Paramétrage de l'interception de la neige par le couvert forestier

[Departement_IRSTEA]Territoires [TR1_IRSTEA]SEDYVINInternational audienceSnow interception drives spatial heterogeneity of snow under forest canopies and displays significant differencesbetween forested, open and alpine areas at a variety of scales. Beyond giving a first order control on snow accu-mulation, interception by canopy drives other processes. A prime example is canopy albedo, as large differencesare readily visible between forest canopy albedo with and without intercepted snow. As such a correct parameteri-zation of interception is necessary as it drives many physical processes important for snow hydrology, climatologyand meteorology studies. However, current parameterizations of snow interception have not always beenable to preserve the large variance of snow beneath canopies at all scales. Various snow interception parameter-izations are applied in land surface models but are generally not validated for different snow climates and/or scales.Here, we developed parameterizations for spatial mean and standard deviation of interception over horizon-tal scales of 50 m. They were developed from a comparison of (1) computed forest structure metrics (sky viewfactor and standard deviation) from a high-resolution Lidar derived digital terrain model and (2) an existingdataset of several thousand interception measurements collected after nine storm events in a coniferous forest inthe Eastern Swiss Alps. By scaling open area snow precipitation with the calculated forest structure metrics, wecomputed spatial mean and standard deviation of forest canopy interception. We obtained similar performancestatistics compared to previously suggested parameterizations, i.e. a RMSE of 1.3 cm (1 mm SWE) for spatialmean and 0.6 cm (0.4 mm SWE) for the standard deviation of canopy interception.Furthermore, we then validated both new interception parameterizations with data from two different geo-graphic regions and snow climates, namely from a study site in Utah, U.S. and one in the French Alps. Thiscomparison suggests that our sub-grid parameterizations for snow interception are applicable in models to describesnow depth heterogeneities for different snow climates and mountain forest environment