Abstract. Detailed studies of snow cover processes require models that offer a fine description of the snow cover properties. The detailed snowpack model Crocus is such a scheme, and has been run operationally for avalanche forecasting over the French mountains for 20 yr. It is also used for climate or hydrological studies. To extend its potential applications, Crocus has been recently integrated within the framework of the externalized surface module SURFEX. SURFEX computes the exchanges of energy and mass between different types of surface and the atmosphere. It includes in particular the land surface scheme ISBA (Interactions between Soil, Biosphere, and Atmosphere). It allows Crocus to be run either in stand-alone mode, using a time series of forcing meteorological data or in fully coupled mode (explicit or fully implicit numerics) with atmospheric models ranging from meso-scale models to general circulation models. This approach also ensures a full coupling between the snow cover and the soil beneath. Several applications of this new simulation platform are presented. They range from a 1-D standalone simulation (Col de Porte, France) to fully-distributed simulations in complex terrain over a whole mountain range (Massif des Grandes Rousses, France), or in coupled mode such as a surface energy balance and boundary layer simulation over the East Antarctic Ice Sheet (Dome C).

Abstract. SURFEX is a new externalized land and ocean surface platform that describes the surface fluxes and the evolution of four types of surfaces: nature, town, inland water and ocean. It is mostly based on pre-existing, well-validated scientific models that are continuously improved. The motivation for the building of SURFEX is to use strictly identical scientific models in a high range of applications in order to mutualise the research and development efforts. SURFEX can be run in offline mode (0-D or 2-D runs) or in coupled mode (from mesoscale models to numerical weather prediction and climate models). An assimilation mode is included for numerical weather prediction and monitoring. In addition to momentum, heat and water fluxes, SURFEX is able to simulate fluxes of carbon dioxide, chemical species, continental aerosols, sea salt and snow particles. The main principles of the organisation of the surface are described first. Then, a survey is made of the scientific module (including the coupling strategy). Finally, the main applications of the code are summarised. The validation work undertaken shows that replacing the pre-existing surface models by SURFEX in these applications is usually associated with improved skill, as the numerous scientific developments contained in this community code are used to good advantage.

[1] Thirty-three snowpack models of varying complexity and purpose were evaluated across a wide range of hydrometeorological and forest canopy conditions at five Northern Hemisphere locations, for up to two winter snow seasons. Modeled estimates of snow water equivalent (SWE) or depth were compared to observations at forest and open sites at each location. Precipitation phase and duration of above-freezing air temperatures are shown to be major influences on divergence and convergence of modeled estimates of the subcanopy snowpack. When models are considered collectively at all locations, comparisons with observations show that it is harder to model SWE at forested sites than open sites. There is no universal ''best'' model for all sites or locations, but comparison of the consistency of individual model performances relative to one another at different sites (and vice versa). Calibration of models at forest sites provides lower errors than uncalibrated models at three out of four locations. However, benefits of calibration do not translate to subsequent years, and benefits gained by models calibrated for forest snow processes are not translated to open conditions.

The Interactions between Soil, Biosphere, and Atmosphere land surface scheme is currently used coupled both to atmospheric models and to a distributed hydrological model. There are two snow-scheme options available for hydrological modeling: the baseline force-restore approach, which uses a composite snow-soil-vegetation energy budget and a single snow layer; and a multilayer detailed internal-process snow model. Only the forcerestore method is routinely used in atmospheric modeling applications. Recent studies have shown that hydrological simulations for mountainous catchments within the Rhone basin, France, are significantly improved using the detailed snow scheme. The main drawback is that the scheme is computationally expensive, and it is not currently feasible for routine application in atmospheric models. For these reasons, a third new intermediatecomplexity scheme has been developed that includes certain key physical processes from the complex model for improved snowpack realism and hydrological depiction while attemping to keep computational requirements similar to those of the simple default scheme. In the current study, the new scheme is described, evaluated, and compared with the results from the two other schemes at a local scale at an alpine site located within the Rhone basin for two contrasting (weather) years. All schemes are able to model the basic features of the snow cover with similar errors averaged over the 2-yr period; however, there are important differences on shorter timescales. When compared with the more complex scheme, it was found that differing surface energy budget parameterizations (turbulent transfer, albedo) were the cause for the largest differences in total snowpack snow water equivalent (SWE) simulated by the models. When compared with the simple scheme, the ability for the intermediate model to simulate snow ripening resulted in the largest differences in simulated SWE and snow temperature during melt and runoff.

The Protocol for the Analysis of Land Surface Models (PALS) Land Surface Model Benchmarking Evaluation Project (PLUMBER) was designed to be a land surface model (LSM) benchmarking intercomparison. Unlike the traditional methods of LSM evaluation or comparison, benchmarking uses a fundamentally different approach in that it sets expectations of performance in a range of metrics a priori-before model simulations are performed. This can lead to very different conclusions about LSM performance. For this study, both simple physically based models and empirical relationships were used as the benchmarks. Simulations were performed with 13 LSMs using atmospheric forcing for 20 sites, and then model performance relative to these benchmarks was examined. Results show that even for commonly used statistical metrics, the LSMs' performance varies considerably when compared to the different benchmarks. All models outperform the simple physically based benchmarks, but for sensible heat flux the LSMs are themselves outperformed by an out-of-sample linear regression against downward shortwave radiation. While moisture information is clearly central to latent heat flux prediction, the LSMs are still outperformed by a three-variable nonlinear regression that uses instantaneous atmospheric humidity and temperature in addition to downward shortwave radiation. These results highlight the limitations of the prevailing paradigm of LSM evaluation that simply compares an LSM to observations and to other LSMs without a mechanism to objectively quantify the expectations of performance. The authors conclude that their results challenge the conceptual view of energy partitioning at the land surface.