Monday, 11 January 2016: 11:00 AM
Room 242 ( New Orleans Ernest N. Morial Convention Center)
Many studies make the claim of undertaking model benchmarking. Unfortunately, there is often confusion about what “benchmarking” means; some undertake true benchmarking, others are undertaking the more traditional evaluation or comparison activities. In this presentation we will attempt to clarify the differences between the three approaches and demonstrate how the interpretation of model results can differ depending on which of the three measures of model performance are used. To enable this, data from the land surface benchmarking experiment PLUMBER (PALS Land sUrface Model Benchmarking Evaluation pRoject) is used.
In addition, a brief overview of the PLUMBER experimental protocol will be presented along with the key findings from the experiment to date. Benchmarking methods are utilised to identify how conclusions can be biased by the statistical methods used for analysis, to highlight systematic issues across many widely used land surface models and to challenge our traditional view of the surface energy balance.
Finally recommendations are given for the development priorities of land surface models, along with suggestions for future community land surface benchmarking projects.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner