Explore the vast range of titles available.

Model Intercomparison Projects (MIPs) are fundamental to our understanding of how the land surface responds to changes in climate. However, MIPs are challenging to conduct, requiring the organization of multiple, decentralized modeling teams throughout the world running common protocols. We explored centralizing these models on a single supercomputing system. We ran nine offline terrestrial biosphere models through the Terrestrial Biosphere Model Farm: CABLE, CENTURY, HyLand, ISAM, JULES, LPJ‐GUESS, ORCHIDEE, SiB‐3, and SiB‐CASA. All models were wrapped in a software framework driven with common forcing data, spin‐up, and run protocols specified by the Multi‐scale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP) for years 1901–2100. We ran more than a dozen model experiments. We identify three major benefits and three major challenges. The benefits include: (a) processing multiple models through a MIP is relatively straightforward, (b) MIP protocols are run consistently across models, which may reduce some model output variability, and (c) unique multimodel experiments can provide novel output for analysis. The challenges are: (a) technological demand is large, particularly for data and output storage and transfer; (b) model versions lag those from the core model development teams; and (c) there is still a need for intellectual input from the core model development teams for insight into model results. A merger with the open‐source, cloud‐based Predictive Ecosystem Analyzer (PEcAn) ecoinformatics system may be a path forward to overcoming these challenges. Plain Language Summary Comparing models is fundamental to our understanding of how the land surface responds to changes in climate. However, these comparisons are challenging to conduct, requiring the organization of multiple, decentralized teams throughout the world. We explored centralizing these models on a single supercomputing system. The models were all run the same way. We ran more than a dozen model experiments. We identify three major benefits and three major challenges. The benefits include: (a) the centralized system takes a lot of burden off individual teams; (b) running models the same way helps to identify differences in how the world is represented in the models; and (c) the system allows us to run many model experiments relatively quickly. The challenges are: (a) lots of models require lots of data storage and transfer needs; (b) model versions lag those from the core model development teams; and (c) there is still a need for intellectual input from the core model development teams for insight into model results. Another system, called PEcAn, which has a lot of tools that can help overcome these challenges, can potentially be used in future work. Key Points We ran nine terrestrial biosphere models centralized on a common computing framework The Farm allows multiple MIP experiments to be run relatively quickly and uniformly Challenges included technological demand, model versioning, and interpretation of results

Methane (CH4) flux estimates from high-latitude North American wetlands remain highly uncertain in magnitude, seasonality, and spatial distribution. In this study, we evaluate a decade (2007–2017) of CH4 flux estimates by comparing 16 process-based models with atmospheric CH4 observations collected from in situ towers. We compare the Global Carbon Project (GCP) process-based models with a model inter-comparison from a decade earlier called The Wetland and Wetland CH4 Intercomparison of Models Project (WETCHIMP). Our analysis reveals that the GCP models have a much smaller inter-model uncertainty and have an average magnitude that is a factor of 1.5 smaller across Canada and Alaska. However, current GCP models likely overestimate wetland fluxes by a factor of two or more across Canada and Alaska based on tower-based atmospheric CH4 observations. The differences in flux magnitudes among GCP models are more likely driven by uncertainties in the amount of soil carbon or spatial extent of inundation than in temperature relationships, such as Q10 factors. The GCP models do not agree on the timing and amplitude of the seasonal cycle, and we find that models with a seasonal peak in July and August show the best agreement with atmospheric observations. Models that exhibit the best fit to atmospheric observation also have a similar spatial distribution; these models concentrate fluxes near Canada's Hudson Bay Lowlands. Current, state-of-the-art process-based models are much more consistent with atmospheric observations than models from a decade ago, but our analysis shows that there are still numerous opportunities for improvement.

Land surface models (LSMs) are essential tools for simulating the coupled climate system, representing the dynamics of water, energy, and carbon fluxes on land and their interaction with the atmosphere. However, parameterizing sub‐grid processes at the scales relevant to climate models (∼${\\sim} $ 10–100 km) remains a considerable challenge. The parameterizations typically have a large number of unknown and often correlated parameters, making calibration and uncertainty quantification difficult. Moreover, many existing LSMs are not readily adaptable to the incorporation of modern machine learning (ML) parameterizations trained with in situ and satellite data. This article presents the first version of ClimaLand, a new LSM designed for overcoming these limitations, including a description of the core equations underlying the model, the results of an extensive set of validation exercises, and an assessment of the computational performance of the model. We show that ClimaLand can leverage graphics processing units for computational efficiency, and that its modular architecture and high‐level programming language, Julia, allows for integration with ML libraries. In the future, this will enable efficient simulation, calibration, and uncertainty quantification with ClimaLand. Plain Language Summary Simulating the Earth's atmosphere, ocean, and land surface is an important method that scientists use for understanding the Earth's climate, including its response to climate change. Due the complexity of the processes involved, approximations are made when representing certain aspects of the land surface, such as vegetation heterogeneity or topographical variation. These approximations can be improved by using data (“calibration”), but doing so has a large computational cost. They can also be improved using machine learning (ML), but this requires models to be easily integrated with ML packages. ClimaLand is a new land surface model which has been designed from the start to incorporate ML parameterizations and to more efficiently calibrate parameterizations with data. This article presents the ClimaLand model, benchmarks its computational performance, and compares model output against data in a variety of regimes. Follow‐on studies will improve the core model using ML parameterizations and by calibrating the model. Key Points ClimaLand, a land surface model, simulates energy, water, and carbon fluxes within and across soil, canopy and snow components The soil model simulates flow and phase changes of water in both saturated and unsaturated zones The model runs natively on graphics processing units and is designed to enable the inclusion of data‐driven parameterizations