CoastGIS’99: geomatics and coastal environment
Most of the time, only mean annual values are available and sampling
conditions and frequencies are not well known. The reliability of such
data is not easy to determine. On our regional scale, the program Classes
Ecofleuves was a TSM survey which provided several data series on
TSM fluxes and water discharge spanning the same period. Clearly, the
values statistical significance could be calculated, but it was also important to estimate the quality of the TSM flux estimations from metadata
and from the coefficient of TSM-Q regressions (Maneux, 1998) for each
calculated sediment yield value in order to determine which data sets
were significant for the period studied. Indeed, only 30 sediment yields
out of 50 monitored gauging stations were selected. The main difficulty
is to get reliable TSM fluxes from drainage basins witli surface areas
under 2000 square kilometres. Nevertheless, on smaller scales, such as
the Nivelle river watershed (165 km 2 ), the spatial correlation can be
performed with point TSM concentrations recorded along the stream
during flood events (De Maisonneuve et al., 1997). The reliability of
such TSM data is good but the spatial and temporal significance is not
well known.
In the same way as the Universal Soil Loss Equation ( Wishmeier et al.,
1958) and the global sediment flux modeling of Ludwig & Probst
(1996, 1998), morphological (slope, altitude), climatological (Fournier index), soil erodibility (texture, lithology) and Land Use and lancl
Cover databases (LUC) were gathered in a GIS in order to describe each
watershed (Band, 1989; Corine, 1992). On the regional scale, the spatial resolution seems sufficient to describe the watershed features. The
minimum spatial integration unit per watershed seems to be statistically
significant. Indeed, the smallest watershed ( 165 km
2 ) could be described
by 660 pixels, whereas only five pixels were used to describe a 9000 km 2
watershed on a global scale.
On the global scale, significant correlations were further determined
between observed sediment yields and the different environmental features for a set of 60 rivers (Ludwig & Probst, 1996, 1998). The range
of sediment yields (4 - 2 500 t.knTcyC 1 ) and the different types of
watersheds (from Arctic to Himalayan) make it possible to cletermine
the significant features for different climatic zones. For the Bay of
Biscay, the best empirical model is that of a temperate wet climate:
Sediment yield = 000.83 (slope. Fournier index).
Similar relationships were not determined on a regional scale, since no
significant correlation between sediment yielcls and basin features were
found. This is probably due to a shorter sediment yield range: from 5
to 150 t.km' 2 .yr _ 1 . Moreover, the watershed features still differed significantly especially between the smallest watersheds, but sediment yield
reliability decreases with basin size. Thus, spatial resolution of the
available databases is probably not the limiting factor.
On the other hand, environmental information for the South-West of
France in available databases is limitecl. For example, LUC was described
250
Most of the time, only mean annual values are available and sampling
conditions and frequencies are not well known. The reliability of such
data is not easy to determine. On our regional scale, the program Classes
Ecofleuves was a TSM survey which provided several data series on
TSM fluxes and water discharge spanning the same period. Clearly, the
values statistical significance could be calculated, but it was also important to estimate the quality of the TSM flux estimations from metadata
and from the coefficient of TSM-Q regressions (Maneux, 1998) for each
calculated sediment yield value in order to determine which data sets
were significant for the period studied. Indeed, only 30 sediment yields
out of 50 monitored gauging stations were selected. The main difficulty
is to get reliable TSM fluxes from drainage basins witli surface areas
under 2000 square kilometres. Nevertheless, on smaller scales, such as
the Nivelle river watershed (165 km 2 ), the spatial correlation can be
performed with point TSM concentrations recorded along the stream
during flood events (De Maisonneuve et al., 1997). The reliability of
such TSM data is good but the spatial and temporal significance is not
well known.
In the same way as the Universal Soil Loss Equation ( Wishmeier et al.,
1958) and the global sediment flux modeling of Ludwig & Probst
(1996, 1998), morphological (slope, altitude), climatological (Fournier index), soil erodibility (texture, lithology) and Land Use and lancl
Cover databases (LUC) were gathered in a GIS in order to describe each
watershed (Band, 1989; Corine, 1992). On the regional scale, the spatial resolution seems sufficient to describe the watershed features. The
minimum spatial integration unit per watershed seems to be statistically
significant. Indeed, the smallest watershed ( 165 km
2 ) could be described
by 660 pixels, whereas only five pixels were used to describe a 9000 km 2
watershed on a global scale.
On the global scale, significant correlations were further determined
between observed sediment yields and the different environmental features for a set of 60 rivers (Ludwig & Probst, 1996, 1998). The range
of sediment yields (4 - 2 500 t.knTcyC 1 ) and the different types of
watersheds (from Arctic to Himalayan) make it possible to cletermine
the significant features for different climatic zones. For the Bay of
Biscay, the best empirical model is that of a temperate wet climate:
Sediment yield = 000.83 (slope. Fournier index).
Similar relationships were not determined on a regional scale, since no
significant correlation between sediment yielcls and basin features were
found. This is probably due to a shorter sediment yield range: from 5
to 150 t.km' 2 .yr _ 1 . Moreover, the watershed features still differed significantly especially between the smallest watersheds, but sediment yield
reliability decreases with basin size. Thus, spatial resolution of the
available databases is probably not the limiting factor.
On the other hand, environmental information for the South-West of
France in available databases is limitecl. For example, LUC was described
250