The PALS Land sUrface Model Benchmarking Evaluation pRoject (PLUMBER)
by Gab Abramowitz (UNSW)
The Met Office (Martin Best) is coordinating a land surface model (LSM) benchmarking experiment through the PALS system – the PALS Land sUrface Model Benchmarking Evaluation pRoject (PLUMBER). While the project is in essence very similar to the PILPS type experiments the community has been performing for many years now, there is one key difference. PLUMBER aims to reinforce benchmarking as distinct from evaluation, where benchmarking aims to define expectations of model performance a priori. Benchmarks in this sense include older conceptual LSM implementations such as Manabe bucket and Penman-Monteith models, as well as three out-of-sample empirical models, as described in Abramowitz (2012, GMD). LSMs are running multi-year simulations at 20 Fluxnet sites covering 10 IGBP vegetation types. Performance of latent heat flux, sensible heat flux and net ecosystem exchange of CO2 is examined with initial metrics covering mean offset, variance, correlation and normalised mean error. Participating LSMs include CABLE, CABLE_SLI, JULES, NOAH, ORCHIDEE, CHTESSEL, COLASSiB, Mosaic, CLM, including several variants of some LSMs. Early results appear to suggest significant under-performance in sensible heat flux relative to benchmarks. All model outputs and a broad range of analysis are available the PALS in the GLASS_benchmark workspace.
