Study WP5.6 Snow Dynamics Impacts on Temperate / High Latitude Climate
Description
This Study is led by Catherine Ottlé, Amélie Cuynet and Philippe Peylin from LSCE-IPSL.
The main CCI ECVs used in this study are Snow Cover Fraction SCF (Viewable and on Ground), Snow Water Equivalent SWE and MR-Land Cover.
The study's main objective is to improve our understanding and modelling of snow-vegetation-atmosphere feedback, with the IPSL climate model (LMDZ-ORCHIDEE) and various CCI products (especially snow products). The study comprises several parts. The first part involves data analysis including consistency check/analysis between snow cover (mass and extent), land cover and other CCI products + albedo with analysis of the differences between short and tall vegetation and deciduous and evergreen trees. The second part involves ORCHIDEE model evaluation with the evaluation of the simulated snow cover dynamics (mass and extent) and snow albedo using simulations with prescribed climate forcing (e.g., ERA5) and define a set of key “homogeneous points” for the identification of model biases. The third part involves model improvement with improved soil thermics (carbon impact on soil thermal properties; ongoing work) and SCF parameterisations and accounts for shrubs and the representation of snow - vegetation dynamics. The fourth part involves snow model parameter optimisation with model sensitivity analysis experiments to identify critical parameters (Morris/Sobol approaches) and multi-site optimisation (local/global approaches) using albedo, SCF and SWE data. The fifth, part (accepted option) is to explore Coupled Model simulations with the use of the Coupled LMDZ -ORCHIDEE model (AMIP type simulation: fixed SST, SIC) and exploring historical simulations to analyse the impact of the “improved snow model” on surface-atmosphere feedbacks.
Results and conclusions (June 2025)
This study led to multiple results, that can be presented in three parts:
- Analysis of the observational datasets (CCI Snow, albedo and land cover products) with a focus on snow-related variables: snow cover fraction (SCF) and snow water equivalent (SWE)
- Preliminary comparison of snow evolution observations with simulations from two versions of the ORCHIDEE land surface model (LSM): V3.0 and Trunk standard version (V4.0)
- Development and optimisation of parameters in the ORCHIDEE LSM for albedo and SCF
Analysis of the observation products: impact of the vegetation on snow behaviour
Several updated versions of the CCI products were released during the study, with noticeable improvements over earlier versions. An example of the MODIS-based viewable SCF product (v3.0) is shown in Figure 1. Analysis of snow and land cover products revealed a clear influence of vegetation type on snow behaviour. For example, snow coverage varied by vegetation type, indicating that SCF and albedo parameters should be adapted accordingly for each vegetation category, called hereafter plant functional types (PFT).
Comparison of CCI Snow products with ORCHIDEE simulation outputs
Two global reference simulations — without optimised snow parameters — were carried out using ORCHIDEE v3 and v4, and compared with CCI Snow and MODIS albedo products across regions above 30°N. Selected results from the V4 version are presented here.
Figure 2 shows the mean albedo Root Mean Square Error (RMSE) over the period 2011-2019 between the monthly averaged albedo from ORCHIDEE and the MODIS albedo product. Large RMSE values, up to 0.20, could be observed in the northernmost regions. Figure 3 illustrates monthly SCF differences between v4 and CCI SCF, showing SCF overestimation in areas with year-round snow and underestimation in May due to too early melting in the simulation.
The two model versions exhibited distinct behaviours in albedo modelling. In v3, albedo errors tended to be more concentrated during winter. In contrast, Trunk followed a dipole pattern: it strongly underestimated albedo in snowy conditions, particularly in high-latitude regions, while overestimating it in non-snow-covered areas and in southern regions, suggesting a strong constraint linked to vegetation albedo. The differences between v3 and v4 indicate that the new albedo scheme present in v4 and not in v3, impacts the overall albedo computation at all seasons, and that both the vegetation and snow albedo dynamics should be analysed and optimised.
![Figure 3: SCFV monthly differences [%] between ORCHIDEE v4 (Trunk) and Snow CCI - 2011-2019. The regions in grey correspond to pixels where data were unavailable](/media/images/Picture3_w2RfI7w.max-2000x2000.png)
Optimising albedo parameters in both versions is necessary to address these disparities, ensuring improved consistency in albedo, SCF, and SWE modelling across different regions and seasons.
Albedo and snow cover fraction optimisation
A three-step optimisation approach was implemented for each plant functional type (PFT):
- Tuning global vegetation albedo parameters
- Tuning snow albedo parameters
- Tuning snow cover fraction parameters
A dedicated protocol was developed to identify optimal regions for site-level calibration. An algorithm was used to select homogeneous pixels in terms of vegetation, SCF, SWE, and albedo. Vegetation evolution was assessed using the new PFT maps by Harper et al. (2023).
In ORCHIDEE v3, vegetation albedo parameters were recalibrated using these maps and albedo datasets. Snow albedo parameters—PFT-dependent—were also adjusted based on CCI Snow, MODIS albedo, and the new PFT maps. This method significantly improved global albedo estimates. Figure 4 illustrates time series of mean albedo and SCF (observations in black, pre-optimisation in blue, post-optimisation in orange) for a boreal broadleaf summergreen forest site in Siberia.
The same optimisation procedure now needs to be applied to the current v4 version.