1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Near Real-Time <strong>Modelling</strong> <strong>of</strong> Regional Scale Soil Erosion 163<br />
model analyses <strong>of</strong> meteorological parameters. This process for warm cloud precipitation<br />
estimation takes into account surface wind direction, relative humidity and terrain. The<br />
combined technique incorporates rainfall from both <strong>the</strong> convective and <strong>the</strong> stratiform cloud<br />
types, producing a final estimate <strong>of</strong> total accumulated precipitation.<br />
The main problem <strong>of</strong> implementing <strong>the</strong> SCS model with <strong>the</strong> FEWS rainfall data is that<br />
<strong>the</strong> model is only applicable for rainfall events whereas <strong>the</strong> rainfall data are on a decadal<br />
timestep. To overcome this, <strong>the</strong> model was adjusted to run on a decadal basis where OF(i)<br />
is calculated in <strong>the</strong> following manner (Zhang et al., 2002):<br />
OF(i) =<br />
n∑<br />
rOFpJ (6)<br />
i=1<br />
OFp is overland flow per n different classes <strong>of</strong> rainfall intensity, i = 1, ...,n,<br />
r = (r max − I a )<br />
(7)<br />
n<br />
r max is <strong>the</strong> maximum rainfall per rain day which is assumed to be 500 mm, and J is <strong>the</strong> rain<br />
day frequency density function which is assumed to be:<br />
J = J 0<br />
e − r i<br />
r 0 (8)<br />
r 0<br />
where J 0 is <strong>the</strong> total number <strong>of</strong> rain days per month, r i is <strong>the</strong> rainfall per rain day and<br />
r 0 is <strong>the</strong> mean rainfall per rain day (mm). This methodology allows <strong>the</strong> use <strong>of</strong> FEWS<br />
data in <strong>the</strong> overland flow model; however, a map <strong>of</strong> <strong>the</strong> number <strong>of</strong> rain days per decad is<br />
required. We produced this by interpolating <strong>the</strong> daily GTS raingauge data using indicator<br />
Kriging (Symeonakis, 2001). Indicator Kriging maps <strong>the</strong> probability <strong>of</strong> a pixel being wet.<br />
By defining a probability level <strong>of</strong> 50% this field was transformed to a binary one, with ones<br />
where rain occurred and zeros where it did not for each day <strong>of</strong> <strong>the</strong> month. These maps were<br />
<strong>the</strong>n summed into decads and used to implement <strong>the</strong> overland flow model.<br />
8.2.2 Sediment Transport and Routing<br />
The process <strong>of</strong> soil erosion by water includes soil detachment, transport and deposition. So<br />
far we have only modelled soil detachment and availability for transport but not transport<br />
itself or when and where it will be deposited. To estimate sediment input into <strong>the</strong> lake we<br />
have to consider <strong>the</strong>se factors.<br />
Eroded soil can be divided into two types: (i) sediment deposited when <strong>the</strong> sediment<br />
concentration is higher than <strong>the</strong> transport capacity or (ii) sediment transported into <strong>the</strong> lake<br />
when <strong>the</strong> concentration <strong>of</strong> sediment in <strong>the</strong> overland flow is less than <strong>the</strong> transport capacity.<br />
Sediment transport capacity can be calculated using models such as <strong>the</strong> European Soil-<br />
Erosion Model (EUROSEM), Water Erosion Prediction Project (WEPP) and <strong>the</strong> sediment<br />
continuity equation (Kothyari et al., 1997). However, such models require numerous input<br />
parameters, many <strong>of</strong> which cannot be readily derived at <strong>the</strong> regional scale. Therefore, such<br />
models are not currently suitable for regional scale modelling, particularly in Africa where<br />
input data are relatively sparse.<br />
An alternative way to model deposition is to estimate <strong>the</strong> sediment delivery ratio (Dr),<br />
<strong>the</strong> ratio <strong>of</strong> sediment yield to erosion. The delivery ratio <strong>of</strong> each cell in a catchment can