The applicability of C-band SAR and optical data for snow monitoring in boreal forest

Optical remotely sensed data is used operationally to map snow extent in several countries. Also space-borne SARs have shown their usefulness in estimation of Snow Covered Area (SCA) under wet snow conditions, i.e. in the spring melt period. However, in boreal forest zone, forest cover deteriorates the accuracy of present empirical algorithms. These algorithms are typically based on pixel-wise or region-wise linear interpolation between reference images representing total (wet) snow cover and totally melt-off conditions. This paper presents two approaches to estimate regional SCA in forested areas 1) linear interpolation method for SAR data using backscattering model to compensate the influence of forest canopy and 2) inversion of empirical reflectance model for optical data (first implemented for NOAA/AVHRR). The SCA estimates are calculated for third class sub-drainage areas used by an operative hydrological model (The Watershed Simulation and Forecasting System, WSFS). The average size of sub-areas is 60 km(2). For SAR-data, the SCA estimation algorithm, as well as the validity of the employed model, is tested for the River Kemijoki drainage area in Northern Finland. For AVHRR-data, the estimates are performed and validated for all drainage basins of Finland. The SCA estimates from both methods are compared with ground observations. In addition, AVHRR-estimates are compared to predictions produced by the WSFS. For SAR-method, the preliminary results show that, in contrast to commonly used linear interpolation algorithms, the developed algorithm typically yields higher values of SCA for forested areas than for open areas, which is in agreement with natural snow melt process. Moreover, comparison of SAR-derived and AVHRR-derived SCA estimates shows the potential of synergetic use of both data.