilation methods can be applied with these model systems, but have a high high computing cost. To allow us to efficiently perform the data assimilation, the parallel data assimilation framework (PDAF) has been developed. I will discuss the application and challenges of coupled ensemble data assimilation on the examples of the data assimilative model system of AWI-CM coupled to PDAF and a coupled ocean-biogeochemical model consistent of the ocean circulation model MITgcm and the ecosystem model REcoM2.

Description: Coupled models simulate different compartments of the Earth system as well as their interactions. For example coupled ocean-biogoechemical models simulate ocean circulation, biogeochemical processes and the carbon cycle. Coupled atmosphere-ocean models like the AWI Climate Model (AWI-CM), simulate the physics in both compartments and fluxes in between then. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. Ensemble data assimilation methods can be applied with these model systems, however the need to compute an ensemble of model integrations strongly increases the already high computing cost of the models. To allow us to perform the data assimilation in supercomputers, the parallel data assimilation framework (PDAF) has been developed. I will discuss the application and challenges of coupled ensemble data assimilation with PDAF on the example of two different coupled model systems: the ocean-biogeochemical model MITgcm-REcoM and the atmosphere-ocean model AWI-CM.

Description: Coupled models simulate different compartments of the Earth system as well as their interactions. For example coupled atmosphere-ocean models like the AWI Climate Model (AWI-CM), simulate the physics in both compartments and fluxes in between then. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. Ensemble data assimilation methods can be applied with these model systems, but have a high high computing cost. To allow us to efficiently perform the data assimilation, the parallel data assimilation framework (PDAF) has been developed. I will discuss the application and challenges of coupled ensemble data assimilation on the example of the data assimilative model system AWI-CM coupled to PDAF.

Description: Efficient ensemble data assimilation with coupled models poses particular challenges due to the comp lexity of the model system and due to its high computational cost. On the methodological side, one h as to account for different time scales, but also distinct correlation lengths, of different model c ompartments like the ocean and the atmosphere. Computationally, one often has to deal with multiple program executables, a coupler software, observation handling for different model compartments, and a large number of processors required to compute a complex coupled model. I will focus on the computational aspects and discuss the steps required to build a highly scalable and flexible data assimilation system can be built on the basis of the Parallel Data Assimilation Framework (PDAF, http://pdaf.awi.de) using the example of the coupled climate model AWI-CM (Sidorenko et al., Climate Dynamics, 44 (2015) 757-780). AWI-CM consists of the finite-element sea ice-ocean model FESOM, which uses an unstructured model grid, and the model ECHAM6 for the atmosphere. The model coupling is implemented with OASIS-MCT and the model system consists of two separate executable programs for the ocean and atmosphere. Next to the implementation steps, the scalability of the assimilation system is discussed with a realistic configuration of AWI-CM. The high scalability is obtained by an online-connection strategy for the data assimilation system. First, the parallelization of the coupled model system is modified so that the coupled model can perform ensemble forecasts. Second, the analysis (solver) step is directly inserted into the time-stepping loops of each model compartment. Augmenting the coupled model in this online way, the ensemble information is kept in memory and transferred by parallel communication when necessary. Thus, one avoids the need to repeatedly write an ensemble of model fields into files and read them again for the analysis step. Further, the coupled model is only started once and there is no need to stop and restart the whole coupled model to compute the analysis step. Instead, the analysis step is performed in between time steps and is independent of the actual model coupler. These modifications of the model are supported by the framework structure of PDAF. In addition to the parallel online connection for the data assimilation system, the analysis step has to be parallelized. Here, the different model compartments are treated like parallel subdomains of the model. In this way, one can one can use the data assimilation algorithms provided by PDAF and can implement and perform the analysis step in analogy to uncoupled models. However, one has to take into account the different model grids and possible distinct ways in which the model compartments store their model fields. This results in a data assimilation system that can perform the assimilations both in-compartment (for weakly coupled assimilation) and cross-compartment (for strongly coupled assimilation).

Description: Data assimilation combines observational information with numerical models taking into account the errors in both the observations and the model. In ensemble data assimilation the errors in the model state are dynamically estimated using an ensemble of model states. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. The coupled models simulate different compartments of the Earth system as well as their interactions. For example coupled atmosphere-ocean models like the AWI Climate Model (AWI-CM), simulate the physics in both compartments and fluxes in between then. Data assimilation is used with coupled models to generate model fields to initialize model predictions, for computing a model state over time as a reanalysis, to optimize model parameters, and to assess model deficiencies. Ensemble data assimilation methods can be applied with these model systems, but have a high high computing cost. To allow us to efficiently perform the data assimilation, the parallel data assimilation framework (PDAF) has been developed. I will discuss the application and challenges of coupled ensemble data assimilation on the examples of the data assimilative model system of AWI-CM coupled to PDAF and a coupled ocean-biogeochemical model consistent of the ocean circulation model MITgcm and the ecosystem model REcoM2.

Description: We discuss how to build an ensemble data assimilation system using a direct connection between a coupled model system and the ensemble data assimilation software PDAF (Parallel Data Assimilation Framework, http://pdaf.awi.de). The direct connection results in a data assimilation program with high flexibility, efficiency, and parallel scalability. For this we augment the source code of the coupled model by data assimilation routines and hence create an online-coupled assimilative model. This first modifies the coupled model to be able to simulate an ensemble. Using a combination of in-memory access and parallel communication with the Message Passing Interface (MPI) standard we can further add the analysis step of ensemble-based filter methods, which compute the assimilation of observations, without the need to stop and restart the whole coupled model system. Instead, the analysis step is performed in between time steps and is independent of the actual model coupler that couples the different model compartments. This strategy to build the assimilation system allows us to perform both weakly coupled (in-compartment) and strongly coupled (cross-compartment) assimilation. The assimilation frequency can be kept flexible, so that the assimilation of observations from different compartments can be performed at different intervals. Further, the reading and writing of disk files is minimized. The resulting assimilative model can be run in the same way as the regular coupled model, but with additional parameters controlling the assimilation and with a higher number of processors to simulate the ensemble. Using the example of the coupled climate model AWI-CM that contains the FESOM model for the ocean and sea ice and ECHAM6 for the atmosphere, both coupled through the OASIS-MCT coupler, we discuss the features of the online assimilation coupling strategy and the performance of the resulting assimilative model.