A comprehensive set of benchmark tests for a land surface model of simultaneous fluxes of water and carbon at both the global and seasonal scale

Evaluating the models we use in prediction is important as it allows us to identify uncertainties in prediction as well as guiding the priorities for model development. This paper describes a set of benchmark tests that is designed to quantify the performance of the land surface model that is used in the UK Hadley Centre General Circulation Model (JULES: Joint UK Land Environment Simulator). The tests are designed to assess the ability of the model to reproduce the observed fluxes of water and carbon at the global and regional spatial scale, and on a seasonal basis. Five datasets are used to test the model: water and carbon dioxide fluxes from ten FLUXNET sites covering the major global biomes, atmospheric carbon dioxide concentrations at four representative stations from the global network, river flow from seven catchments, the seasonal mean NDVI over the seven catchments and the potential land cover of the globe (after the estimated anthropogenic changes have been removed). The model is run in various configurations and results are compared with the data. A few examples are chosen to demonstrate the importance of using combined use of observations of carbon and water fluxes in essential in order to understand the causes of model errors. The benchmarking approach is suitable for application to other global model