Snow Cover Mapping


The use of remote sensing data in arctic and alpine areas is especially useful because of the inaccessibility of these areas. Snow cover mapping is particularly important in relation to hydropower applications in areas where the catchment is covered in snow during part of the year. If the Snow Water Equivalent (SWE) of the snow pack can be estimated, the amount of runoff can be assessed. Local adaptations of algorithms are usually necessary in order to obtain the best results. GRAS is currently involved in several snow cover monitoring projects.

Methods
Remote sensing data from satellites can be used to assess snow water equivalents (SWE) using two different approaches. Either directly by using passive microwave data or by mapping changes in snow cover from optical satellite data.

Microwave data
SWE can be estimated directly from passive microwave remote sensing data, since the density and thickness of the snowpack is proportionally related to the temperature of the snow. Two brightness temperatures (apparent temperatures in different wavebands) are used and these are assessed from 37 and 18 GHz passive microwave signals. Data are available from the American satellites Scanning Multispectral Microwave Radiometer (SMMR) (from 1978 to 87) and Special Sensor Microwave Imager (SSM/I) (since 1987). The spatial resolution is approximately 25 x 25 km for these types of data and 10 day composites or daily data can be obtained. All optical remote sensing data can be used to assess snow cover; however, data with a high temporal coverage are preferred.

Optical data
Data from NOAA AVHRR and Modis are being used at the present. NOAA AHVRR has a spatial resolution of 1 x 1 km and Modis has 250-1000 m resolution. Both systems can provide daily and 10 day composite data. Time series of snow cover can be used to compute the number of days each grid cell is covered with snow. Applying a simple degree-day model along with a stratification of snow days in different altitude intervals, allows to assess SWE (must be calibrated on run-off data). NOAA AVHRR data are available since 1981 in 10 day composites aggregated to 8 x 8 km grid cells. Studies where snow cover mapping from NOAA AVHRR 8 km data have been compared to mapping using high resolution Landsat data (30 m grid cells) shows a good agreement. Modis data are only available since 2000.

Operational setup
SMM/I, NOAA AVHRR and Modis data can be obtained free of charge; however, processing must be done. Both of the suggested methods, the direct assessment of SWE using passive microwave data and the assessment of SWE through the mapping of changes in snow cover from optical data, can be applied in near real time (NRT) mode and are therefore suitable for forecasting purposes. NRT processing will be based on an application where remote sensing data are uploaded from a data server within a period of 3 to 12 hours upon reception. The data are then processed and made available for the hydrological modelling. The process can be fully automated once the SWE estimation method has been calibrated and the remote sensing data source has been identified. Usually it is necessary to analyse 25 to 30 years of inflow data to the upstream reservoir (and if available the inflow from sub-basins) and satellite images to calibrate the model.