SOIL EROSION MODELS
= give quantitative estimates of the amount of erosion for a given area over a given period (digital
models)
Kind of models
- Digital soil erosion models: dominant models, they are used for simulations (approximations
of the real world) that can be divided into forecasts (simulations that are intended to reflect
on the actual conditions now, in the future and in the past) and predictions (simulations
under conditions with a certain likelihood).
- Physical models: soil erosion process is scaled down to be studied in laboratory.
Nature of models
- Empirical: the simulated erosion rates are based on statistical relationships between input
variables and observed erosion rates.
- Physically based: erosion models in which erosion is computed with equations that are
based on physical principles (conservation of mass, energy and momentum).
Context of models
- Prognostic: the simulations are intended to reflect what could happen (predictions and most
forecasts).
- Diagnostic: soil erosion models are used to understand what the influence different
processes or conditions could have on the amount of erosion.
Use of models
- Screening: explorative use of soil erosion models to identify locations or conditions that are
more or less prone to soil erosion. The models can be simpler and focus on the main drivers
of soil erosion.
- Assessment: quantify and compare the actual amount of erosion between locations or
treatments and are more detailed and accurate.
Scale of models
- Spatial scale: the scale of a given process in the landscape (point scale = closed plot, hillslope
scale = open plot and catchment scale)
- Temporal scale: the scale of a given process over time (steady-state(graded time) = soil
erosion models that estimate soil loss as a long-term average, transient(continuous time) =
soil erosion models that estimate soil loss as a rate over time, event-based = models focus on
a specific time period (e.g. a rainstorm)
Spatial resolution of models
- Lumped: define a single estimate of soil loss for the area of application (point, slope or
catchment) and are often on graded time. (empirical)
- Spatially distributed: the input is spatially (and temporally) variable so local effects on soil
and vegetation are better captured. They ned a lot of data and time and has
uncertainties(physically)
- Semi-distributed: the input requirements are reduced by changing spatially distributed to
this. The finer details are aggregated into larger fuctional units.
, Boxes of models
- Black-box: testing with zero knowledge (empiral)
- Grey-box: testing with some knowledge, all soil models include a level of empiricism so they
are a grey-box
- White-box: testing with full knowledge (physically based)
Kind of models:
- Deterministic: same input always leads to the same output.
- Stochastic: covers uncertainties in the input conditions. Probability frequency functions in
the input conditions are translated into the probability of the computed erosion range. It
often leads to a range.
Validation = erosion models should be compared and tested with observed erosion rates.
Calibration = models that are not based on underlying measurements or physical boundaries,
need fixing to improve their performance.
Continuity of erosion and deposition by water
Flow diagram of water-induced erosion(right) and the interpretations in terms of detachment and deposition (left)
Slopes
- Transport limited: produce more material that can be eroded and become mantled by soil.
Detachment > transport capacity
- Production limited: (weathering limited) cannot produce enough material and become
stripped of soil, resulting in barren slopes near the watersheds,
Detachment < transport capacity
= give quantitative estimates of the amount of erosion for a given area over a given period (digital
models)
Kind of models
- Digital soil erosion models: dominant models, they are used for simulations (approximations
of the real world) that can be divided into forecasts (simulations that are intended to reflect
on the actual conditions now, in the future and in the past) and predictions (simulations
under conditions with a certain likelihood).
- Physical models: soil erosion process is scaled down to be studied in laboratory.
Nature of models
- Empirical: the simulated erosion rates are based on statistical relationships between input
variables and observed erosion rates.
- Physically based: erosion models in which erosion is computed with equations that are
based on physical principles (conservation of mass, energy and momentum).
Context of models
- Prognostic: the simulations are intended to reflect what could happen (predictions and most
forecasts).
- Diagnostic: soil erosion models are used to understand what the influence different
processes or conditions could have on the amount of erosion.
Use of models
- Screening: explorative use of soil erosion models to identify locations or conditions that are
more or less prone to soil erosion. The models can be simpler and focus on the main drivers
of soil erosion.
- Assessment: quantify and compare the actual amount of erosion between locations or
treatments and are more detailed and accurate.
Scale of models
- Spatial scale: the scale of a given process in the landscape (point scale = closed plot, hillslope
scale = open plot and catchment scale)
- Temporal scale: the scale of a given process over time (steady-state(graded time) = soil
erosion models that estimate soil loss as a long-term average, transient(continuous time) =
soil erosion models that estimate soil loss as a rate over time, event-based = models focus on
a specific time period (e.g. a rainstorm)
Spatial resolution of models
- Lumped: define a single estimate of soil loss for the area of application (point, slope or
catchment) and are often on graded time. (empirical)
- Spatially distributed: the input is spatially (and temporally) variable so local effects on soil
and vegetation are better captured. They ned a lot of data and time and has
uncertainties(physically)
- Semi-distributed: the input requirements are reduced by changing spatially distributed to
this. The finer details are aggregated into larger fuctional units.
, Boxes of models
- Black-box: testing with zero knowledge (empiral)
- Grey-box: testing with some knowledge, all soil models include a level of empiricism so they
are a grey-box
- White-box: testing with full knowledge (physically based)
Kind of models:
- Deterministic: same input always leads to the same output.
- Stochastic: covers uncertainties in the input conditions. Probability frequency functions in
the input conditions are translated into the probability of the computed erosion range. It
often leads to a range.
Validation = erosion models should be compared and tested with observed erosion rates.
Calibration = models that are not based on underlying measurements or physical boundaries,
need fixing to improve their performance.
Continuity of erosion and deposition by water
Flow diagram of water-induced erosion(right) and the interpretations in terms of detachment and deposition (left)
Slopes
- Transport limited: produce more material that can be eroded and become mantled by soil.
Detachment > transport capacity
- Production limited: (weathering limited) cannot produce enough material and become
stripped of soil, resulting in barren slopes near the watersheds,
Detachment < transport capacity
