E2.1 ELM v2 Data Architecture

                    

Poster TitleA new data architecture for ELM v2: implications for coupling and simulation of disturbance
AuthorsPeter Thornton, Teklu Tesfa, Michael Brunke
First AuthorPeter Thornton
Session TypeE3SM Session
Session IDE2
Submission TypePoster
GroupLand
Experiment
Poster Link




Abstract

A new data architecture is being implemented for ELM v2, which introduces a new level in the nested sub-grid hierarchy, standardizes the science basis for coupling to other E3SM components, and provides a foundation for new functionality that represents the influence of time-since-disturbance on states and fluxes tracked at the level of the soil column. The first stage of this development work, now complete, has standardized the topology information stored for each level of the ELM sub-grid hierarchy, and has introduced the topographic unit as a new sub-grid level. That creates a hierarchy arranged as: gridcell - topounit - landunit - column - vegetation (pft) - cohort (used by FATES). In ELM v1, near-surface weather forcings from the atmosphere to the land were delivered at the gridcell level, with some downscaling applied to some fields to produce column-level quantities, in support of the dynamic land ice model. In ELM v2, all surface weather forcing is delivered from the atmosphere to the land at the level of the topographic unit (topounit), where differences in elevation, slope and aspect can be downscaled consistently, or delivered from independent atmospheric physics columns. The new architecture also provides flexibility for forcings from other coupled components (e.g. river model, or eventually a coastal ocean) to be applied to the land sub-grid at other levels, as appropriate. The second stage of this development, underway now and scheduled for completion in CY 2018, divides the state and flux variables for water, energy, carbon, and nutrients into subsets that are native either to the soil column or to the vegetation or plant functional type (PFT) sub-grid level. This structural change provides more clarity in the code about which variables are maintained at which sub-grid level, and where to find variables within the fundamental data types. It also lays the foundation for new science by implementing an arbitrary number of soil columns on each vegetated landunit, and providing a data mechanism for tracking both the land area associated with each soil column and an area-associated distribution of time-since-disturbance within each of those columns. Examples of the new data structures are shown, illustrating the standardized topologies, atmospheric coupling, potential for component coupling at other sub-grid levels, and implementation for the purpose of improved simulation of disturbance effects.