The aim of the study was to investigate the relative
performance of the Snowmelt Runoff Model (SRM) to simulate streamflow in five
sub-catchments of the High Atlas Mountain range when using two snow extent
products of limited precision: i) a snow cover information derived from the
VEGETATION sensor onboard the SPOT satellite with a mean weekly interval, ii)
snow cover information computed solely from the meteorological data acquired at
a few climate stations.
At the seasonal scale, snow cover information obtained from
SPOT-VEGETATION and generated with the degree day method is quite comparable
for the five tributary sub-catchments. In general, streamflow simulation is
good in the Rheraya and Ourika sub-catchments where snow processes are
important and hydrometeorological data are relatively good. In the other hand,
SRM performances were poorer in the Nfis, Zat and R'Dat sub-catchments where
snow plays a smaller role in the hydrological budget.
In this study, the snowmelt contribution to streamflow was
computed in all sub-catchments from 2002 to 2005 using snow maps derived from
SPOT-VEGETATION sensor. Generally, it was shown that 25 % of streamflow
arriving from the North sides of High Atlas is derived from snowmelt.
At annual timescales, the simulated and observed hydrograph
using the two snow products in all sub-catchments are similar. Due to local
intense rainfall events not measured by the weather stations, where streamflow
tends to be dominated by rapid responses, the multiple peak discharge simulated
was often lower than observed. During recessions, the streamflow simulations
are acceptable. However, using snow cover information derived from remote
sensed data can significantly improve streamflow prediction for individual
interstorm periods were rainfall events are not observed by the given network
or when the temperature lapse rate is badly estimated.
Finally, the Remote Sensing and the meteorological data were
used separately to compute snow cover extent as an input in the SRM model.
Since the results with either data sources are encouraging, combining both
products to estimate the snowpack evolution between two image acquisitions
(instead of linearly interpolated snow depletion curves, as it is classically
done in most SRM applications) should improve the streamflow prediction
performance in the High Atlas. This has not been tested in this study but will
be done in the next future.
ACKNOWLEDGEMENTS
This study was supported by the research projects SUDMED
(IRD-UCAM), PAI (`Programme d'Action Intégrée du Comite Mixte
Interuniversitaire Franco-Marocain, Jeune Equipe IRD (CREMAS), `Volubilis'
MA/06/148) and PLEADeS see
http://www.pleiades.es/ project
funded by the European Commission (6th PCRD). The authors are
grateful to ABHT (Agence du Bassin hydraulique de Tensift, Marrakech, Morocco)
for the acquisition of the hydro-meteorological data on the Tensift watersheds.
We also thank the SPOT-VEGETATION program for provided the series of satellites
images.
