Cascada Iterativa versión #2 para la aplicación

The severe problems due to soil erosion in the USA at the beginning of the 20th century (cf. Section 1.1.2) initiated the development of erosion models. According to Richter (1998),

the first practically usable approach is represented by theUSLE(Wischmeier & Smith, 1965).

TheUSLEis an empirical equation, derived from extensive measured data sets. It calculates

the mean annual soil loss based on factors comprising rainfall, soil erodibility, slope length, slope gradient, cropping management and erosion control practice. In recent decades the

interest in off-site effects has grown and as theUSLEis not capable of modelling transport and

deposition, numerous derivatives and advancements for the equation have been developed. These models are based on process-oriented, physically-based approaches, but often rely on

some components from theUSLE. As recently as in the 1990s another generation of models

evolved, also on a physical basis, but intentionally avoiding the inclusion of any components

of theUSLE. Tab. 4.4 summarises some of the established models and their basic properties.

For a more extensive overview of erosion models the reader is referred to Merritt et al. (2003).

Tab. 4.5 shows the most important factors influencing soil erosion. Soil erosion models consider these factors to various degrees.

Empirical models usually calculate soil loss with only a few of the factors, by statistically cor- relating those, which are considered by the developer to be the most important ones, to soil loss. This requires extensive field measurements and hence is limited to the regional condi-

tions where it was developed. The most popular example is theUSLE, which was developed in

the USA over 20 years, but in order to be applicable in other regions, it was comprehensively adapted (e.g. for Germany by Schwertmann (1981), West Africa, or India (both stated in Klisch (2003))). The advantage of such an approach, once it is adapted, is the user-friendly applicability, which often even allows for calculation in situ by the farmer himself, i.e. without

extensive hard- and software requirements (e.g. anLfL-booklet (LfL, 2005)). A drawback is,

that only a long-term annual mean value can be calculated, i.e. simulation of single events is not possible, and further on, temporal and spatial variability of input parameters is not con-

sidered. This basically renders e.g. theUSLE unsuitable for Global Change assessment,

respectively requires further adaptations, as e.g. incorporated in theRUSLE.

Physically based models in contrast, make use of many more input parameters, often even more than those listed in Tab. 4.5, and link these with physical equations to calculate erosion processes. The vast number of input parameters they require (which in contrast are implicitly included in empirical equations in their lumped parameter set), usually demands a lot of work until they can be operated on the desired plot or catchment. The physical basis of such models generally provides them with spatial and temporal independence, making them universally applicable in theoretically any arbitrary region. Nevertheless they often rely on or incorporate empirical components, as the processes involved are too complex to model them on a truly physical basis. But process orientation and the usage of physical base units makes sub-components of the models more exchangeable and extensible.

Tab. 4.4: Basic properties of selected erosion models (compiled from_Richter(1998);_Klisch(2003);

Deinlein & Böhm(2000) and quoted literature, pb: physically based).

Model Approach

(detach- ment/transport)

Spatial represen- tation

Temporal basis Developer

USLE empirical (n/a) plot/field long-term annual

mean

Wischmeier &

Smith (1965)

ABAG USLEadaptation plot/field long-term annual

mean

Schwertmann (1981)

RUSLE revised eqs. of

USLE, pb

plot/field long-term annual

mean Renard et al. (1996) CREAMS USLE- based/transport capacity, pb

plot/field, pb single event Kinsel (1980)

EPIC USLE and deriva-

tives, pb

plot/field continuous Williams et al.

(1983)

EUROSEM transport capac-

ity/sediment con- centration, pb

plot/field & catch- ment

single event Morgan et al.

(1998)

EROSION 2D momentum

flux/transport capacity, pb

plot/field single event Schmidt (1996)

WEPP shear

stress/transport capacity, pb

plot/field & catch- ment

continuous Flanagan &

Nearing (1995)

KINEROS transport capac-

ity/sediment con- centration, pb

watershed single event Woolhiser

et al. (1990)

Tab. 4.5: Basic factors influencing soil erosion by water (Klisch, 2003).

Climate Relief Soil Vegetation Management

rainfall intensity slope length particle size dis-

tribution

land use date of tillage

rainfall duration slope gradient bulk density canopy cover type of tillage

rainfall frequency slope shape soil structure vegetation height depth of tillage

catchment area water content rooting depth protection mea-

sures organic matter content permeability cation exchange capacity shear strength sealing plant litter roughness

gle long-term annual mean value (empirical models), or on a single event (physically based models). Some models are also capable of a continuous simulation, as they include sub- components for dynamic computation of plant growth, soil consolidation, etc. Concerning the spatial representation, models deal with input parameters and process calculations ei- ther in a lumped or a distributed way. Lumped models treat the input parameters as lumped

over the whole area of analysis, e.g. theUSLEdescribes a plot or a field with a single param-

eter set and calculates the output for this unit. Distributed models reflect the spatial variability by sub-dividing the whole area into smaller units, which are calculated (and parameterised) separately. This distribution may consist of homogeneous irregular-sized units, or a regular grid.

In the early years of development of erosion models, these have been kept simple, because extensive use of computer aided calculation was not available. But also nowadays these early models are often used for assessment of soil erosion or incorporated in other models, espe-

cially when considering large areas (e.g. theUSLE used in Modelling Nutrient Emissions in

River Systems (MONERIS) (Behrendt et al., 2007)). Application of physically based models

above the watershed/catchment scale is rarely executed, on the one hand due to the exten- sive parameterisation, and on the other hand these models usually need calibration data to work properly, which is not available at larger scales. So generally only for modelling on the catchment scale or at smaller scales, there is a clear trend to physically based modelling, as they respond to single events and allow for physical interpretation of modelling results.