Meeting report: WG Model Evaluation and Benchmarking, February 2013

On Friday 15th February working group 2 (Model Evaluation and Benchmarking) had their first meeting by phone. A summary of the discussion is below.

If you are interested in joining this group please contact Gab Abramowitz

Attending: Gab Abramowitz, Jason Evans, Andrew Frost, Rachel Law, Adam Smith, Matthew Stenson, Albert Van Dijk, Neil Viney, Jeff Walker

Apologies: Beth Ebert, Valentijn Pauwels

Absent: Barry Croke, Rachel Gilmore, Juan Guerschman, Garth Warren, Qiang Yu

At its inaugural meeting, WG2 reinforced its stated aim to develop an automated, online, evaluation and benchmarking environment that is common to hydrological and land surface models. As a starting point, we will cater to CABLE and AWRA models as key representatives, recognising the need in both of their communities for an extended automated benchmarking system. We will use PALS as the base application system to build from, so that analyses scripts will likely be developed in R or Python and data formatted in CF- or ALMA compliant netcdf where appropriate. The first experiment in PALS to explore this dual role will likely be continental scale streamflow, using gauged unimpaired catchments, although the precise number and nature of catchments will depend on the availability of appropriate driving data and resolution that will suit both model types. Ultimately, we hope for integration of existing groundwater and flux tower data as additional constraints, as well as other relevant data streams considered by WG1 (Observational data). We also noted the interest from WG3 (Model-Data Assimilation) in an assimilation comparison exercise, particularly the possible role of WG2 in contributing to the evaluation schema for such a project.

Action points from this meeting:

– Gab Abramowitz will liaise with Matthew Stenson regarding the AWRA model input and output formatting, as well as Neil Viney regarding the formatting and availability of runoff data.

– Gab Abramowitz will update the group at the next meeting on the progress of integration of recent OzFlux level 4 data into the PALS system

– we resolved to attempt future meetings with h.323 video conferencing as most parties seem to have access to appropriate facilities already

Leave a Reply

Your email address will not be published. Required fields are marked *