Abstract
Water stored in the snowpack is an indispensable resource for sustaining human life, especially for mountainous regions like the Alps. Monitoring snowpack over remote areas is challenging and its heterogeneity limits the representativeness of in-situ measurements. A valid alternative is the use of physically based snow models, that provide high-resolution information on snow evolution for large areas. However, these models require highly accurate meteorological input data to produce reliable results and struggle to accurately represent snow redistribution caused by gravity and wind transport. Flexible snow model (FSM2) is a multi-physics energy balance snow model that can reconstruct the snowpack evolution using meteorological forcings as temperature and radiation as input (Essery et al., 2024). Another option is remote sensing, which provides free access to large-scale data. Current optical sensors provide accurate information about the snow extent. By integrating multi-source satellite data, such as high spatial resolution snow cover derived from Sentinel-2 and daily low-resolution snow cover derived from MODIS, we can derive daily high-resolution Binary Snow Cover time series (BSC-TS) (Premier et al., 2023). In this work, in order to investigate the impact of spatialization errors in meteorological input data on snow model simulations, we analyze how perturbations in temperature and radiation affect the timing of snow disappearance (i.e., time when SWE=0) in a complex terrain environment. Specifically, we aim to identify the minimum (co)perturbation required to minimize the discrepancy between snow disappearance dates simulated by the FSM2 model and those derived from remote sensing BSC-TS data. The perturbation analysis involved modifying temperature in a range from -1°C to +1°C and then altering the radiation (including both direct and diffuse components) with a multiplicative factor ranging from -20% to +20%. The area of interest is the Dischma catchment in Switzerland where high-resolution hydrometeorological and snow data are available as input to the snow model (Magnusson et al., 2024). The results are further analyzed in relation to altitude, slope, and aspect. The analysis reveals that the model tends to overestimate the time of snow disappearance with increasing altitude and slope. As expected, the sensitivity analysis confirms that an increase in both temperature and radiation decreases the bias in the timing for high altitudes and slopes while lower temperature and radiation is required for low altitudes and slopes. This result might be linked to inaccuracies of interpolation techniques due to missing in-situ data especially at high elevation, resulting in inaccurate trends of the forcings with elevation. Regarding slope, the overestimation trend can be attributed to gravitational forces which lead lateral snow redistribution, a process not implemented in FSM2. Also, wind redistribution might not be well represented by the method but better captured by the satellite product. For aspect, there is no clear trend but FSM2 tends to slightly overestimate the time of disappearance for North-exposed slopes, requiring an increase of temperature/energy to fit the benchmark contrarily to what is expected, while South-exposed areas tend to underestimate. On the other hand, it was expected to be an inverse trend of overestimation of temperature/energy for North-exposed slopes, given the fact that usual interpolation techniques do not count for casted shadow areas. This preliminary assessment highlights the potential for more complex data assimilation schemes while considering the limitations of input forcing and model. Integrating remote sensing data into snow models could enhance the accuracy of snow evolution simulations. The next step involves implementing improvements to simulate snow disappearance and the onset of melting, leveraging SAR observations from Sentinel-1. Further research could also improve an alternative method to abandon inputs such as meteorological forcing and implement satellite inputs. Acknowledgement: This work has been produced with co-funding from the European Union - Next Generation EU.
References:
Essery, R., Mazzotti, G., Barr, S., Jonas, T., Quaife, T., Rutter, N., 2024. A Flexible Snow Model (FSM 2.1.0) including a forest canopy. https://doi.org/10.5194/egusphere-2024-2546
Magnusson, J., Bühler, Y., Quéno, L., Cluzet, B., Mazzotti, G., Webster, C., Mott, R., Jonas, T., 2024. High-resolution hydrometeorological and snow data for the Dischma catchment in Switzerland. https://doi.org/10.5194/essd-2024-374
Premier, V., Marin, C., Bertoldi, G., Barella, R., Notarnicola, C., Bruzzone, L., 2023. Exploring the use of multi-source high-resolution satellite data for snow water equivalent reconstruction over mountainous catchments. The Cryosphere 17, 2387–2407. https://doi.org/10.5194/tc-17-2387-2023