LMWG Near-term Plans
27 March 2013
Immediate plans for the LMWG are to finalize and document CLM4.5, which will be released in May 2013. CLM4.5 will alleviate several significant biases in CLM4 including excessive tropical GPP, LAI and LH, low soil carbon stocks especially in high latitudes, unrealistic cold region hydrology, and problems with nitrogen dynamics. Preliminary simulations indicate increased carbon uptake over 20th century more in line with observed estimates. CLM4.5 will provide a basis for further model development on the path towards CLM5 and we anticipate that it will be a useful and more realistic model for (primarily offline) CLM users for carbon cycle, water cycle, and ecosystem studies over the next several years. The modifications slated for inclusion in CLM4.5 are listed below. Details can be found in the associated papers and will be documented in a paper describing CLM4.5:
- Maintain backwards compatibility with option to switch between CLM4 and CLM4.5
- Revised photosynthesis model, multilayer canopy (optional), temperature acclimation, iterative calculation fix (Bonan et al., 2011, 2012; Sun et al., 2012)
- Cold region hydrology improvements including ice impedance, perched water table, corrections to soil water equations where ice is present (Swenson et al. 2012)
- Snow cover fraction update with improved accumulation and ablation characteristics (Swenson and Lawrence, 2012)
- 2-way CLM/RTM interactions, flooding (default off) (Swenson and Lawrence, in prep)
- CENTURY-like vertically resolved soil biogeochemistry, including some revisions to N-dynamics (Koven et al., in prep)
- New lake model and lake datasets (Subin et al., 2012)
- Methane emissions from wetlands (Riley et al., 2011; Meng et al. 2012)
- Prognostic wetland distribution model, surface water store (Swenson and Lawrence, in prep)
- Revised fire model including anthropogenic triggers and suppression and agricultural and peat fire (Li et al., 2012; 2013)
- Fertilization, irrigation, organs pool, and other updates to crop model (Drewniak et al., 2013; Levis et al., 2012; Sacks et al. 2009)
- VIC hydrology (optional alternative hydrology scheme) (Li et al., 2012)
- C13/C14 enabled
- Multiple urban classes
- Several additional minor and major bug fixes, speedup of BGC spinup
The intention is for CLM4.5 to be used primarily for uncoupled studies and no scientifically supported coupled compsets will be included with the release. Time permitting, the LMWG intends to evaluate CLM4.5 coupled to CAM5 and add a scientifically supported CAM5/CLM4.5 compset sometime this year. Scientifically supported compsets, including simulations, will be conducted and provided for a transient CLM4.5 historical period (probably forced with the CRUNCEP historical forcing dataset).
A main priority over the next several months will be the documentation of CLM4.5. Contributors to CLM4.5 are working on a revision to the CLM Tech Note that will go out with the model release. In addition, a journal paper documenting the changes, improvements, and remaining high priority weaknesses/limitations in CLM4.5 is planned. Discussions across the working group are pushing for this paper to also serve as a land model metrics paper in which a set of metrics are defined that robustly test CLM water, energy, and carbon cycles. This effort will be done in collaboration with the International Land Model Benchmarking project (ILAMB). We recognize that, in a complex model like CLM, there are many ways to improve some aspects of a simulation (e.g., NEE) while degrading other aspects (e.g., hydrology). The current group of active contributors to CLM is very large, and without a consistent formalism for model diagnostics, there is a real risk of degrading important aspects of the simulation, or worse, not realizing at the time that such degradations occurred. To improve the model development process, a goal of the LMWG is to 1) develop an appropriate suite of metrics for model parameterizations (e.g., stomatal conductance response to environmental factors) and "emergent responses" (e.g., NEE trends) and 2) development of a capability to do unit testing on model components. Where possible, criteria will be established on the basis of this testing that must be met before a new model parameterization can be integrated into a released version of CLM. Human resources are a major limiting factor and additional funding is likely required to bring this idea to complete fruition.
Once CLM4.5 is finalized, model development will shift towards bringing in several large model developments that can be brought in as non answer-changing options. A high priority is to bring the Ecosystem Demography (ED) model onto the trunk so that developers can start investigating how the ED model interacts with other parts of CLM. Infrastructure for riverine transport of nutrients and model changes to enable dynamic landunits (i.e., allowing transient changes in landunit weights like glacier to vegetated or vegetated to crop; note that this work will permit us to run the crop model and/or the dynamic global vegetation model in transient land cover change experiments) are also high on the priority list. The dynamic landunit work will also help pave the way for software changes required to embed an Integrated Assessment Model (initially, GCAM) within CESM. There are many other active development projects across the LMWG including a revisions to the nitrogen cycle, a new river flow model, phosphorous dynamics, a reactive transport model, soil microbial dynamics, water isotopes, photosynthesis, permafrost excess soil ice, ozone poisoning of vegetation, data assimilation, sub-surface lateral redistribution of water, and global crops (not an all inclusive list). Over the course of the next year, LMWG members will continue to work and evaluate these model developments and the most promising and mature projects, especially those that advance our ability to address priority CESM science questions, will be considered and tested for possible inclusion in the next version of CLM.
Finally, the CLM software engineers have outlined a set of code refactoring that will increase the robustness and extensibility of the model and will hopefully reduce the amount of time that it takes to bring in new science developments. Among the proposed refactoring projects is an effort to pull hard coded parameters out of CLM into external input files, which will enable parameter uncertainty and parameter estimation work, a priority of CESM and CLM.