Geography Reference
In-Depth Information
Figure 10.21. (Left) Clay content averaged over the top 40 cm of the soil across the Illinois River basin in Oklahoma, USA, obtained from
STATSGO. (Right) Corresponding saturated hydraulic conductivity, derived from pedo-transfer functions. From Li et al.(
2012
).
Richards equation (or similar equations), soil data are also
widely used to set a-priori parameters of conceptual
hydrological models where the model parameters are
defined at the catchment or element scale rather than at
the local scale. Koren et al.(
2000
) and Anderson et al.
(
2006
) developed a framework in which the model param-
eters of a conceptual model were related to observable soil
properties. They assumed that plant-extractable and gravi-
tational soil moisture can be derived from soil properties
such as saturated moisture content, field capacity and
wilting point, which in turn can be estimated from domin-
ant soil texture available in spatial grids for different soil
layers. Mapping from soil texture to physical property is
performed via a look-up table constructed from empirical
relationships documented in Clapp and Hornberger (
1978
)
and Cosby et al.(
1984
). The results of Anderson et al.
(
2006
), Mednick (
2010
) and Zhang et al.(
2011
) indicate
that the model predictions can be further improved when
based on a finer-scale soil database and combined with
high-resolution land cover data. As suggested by Tesfa
et al.(
2009
), detailed experimental soil observations at
the catchment of interest may considerably improve esti-
mates above those derived from the SSURGO data set and
also allow the creation of spatial patterns of soil character-
istics on a small catchment scale.
Droogers,
2000
). Land cover classes, in turn, can be esti-
mated from a range of satellite data based on indices such
as the normalised difference vegetation index (NDVI).
There exist numerous satellite (e.g., AVHRR and Landsat)
based land cover maps such as the European CORINE land
cover data set (
Büttner et al., 2002
), data sets for North
America (see e.g., Gallo et al.,
2001
) as well as global data
sets (e.g., Hansen et al.,
2000
;
Tucker et al., 2004
, see
Chapter 3
). Recent developments use LiDAR (light
detecting and ranging) data to identify the vegetation struc-
ture. LiDAR measures the signal travel time of laser pulses
between the terrain and the airborne platform, which is
directly proportional to the distance from which the struc-
ture of the microtopography and vegetation can be
inferred. LiDAR methods have been developed for map-
ping of vegetation in forests, shrublands and other land-
scapes (Farid et al.,
2008
; Mitchell et al.,
2011
; Eysn et al.,
2012
). Also, LiDAR has been combined with other
remotely sensed data for vegetation classification to give
more detailed information on the canopy characteristics
(e.g., Puttonen et al.,
2011
). The main strength of LiDAR
data for hydrological modelling is the high spatial reso-
lution (e.g., Cobby et al.,
2003
).
Surface roughness and hydraulic geometry
Surface roughness parameters such as Manning
s n are
usually determined by in-situ experiments on irrigation
plots (e.g., Hessel et al.,
2003
). If measurements are
unavailable, tabulated values in the literature are often used
that are a function of land cover and sometimes topo-
graphic slope (Engman,
1986
). Ideally, the locality to
which the roughness values are applied should be similar
to those where they have been measured. Land cover type
'
Vegetation characteristics
The leaf area index, the fraction of green vegetation, and
the fraction of absorbed photosynthetic active radiation are
vegetation characteristics that can be used in rainfall
-
runoff models to estimate evaporation. These vegetation
characteristics can be related to land cover classes although
the relationships are not always unique (e.g., Kite and
Search WWH ::
Custom Search