Snowmelt Contributions to Runoff in an Extremely Wide Altitude Range from Large Area Satellite Imagery

Based on periodical snow cover mapping by NOAA satellites, contributions to runoff from seasonal snow cover, intermittent snowfalls and rain are evaluated in relation to the altitude in the basins of the rivers Ganges (917 444 km) and Brahmaputra (547 346 km) in an altitude range from 0 to 8848 m a.s.l. In spite of insufficient precipitation and temperature data, runoff was simulated with an acceptable accuracy by the SRM model, as an indirect confirmation of the respective input quantities. Snowmelt including new snow contributes by 9% to runoff in the Ganges basin and by 27% in the Brahmaputra basin, in line with the shape of area-elevation curves. With the climate scenario assumed for this region, an increase of runoff and flood risks by the year 2030 was modelled. Based on the NOAA/AVHRR imagery, a relation between the percentage of snow in precipitation in relation to altitude was derived. The 50% point is by 1500 m higher than that of a similar relation derived from the Swiss snow gauging stations in a limited elevation range. This is in line with the more southern geographical latitude of the Himalayan basins. INTRODUCTION Large scale snow cover mapping in the Himalayan basins of Ganges (917 444 km) and Brahmaputra (547 346 km) was conducted in the framework of the ESA-DUP SPIHRAL project. This application of remote sensing in cold regions provided data for studying the varying contribution of snowmelt to runoff in an altitude range from 0 to 8848 m a.s.l. The year round NOAA/AVHRR monitoring of the seasonal snow cover was evaluated separately for 7 elevation zones. From the varying snow cover areas, precipitation and temperature data, the respective contributions to runoff from snowmelt and rain were computed by a hydrological model. Seidel and Martinec Snowmelt Contributions to Runoff in an Extremely Wide Altitude Range from Large Area Satellite Imagery 1 LARGE AREA SATELLITE SNOW COVER MAPPING FOR RUNOFF MODELLING The situation of the studied basins is shown in Fig. 1. The periodical snow cover mapping was based on 10-days composites of NOAA/AVHRR data recorded in 1995. In each month, the best cloud-free composite was selected so that the snow coverage was evaluated in terms of monthly values for the respective elevation zones by means of a digital elevation model (DEM). Precipitation data were available only as monthly totals and so it was not possible to identify daily snowfalls. In the normal use of depletion curves of the snow coverage for snowmelt runoff computations (Hall and Martinec, 1985) a short-lived snow cover seen by satellites after occasional summer snowfalls is disregarded so that these curves refer to the seasonal snow cover. In the present case, the average snow covered area in each month had to be used. Another deficiency were temperature data which were also available only as monthly means. In order to use the SRM model (Martinec et al., 1998) for runoff modelling, the monthly precipitation totals were empirically desaggregated (Baumgartner, 1999) by taking into account the statistically determined numbers of daily precipitation events in each month. In spite of these inadequacies, it was possible to simulate runoff in both basins as shown in Fig. 2 and Fig. 3. The combined annual measured runoff in both rivers is 1 037 932 10m and the computed runoff 1 009 732 10m. This represents 2.3% of the mean annual streamflow volume of the world (Shiklomarow, 1990). By contrast, the area of Bangladesh is only 0.1% of the world land area without Antarctica. By relation to the respective land areas, the combined runoff of Ganges and Brahmaputra in 1995 was for example 23 times larger than the mean annual runoff of all French rivers and 7 times larger than the mean annual runoff of all Swiss rivers. Seidel and Martinec Snowmelt Contributions to Runoff in an Extremely Wide Altitude Range from Large Area Satellite Imagery 2 Figure 1. Situation of the basins of the rivers Ganges (G) and Brahmaputra (B) asseen by NOAA/AVHRR (10-days composite from September 1995) Figure 2. Measured and computed runoff in the river Ganges, 1995 Seidel and Martinec Snowmelt Contributions to Runoff in an Extremely Wide Altitude Range from Large Area Satellite Imagery 3 Figure 3. Measured and computed runoff in the river Brahmaputra, 1995 The SRM model, originally developed in a small basin of 2.65 km, was for the first time applied in basins of this order of magnitude (Seidel et al., 2000). It was possible thanks to remote sensing data as a direct input and thanks to a special treatment of the recession flow incorporated in the model. The recession formula reads Kn + 1 = x • Qn (1) Where k = variable recession co Q = daily runoff [ms] X, y = constants to be determ N = sequence of days Taking into account the size of the b explained elsewhere (Martinec, et al., Seidel and Martinec Snowmelt Contributions to Runoff i Qn + ____