Supported by USA CERL developed with GRASS5.0

Landscape soil erosion modeling for spatial conservation planning:
GIS-based tutorial

Under preparation by
Helena Mitasova, GMSL UofI, MEAS NCSU, Bill Brown GMSL UofI

Co-Authors:
Matt Hohmann, D. Gebhart, S. Warren (USA CERL), D. Jones (Ft. Hood), Fred Schrank(NRCS)

TODO;

- redesign according to the FG regulations
- add side frame for easy navigation
- split into sections, add accompanying slides
- prepare the images for linking (photo's, 3D views, maps)
- add the text
- link-in the tutorials for the models
- zmen styl textu -ask questions and answers, oslov citatela

1. Introduction

Erosion modeling with GIS focuses on describing spatial distributions of sediment flow, soil detachment and net erosion/deposition. Capability to predict the pattern of erosion and deposition and to identify the location of high risk areas for various land use alternatives is critical for effective land use management. Spatial analysis and simulation can also provide supporting information for allocation of resources to those areas and those types of practices which will provide the most effective protection.

The emphasis on integration of local actions with watershed-scale approaches has a significant impact on the development of supporting GIS and modeling tools. Complex, distributed, physics-based models are needed to improve understanding and prediction of landscape processes at any point in space and time. At the same time, land owners and managers working in the watersheds and fields need fast and easy to use models for which the input data are readily available.

To reflect the need for modeling at different levels of complexity a set of models was developed (Mitas and Mitasova, 1998; Mitasova et al., 1999). The simple models RUSLE3D (Revised Universal Soil Loss Equation for Complex Terrain) and USPED (Unit Stream Power-based Erosion Deposition) are based on modifications of well established equations representing special cases of erosion regimes. The basic empirical parameters for these models are available, however their applicability to a wide range of conditions is limited. The new distributed, process-based model SIMWE (SIMulated Water Erosion) provides capabilities to simulate more complex effects, however both the experimental and theoretical research are still very active and underlying equations, as well as the input parameters,are under continuing development.

This tutorial focuses on the the simpler models with some links and comments on the more complex simulations. The comparison of these models with traditional model results and field observations are in the related document at: modviz/hoodn.html : owl creek and house creek

2. Multiple scale approach

Geographic Information Systems linked with landscape process models support coordination of conservation efforts at different management levels. They provide tools for evaluation of land use alternatives at both local and watershed levels and planning of prevention practices based not only on its, but also on the location within the watershed.

In the following section we provide a quick reference to models in this tutorial and on the Internet which are applicable at different scales to support conservation planning at a hierarchy of land management levels. We then describe selected models and provide step-by-step instruction on how to run them in GIS.

2.1 Look-up table for multiple scale approach

The following table explains the type of applications/tasks that are solved at different resolutions, relevant data needed, types of processes that need to be taken into account and examples of models that can be used for each resolution/level of detail. It has been designed for quick access to the relevant parts of the tutorial . More detailed description of each item is in the following sections of the tutorial. Models shown in bold are covered by this tutorial, links to Internet resources are provided for the rest of the models.

Scale / unit	Regional/large watershed XX sq. miles/km (e.g., installation : map)	Landscape/small watershed XX sq miles/acres/ha (e.g., training area: map)	Field /hillslope XX acres/square m (e.g. hillslope, map)
Tasks	assessment of averaged erosion risk for entire region/installation identification and planning for: high risk subwatersheds large conservation areas	identification of hot spots in watersheds location of depositional areas: photo) location of major concentrated flow areas: photo) planning vegetation based conservation measures (dense vegetation, stream buffers: photo) planning potential locations of sedimentation ponds/constructed wetlands: photo)	detailed erosion, deposition pattern including the effects of conservation measures: photo) detailed impact of erosion on roads(photo) design of measures (photo)
Data	Spatially averaged polygon (hydrologic unit/subwatershed) data Distributed 100-30m resolution grid	Distributed existing and planned use at grid level 10-20m resolution data or detailed polygons	Distributed existing and planned use including man-made features 1-2m resolution raster + linear features/borderlines
Processes	flow in rivers and streams depressions are neglected spatial variability is averaged out	overland flow (averaged sheet+rill erosion) concentrated flow (gullies) stream flow (averaged bank erosion) large depressions included	overland flow (sheet, indiv. rills) concentrated and preferential flow (indiv. gullies, roads, tracks) stream flow (specific bank erosion) all depressions included
Models	Spatially averaged e.g., SWAT Distributed RUSLE3D (simplified)	Distributed RUSLE3d USPED SIMWE	Distributed RUSLE3d USPED SIMWE RILLGROW2 WEPP

2.2 Theory and algorithms used in the models

To model spatio-temporal distribution of sediment transport and erosion/deposition at any point and time, a complex system of interacting processes has to be simulated, including rainfall events, vegetation growth, surface, subsurface and ground water flow, soil detachment, transport and deposition. Excellent examples of continuous time simulation systems, which integrate a wide range of interacting processes important for land use management are WEPP , LISEM, or SWAT modeling systems. Spatial components of these systems are usually based on one dimensional routing of water and sediment through homogeneous hydrologic units (e.g., hillslope segments, subwatersheds), limiting the range of spatial effects that can be simulated. To more accurately capture the impacts of spatially variable conditions a new generation of hydrologic and sediment transport models introduced 2 dimensional flow routing capabilities (e.g., SIBERIA, CASC2d; CHILD, SIMWE).

To keep the tutorial simple and easy to use we focus on the soil detachment, sediment transport and deposition processes while providing links to documents which describe the relevant interacting processes, such as rainfall, infiltration, water flow, vegetation growth, etc.

2.2.1 Erosion processes at multiple scales and related models

Similarly as other landscape processes erosion has multiscale character. Different processes are dominant at different scales and it is therefore important to apply the models at the scales for which they were designed. The scales can range from molecules through raindrops, plots, fields, watersheds, regions to entire continents, however, from the point of view of landuse managment we will discuss the scales ranging from field to the region. For each scale it is possible to model full dynamic of the processes during a rainfall event or a steady state.

ADD DHI MODELS!!!

a) plot and field scale, resolution 1cm-1m:
At this scale the sheet, rill and gully erosion (PHOTOS) is distinguished and should be modeled individually. In more detailed, advanced models rills and gullies are dynamic, evolving 3D features and change in elevations (topography) is captured even for short time periods (rainfall event). Spatial variability in land cover (vegetation density, canopy, stones, roots) are important as well as the human impact such as vehicle tracks or ditches.All conservation measures are represented by their shape and properties - e.g., contour filter strips, grassways, hedges, dry dams. The goal of the modeling is detailed assesment and prediction of different types of erosion and deposition and especially design of conservation measures.

include side-by-side photo+3dview+map

RillGrow2: Land Degradation and Rehabilitation Programme, University of Oxford Environmental Change Institute
EUROSEM
WEPP (hillslope)
SIMWE

b) landscape scale, small watershed, resolution 5m-20m
At this scale sheet, rill and small gully erosion are averaged and modeled as overland flow erosion, only large concentrated flow erosion features are captured. Variability in topography (covex/concave features) are important while net erosion and deposition is averaged for larger topographic forms. Spatial variability in land cover is averaged and only larger man made features are captured. Man made features, conservation measures and other land cover impacts that would require higher resolution (dirt roads, ditches, tracks, grassways, hedges) are usually captured through factors increasing or reducing detachment and sediment transport. For short term modeling the changes in topography is neglected. Stream routing may be included. The goal of the modeling is conservation/land use planing and erosion/deposition risk assessment.

photo+map

USLE, RUSLE, RUSLE3D
SIMWE
Govers erosion modeling web site
SedSpec
WEPP (small watershed) on-line

c) regional scale, large watersheds
At this scale sheet, rill and concentrated flow erosion are averaged, often into homogeneous hydrologic units (subwatersheds and their subareas) as total soil detachment and sediment transport. Impact of spatial variability in land cover is averaged and often expressed on as a percentage of given area independent of the location. Stream processes dominate at this resolution.

photo+map

SWAT
RUSLE/ANSWERS australia
Conservation measures websites (see bookmarks)

2.2.2 Erosion regimes and relevant models

The amount of eroded and transported soil is controlled by the capacity of water flow to detach and move soil through the landscape. Depending on the relation between the the detachment and transport capacity of flow erosion processes can be modeled as

a) detachment limited: we assume that the water flow can transport unlimited amount of sediment and amount of eroded soil is limited only by the capacity of water to detach soil. Because of this assumption no deposition is predicted (FIGURE). USLE, RUSLE model this case. These estimates can be also handled as estimates of soil detachment and input to models which will route the detached soil through the watershed (e.g. CASC2d, ANSWERS).
b) transport capacity limited : we assume that the water flow transports as much sediment as it possibly can (at its capacity), therefore wherever the transporting capacity increases also erosion increases and where transport capacity decreases, sedimentation occurs (FIGURE).USPED. Moore et al. models, geomorphological models (examples) .
c) general models which simulate erosion processes between these two extreme cases (SIMWE, CASC2D, WEPP, LISEM,...), see animation

Detachment limited case is more probable during large events when transport capacity of water flow is large, transport capacity limited is more typical for evolution of long term pattern which includes impact of both large and small events. Also detachment limited case is more typical for soils with large portion of very small particles such as clay (assuming minimum agregation) while transport capacity limiting case is highly probably for sandy soils with larger particles. add photo with erosion and deposition Special cases a) and b) are in general easier to compute than the case c) and can be used as minimum and maximum extent of deposition for the given landscape - areas which are identified as high risk from both models should be targeted for field inventory and if the high risk is confirmed, conservation measures should be implemented in these areas.

2.2.3 Models

From empirical, spatially averaged towards process-based, distributed. Process based principles in empirical models (physical representation of LS factor...)

Selected models and the accompanying equations are described in the follwing subdocuments:

links to additional models are provided in the table and references

3. Runing the analysis in GIS

3. 1 Installation scale spatially averaged modeling:

see SWAT

3. 2 Installation scale distributed, low resolution (30-90m)

preparing the data:

DEM,
soils,
land cover,
rainfall,
prevention measures

RUSLE3D (simplified)

ArcGIS8
GRASS5

USPED (simplified)

ArcGIS8,
GRASS5

analyzing, interpreting and presenting the results:

maps with continuous values,
maps with categories,
maps of hot-spots,
maps of erosion status indicators (see Hohenfels)
summary reports and histograms

Planning applications

roads under risk of increased sedimentation and damage by erosion
wetlands, ponds and lakes with high risk of unsustainable sedimentation
training areas requiring increased maintance or redesign
...

3.3 Lanscape/subwatershed scale, moderate resolution (10-20m)

preparing the data
RUSLE3D

ArcGIS8,
ArcView3
GRASS5

USPED

ArcGIS8
ArcView3,
GRASS5

analyzing, interpreting and presenting the results
planning applications

identification of high erosion risk hot spots including concentrated flow erosion
eroding roads
stream buffers
vegetation planning for reduction of sediment delivery into wetlands
....

3.4 Small watershed/field/design scale, high resolution

preparing the data
RUSLE3D

ArcGIS8
GRASS5

USPED

ArcGIS8
GRASS5

WEPP - access on-line version
analyzing, interpreting and presenting the results
planning applications

location, size and shape of protective vegetation areas
erosion on roads and its prevention

3.5 Modeling with spatially variable resolution:

finite element/finite difference meshes (WMS)
path sampling and nested grids (SIMWE)

4. Notes on preparation of data

The data needed for erosion modeling are often already available from other projects and mapping efforts. Then simple checks of their suitability is sufficient and the methods outlined above can be directly applied. However, in some cases further processing of data is needed to extract the necessary parameters with sufficient accuracy and realism. It is impossible to address all issues that can arise so the focus of the next sections will be on the most common problems and approaches to solution.

Erosion, sediment transport and deposition involves complex interactions between rainfall, surface and subsurface hydrology, soil properties, land cover and topography. Modeling erosion and deposition in complex terrain using GIS therefore requires adequate digital elevation model (DEM) as well as digital data describing spatial distributions of rainfall, soil and land cover/land use properties.

4.1 Digital elevation model

Types of digital elevation data and their suitability for erosion modeling (link examples, any other types relevant to installations?):

USGS DEM,
IFSARE,
LIDAR,
Digital photogrammetry (points/TIN, contours)
Rapid kinematic survey data
Ground survey

Finding the elevation data on Internet
Assessing the quality of DEM (include link/reference to Joe Wood's work)
Selecting the resolution
Reinterpolating, smoothing and stream enforcement(compare the estimates from integer and FP DEM)
Deriving the model parameters (algorithms, impact of resolution, relative importance, comparison with field data (LCTA, NRC inventory):

slope,
aspect (direction of flow)
upslope area (steady state water flow)

4.2 Land use/land cover

USGS 30m Landsat based data
New 1m vegetation maps (note of 10m resolution DEM used and uniform buffers along streams for bottomland forest)
Estimating the C-factor for GIS-based data
Resampling to lower and/or higher resolution, when to resample, sharp versus smooth boundaries, smoothing versus aggregating
Finding the land cover and C-factor data on the Internet

sources, resolution, level of detail - broad categories at 20-30m resolution, %vegetation cover for design at sites (1m resolution)

4.3 Soils

SSURGO data - K-factor based on soil map (polygons)
Continuous maps of soil properties - future trend
Soil data on Internet

4.4 Rainfall

Rainfall - R-factor - annual, monthly, storm

5. Notes on running the models

5.1 Selecting the level of detail, resolution and appropriate model (explain resolution in the equations - area versus width, issue of running large area at high level of detail - loosing the "big picture")
5.2 Selecting the model parameters (exponents)
5.3 Trouble shooting -

treating unrealistic erosion rates,
avoiding/fixing noisy results
checking for artificial patterns

5.3 Calibrating the model (future?, data are needed for this)

6. Notes on creating and analyzing the resulting maps

6.1 Continuous value maps - what they mean, legend, color tables (automatic creation of exponential legend with intesity given by parameter p=0.01, colors see Cebecauer/GRASScurvatures for erosion/deposition)

exponential color table for soil detachment (RUSLE3D) in tons/(acre.year) with m=0.6 and n=1.3 (detachment value=exp(n), where n=1,2,3.... and each n has a color associated with it.

0 white : stable

2 light green : low

7 yellow : moderate

20 orange : high

50 red : severe

150 magenta : extreme

500 violet : upper limit

6.2 Class maps - reclassifying the continuous maps, standardized classes. Need for classification for the management purposes

6.3 Sumary statistics, reports:

histograms,

%area from each class,

average rate,

total soil detached, total net soil loss, ...

7. Notes on applications for planning

Change laduse/implement conservation measure,...- installation wide, sub-installation-landscape scale

Minimize detachment and net erosion:

model based: set max detachment treshold - create new cover, (necessary C, suggested cover invertly derived from the table), set max net erosion treshold, set elimination of concentrated flow and other criteria

feature based: set buffers along the streams (uniform with given width, adjusted by model,..), set hedge along contour, grass filter strip, conservation area - compute necessary C and adjust shape,

Analyze high risk locations:

find roads affected by high erosion risk

wetland areas and streams affected by high erosion

deposit sediment: increase sedimentation rate in deposition area, create sedimentation pond, ....

Notes - for the management purposes there seems to be a need for categorization/classification - first streams, land are split into discrete homogeneous units and then they are classified/zoned for what we can do with them - is this the most effective approach??? (this results in uniform buffers, interactions between various landscape phenomena such as streams/topography ignored, etc. this also results in such observations that we have more sediment coming from the forested watershed that from agricultural/developed one.

8. References

Plot and field scale, resolution 1cm-1m:

RillGrow2: Land Degradation and Rehabilitation Programme, University of Oxford Environmental Change Institute

WEPP

Landscape scale, small watershed, resolution 5m-20m

general, overviews, summaries:
Watershed management models (relevant are ANSWERS, AGNPS, WEPP, ...:
http://web.aces.uiuc.edu/watershed/model/models_index.htm

Govers erosion modeling web site
SedSpec
WEPP on-line
SWAT
RUSLE/ANSWERS australia
DHI models
Conservation measures websites (see bookmarks)

Regional scale, large watersheds, resolution 30-100m - stream processes dominate, spatial variability on hillslopes is averaged-out

Publications
Bivand Neteler GRASS+R, show impact of integers depending on resolution
Harmon Doe (Eds) 2001 Lnadscape evolution and erosion modeling. Kluwer.
Proceedings from hawai conference, gully conf. in Belgium etc.

Link-in the following material (some need modifications)

- USLE, USPED on-line tutorials
- GRASSBook erosion modeling
- preparation of data from reports and grassbook
- processing of results from grassbook and reports

Standards

Standardized filed names (e.g. used in ATTAC document) - is there any official list for that?
Standardized categories for erosion rates severe/high/moderate/low/stable
Standardized color tables for inputs and outputs

Design standard for 3D illustrations maps

3D view entire study area

topography
theme (slope, water flow, detachment rate, net erosion/deposition, hot spots)
vector features: streams, roads, inventory sites
legend/north/scale

3D view zoom-it to show the detail

Design standard/template for 2D ArcMap files - has DOD any standard for that?

the sample maps are at:

Bill create the following maps

Add legends/scale/north to images (create template, write script?)
Predicted erosion map + sites sized and colored according to the observed sheet /rill erosion
Predicted erosion map + sites sized and colored according to the observed gully/roaderosion
C-factor map derived from vegetation map + sites colored/labeled according to observed C-factor

This document describes "field to landscape scale" modeling of selected hydrologic and erosion processes based on representation of input data as multivariate functions (as opposed to homogeneous hillslope segments or subwaterhseds), simulations performed by solution of equations describing relevant physical processes, advanced multidimensional GIS support for storing, processing and visualizing the data and results. The concept, theory and some applications are described in the following papers, proceedings, and reports.

Methods

The following methods and tools support hydrologic and erosion modeling within GIS at an increasing level of complexity and realism:

The presented work was supported by the Strategic Environmental Research and Development Program, USA CERL, Illinois Department of Natural Resources and C-FAR.