| Session Details | |
| Section | SS - Special Sessions |
| Session Title | International Land Model Benchmarking (ilamb) Package Tutorial |
| Main Convener | Dr. Forrest Hoffman (Oak Ridge National Laboratory, United States) |
| Co-convener(s) | Dr. Nathan Collier (Oak Ridge National Laboratory, United States) |
| Session Description | To advance understanding of biogeochemical processes and their interactions with hydrology and the Earth system under changing environmental conditions, new methods are being developed that use observations to constrain model predictions, inform model development, and identify needed measurements and field experiments. The International Land Model Benchmarking (ILAMB) package embodies those methods to provide quick and easy model–data comparison capabilities for multiple Earth system models (ESMs). The ILAMBv2.2 Package assesses model fidelity on 25 variables in four categories from about 50 data sets; produces graphical global-, regional- and site-level diagnostics; and provides a hierarchical scoring system based on model performance for the annual mean, bias, relative bias, root-mean-square error (RMSE), seasonal cycle phase, spatial distribution, interannual variability, and variable-to-variable comparisons (functional benchmarks). ILAMB is designed for use by individual modelers or model developers for verification and rapid model development cycles, by modeling centers to track the evolution of model performance, and by model intercomparison projects for multi-model analysis. ILAMB is being integrated into the workflow systems at multiple major modeling centers and are being extended for use in high-latitude and hydrology studies. The package is also expected to provide a platform for engaging experimentalists in identifying model weaknesses and needed measurements and field experiments. This hand-on tutorial session will provide students, postdocs, and researchers with experience in installing, running, and extending the ILAMB pckage for assessing model performance in comparison with observations. Participants will be guided through installing the package, written in Python, and observational data on their Linux or MacOS system. Instruction will be provided for evaluating sample model output (provided) or their own model results. In-depth explanation of the software structures and classes will be provided, enabling users to develop their own custom evaluation methods or to use the framework in their own diagnostic packages. Participants should bring their own laptops for either installing the software on their Python-enabled systems (with netcdf4, numpy, and basemap modules) or for accessing a server, where a login account will be provided for the tutorial. Knowledge of Python, netCDF files, and the Climate & Forecast convention is useful, but not required. |