Supplementary material
A Landscape Planning and Management Tool for Land and Water Resources Management: an example application in Northern Ethiopia
Water Resources Management
Lulseged Tamenea*, Quang Bao Leb, Paul L.G. Vlekc
a International Center for Tropical Agriculture (CIAT), Chitedze Agricultural Research Station, P.O. Box 158, Lilongwe, Malawi. E-mail:
b Natural and Social Science Interface (NSSI), Institute for Environmental Decisions (IED), ETH Zurich, Universitaestrasse 22, CH-8092, Switzerland. E-mail:
c University of Bonn, Center for Development Research (ZEF), Walter-Flex-Str. 3, 53113 Bonn, Germany. E-mail:
* corresponding author
This Online resource gives details related to some of the processing related to deriving data inputs and some results that are intended to support the main manuscript.
Estimation of erosion factors for the RUSLE model
The RUSLE equation
The RUSLE model is more appropriate for developing regions were data of reasonable spatial and temporal resolutions are not easily available. It is has been selected for application in Ethiopia because the USLE soil erosion factors are calibrated for the Ethiopian condition (Hurni 1985). The RUSLE model is used to estimate soil erosion mainly by integrating terrain, soil, rainfall, land use and management factors in the form:
RUSLE (t ha-1 y-1) = RKLSCP(1)
where R = rainfall erosivity (MJ mm ha-1 h-1 y-1); K = soil erdibility (t ha h (ha MJ mm)-1); LS = slope length-steepness (-); C = land use/cover (-); and P = conservation/management (-) factors.
The approaches employed to acquire data for the above fix factors are described below. All the data were integrated in a GIS at a spatial resolution of 10 m2.
Rainfall erosvity (R) factor
The R- factor is defined as the product of kinetic energy and the maximum 30-minute intensity and shows the erosivity of rainfall events (Wischmeier and Smith 1978; Renard et al. 1997). The R- factor is usually calculated using rainfall intensity data. However, such data is not available in most gauging stations especially in developing and remote regions. As a result, different studies calibrated relationship between mean rainfall amount and rainfall erosivity for their respective conditions (Morgan 1974; Stocking and Elwell 1976; Roose 1977; Hurni 1985; Renard and Freimund 1994; Loureiro and Couthino 2001; Petkovsek and Mikos 2004). For this study, the relationship between mean annual rainfall and rainfall erosivity established for Ethiopian condition based on the analysis of monthly rainfall data of different stations (Hurni 1985) was used:
R =-8.12 + 0.562Pr (2)
where R = annual rainfall erosivity (MJ mm ha-1 h-1 y-1), and Pr = average annual precipitation (mm) of nearby stations.
Annual rainfall data acquired over the last 35 years from the nearest rainfall stations of each catchment were used in this study. Since the spatial variability of rainfall within catchments is not significant (e.g., Haregeyewn et al. 2005), single rainfall data of the nearby station to each catchment (within 2-5 km) was used to estimate precipitation of the catchment. Accordingly, Pr = 599mm for Adikenafiz catchment was used to calculate R-values for this study.
Soil erodibility (K) factor
The K-factor refers to rate of soil loss per unit of rainfall on a unit plot and indicates the relative ease with which soil is detached and transported (Renard et al. 1997). Soil erodibility is a function of texture, organic matter content, structure and permeability and can be determined based on (Wischmeier et al. 1971; Renard et al. 1997):
(3)
where K = erodibility factor in t ha h (ha MJ mm)-1; M = particle size parameter = (%silt + %sand)* 100 - %clay); OM = organic matter (%); s = soil structure code (-); p = permeability class (-). Division by 7.59 gives values in SI units.
Equation (3) can be used to derive K-factor if data on soil properties such as texture, organic matter content, structure and permeability is available or can be derived for the areas of interest. For this catchment, information on basic soil properties necessary to derive K-factor was not available. As a result, soil erodibilitis potentials derived based on Eweg and Lammeren (1996); Machado et al. (1996a, b) have been used in this study as shown in Table 1 (main text).
The Slope Length - Steepness (LS) factor
Terrain slope, length, shape and form are key components that determine the energy and direction of flow of running water (Moore and Burch 1986; Desmet and Govers 1996a; Mitasova et al. 1996; Van Oost et. al. 2000; Wilson and Gallant 2000). The LS factor is thus one of the most commonly used representations of the effect of terrain on soil erosion (Wischmeier et al. 1971; Renard et al. 1997). In this study, the LS equation developed by Moore and Burch (1986) and Moore et al. (1991) designed to represent the impact of not only slope-length but also terrain shape in a better way has been applied.
Terrain attributes can be derived using Geographical information system (GIS) and associated hydrological models provided that sufficiently detailed digital elevation model (DEMs) are available. In this study, contours, streams, and spot heights were digitized from contour maps of scale 1:50000 (Ethiopian Mapping Authority 1997) and DEMs were constructed for each catchment using the TOPOGRIDTOOL of ARCINFO (Hutchinson 1989). Considering the contour spacing of catchments (20 m) and in order to obtain a reasonably detailed representation of terrain parameters and their derivations, the DEMs were interpolated at 10 m grid cell size. The spline function with the drainage enforcement interpolation facility in the TOPOGRID was applied to compute hydrologically correct DEMs with minimum sinks (Hutchinson 1989; Quinn et al. 1991; Martz and Gabreecht 1992; Zhang and Montgomery 1994). After the DEM was created, pits/sinks were filled before any processing was undertaken in order to “route” runoff to the catchment outlet without facing “unnecessary obstacles”.
Considering the complexity of the landscape of the study area, equation 4 is used to calculate the LS-factor (Moore and Burch 1986; Moore et al. 1991):
(4)
where m and n are slope length and angle coefficients; As is the specific upslope contributing area per unit length of contour; β is the local slope gradient (degrees).
For all erosion models, the m and n coefficients need to be calibrated for specific areas and for the specific prevailing type of flow and soil conditions (Foster 1990; Moore and Wilson 1992). While larger values of m exponent are associated with increasing concentrated flow and rill erosion, it approaches zero when the dominant processes are sheet and interill erosion (Meyer et al. 1975; McCool et al. 1987). Based on literature, m and n coefficients of 0.5 and 1.3, respectively, were used for catchments more vulnerable to rill and gully formation. For sites with good surface cover, where interill erosion dominated, m and n values of 0.4 and 1, respectively, were used (Moore and Wilson 1992; Liu et al. 2000; Mitasova et al. 2001).
Land-use and land-cover (C) factor
The C- factor is defined as the ratio of soil loss from land with specific vegetation to the corresponding soil loss from continuous fallow (Wischmeier and Smith 1978). To incorporate the impact of surface cover on erosion processes, land use/cover data (LUC) and its translation to C-factor values are necessary. In this study, the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) images of 15 m resolution were used to derive LUC maps of catchments. Different enhancement and transformation such as Normalized Difference Vegetation Index (NDVI), Soil Adjusted Vegetation Index (SAVI) and Principal Component Analysis (PCA) techniques were used to aid adequate separation of cover types. The maximum likelihood supervised classification algorithm in IDRISI (Eastman 2001) was then performed on bands 1, 2, 3, and PCA-1 images to derive LUC types for the study areas. Ground data acquired for representative locations using Global Positioning System (GPS) were used to assess the accuracy of the LUC classification results. The classification produced an overall accuracy value of 82%. The LUC maps were then translated into C-factor values (Table 1) based on calibrations made for the Ethiopian conditions Hurni (1985).
Management/Support practice (P) factor
The P-factor gives the ratio between soil loss with a certain soil conservation practice to that with up-and down-slope ploughing (Wischmeier and Smith 1978). Specific cultivation practices and conservation activities affect erosion by modifying flow amount, pattern and direction. In areas where there is terracing, runoff speed could be reduced with increased infiltration, ultimately resulting in lower soil loss and sediment delivery (Renard and Foster 1983). For this study, field observation was made to assess catchments in terms of conservation practices and identify locations within catchments where major conservation activities exist and evaluate their conditions. Once such data was available, P-factor values were assigned (Table 1) based on values suggested by Hurni (1985) and Eweg and Lammeren (1996) for Ethiopian condition. The dominant part of the catchments was assigned a value of P=1 due to inadequate or poorly maintained conservation activities.
Estimating sediment delivery efficiency potential (SDEP) of catchments
Understanding of the linkages between sediment sources, transport, delivery and deposition in floodplains or reservoirs is essential for the appropriate design and implementation of efficient catchment management options. Generally, the majority of soil eroded from hillslopes will be redistributed within the basin or catchment depending on energy of flow, material type and size, surface gradient and roughness and distance from the ‘source to destination’ (Walling 1983; Dickinson et al.1990; Walling 1990; Lane et al. 2000; De Vente et al. 2006, 2007). Total sediment yield thus depends on erosion at various sediment sources and the efficiency of the system to transport the eroded material. Sediment delivery ratio (SDR) is a commonly used indicator of sediment transport efficiency of watersheds (Walling 1983; Dickinson and Collins 1998; Ferro and Porto 2000; Di Stefano et al. 2005).
Different approaches attempted to estimate sediment delivery efficiency of catchments by integrating basin catchment attributes in a GIS (Ferro and Minacapilli 1995; Ferro et al., 1998; Ferro and Porto 2000; Jain and Kothyari 2000; Fernandez et al. 2003; Ferro et al. 2003; Di Stefano et al. 2005). This study adopts similar procedures to determine sediment delivery efficiency potential (SDEP) of the studied catchments. According to Ferro and Minacapilli (1995), the fraction of the gross soil loss from a given cell that actually reaches a continuous stream system can be estimated as:
(5)
(6)
(7)
Where, β= routing coefficient; ti= travel time from a given cell (h); li= channel length in the flow path and usually equal to the length of the side or diagonal of a cell depending on flow direction in the cell (m); vi= runoff velocity of a given cell (m/s);Ri= coefficients based on surface roughness characteristics (m/s); Si= slope gradient (m/m).
Experimental and simulation studies by Ferro et al. (2003) and Di Stefano et al. (2005) suggest different values of βdepending on basin size, slope and the LS-factor used in determining soil erosion. For this study, we adopted a βvalue of -0.0051, which is suggested for catchments with relatively higher LS-factor estimated based on Moore and Burch (1986) equation (Ferro et al. 2003; Di Stefano et al. 2005). Coefficients for surface roughness characteristics (Ri) are adapted from Haan et al. (1994) and Mutua et al. (2006).
Key features of LAPMAT in supporting adaptive land-use planning
One of the important capabilities of the LAPMAT framework is its visualization power (Fig S1). Once data are pre-processed and prepared in a GIS environment, users can import them to NetLogo for further proessing and analysis. In this sudy the main intension is to import RUSLE factor maps and visualize them to assess if there are any anomalies.
Fig S1 Navigation landscape data with LAPMAT
In addition to checking any issues with input data (maps), LAPMAT is also designed to facilitate end users identify high erosion risk areas and define suitable management options. Initially, users are able to choose the model of interest (in this case RUSLE) and to input (import) the erosion factors associated with the model. Once erosion factors relevant for the model of interest are defined, users will have the option to display and visualize the different drivers of erosion (Fig S2, upper part of block A). The users, for example, can visualize the spatial dynamics of land cover types, terrain configuration (elevation, slope, upslope contributing area, wetness index) and any other spatial data that is available for display. The major purpose of this option is to help users assess the spatial pattern of the drivers and have a general feeling of their possible impacts across landscapes.
After the model parameters are imported and visualized, the interface will automatically adjust its panel to a range of coefficients that are relevant for the model and also offers default values (Fig S2, upper part of block B). For instance, instead of single R-factor for a small catchment, the users will have an option to choose from a possible range. The users are also provided with a range of soil erosion coefficients (e.g., m and n slope length and steepness coefficients), where they can choose one based on literature review for locally calibrated coefficients (see Fig S2, block B). However, to increase flexibility for the user (especially users who are not well exposed to soil erosion theory), commonly used values are provided as default.
Once relevant parameters and coefficients are set, the model can be run and results (rates, graphs, maps) be displayed (Fig S2, blocks C, D, E). The output at this stage represents the rate and spatial pattern of soil loss without land management options. This output can be the basis where users identify the priority areas of management intervention based on their own criteria or considering the tolerable threshold.
After having an impression of the severity as well as spatial patterns of soil loss and possibly identify hotspot areas, users can define and assign suitable management options to tackle erosion and sediment yield. For instance, the users can change (re-design) land use/cover types or introduce other management options such as enclosures or terraces (Fig S2, lower part of block B). Ultimately, the modelling framework allows users to assess the rate and spatial patterns of sediment yield after each management option.
Fig S2 LAPMAT’s Graphical User Interface (GUI) designed to facilitatedata input, input visualization, adjust coefficients, and select simulation options
References
Desmet PJJ and Govers G (1996a) A GIS procedure for automatically calculating the USLELS factor on topographically complex landscape units. Journal of Soil and Water Conservation, 51: 427-443.
Desmet PJJ and Govers G (1996b) Comparison of routing algorithms for digital elevationmodels and their implications for predicting ephemeral gullies. International Journal of Geographical Information Systems, 10: 311-331.
De Vente J, Poesen J, Bazzoffi P, Van Rompaey A, Verstraeten G(2006) Predicting catchment sediment yield in Mediterranean environments: the importance of sediment sources and connectivity in Italian drainage basins. Earth Surface Processes and Landforms 31: 1017: 1034.
De Vente J, Poesen J, Arabkhedri M, Verstraeten G(2007) The Sediment Delivery Problem Revisited. Progress in Physical Geography 31(2):155-178.
Dickinson WT, Rudra RP, and Wall GT (1990) Targeting remedial measures to controlnonpoint source pollution. Water Resources Bull. 26: 499-507
Dickinson A and Collins R (1998) Predicting erosion and sediment yield at thecatchment scale. In: Penning de Vries FWT, Agus F, Kerr J (eds) Soil Erosion at Multiple Scales, Principles and Methods for Assessing Causes andImpacts. CABI Publishing, Wallingford, UK in association with the International Board for Soil Research and Management, pp. 317–342.
Di Stefano C, Ferro V, Minacapilli M(2005) Testing SEDD model in Sicilian basins. Sediment Budgets 2 (Proceedings of Symposium SI held during the seventh IAHS Scientific Assembly at Foz do Iguacu, Brazil, April, 2005). IAHS Publ. 292: 152-161.
Eastman JR (2001) Idrisi32 Release 2, Guide to GIS and Image Processing, Volume 2.
Ethiopian Mapping Authority (EPA) (1997) Ethiopia 1:50000 topographic maps South Mekelle and Adigudom, Sheets.
Eweg H, Van Lammeren R (1996) The application of a Geographical InformationSystem at the rehabilitation of degraded and degrading areas. A case study inthe Highlands of Tigray, Ethiopia. Centre for Geographical InformationProcessing, Agricultural University Wageningen, 79p.
Fernandez C, Wu JQ, McCool DK, Stöckle CO (2003) Estimating water erosion andsediment yield with GIS, RUSLE, and SEDD. Journal of Soil and Water Conservation 58(3): 128-136
Ferro V, Minacapilli M(1995) Sediment delivery processes at basin scale. Hydrol. Sci. J. 40: 703-717.
Fero V, Porto P, and Tusa G (1998) Testing a distributed approach for modellingsediment delivery. Hydrological Sciences Journal 43(3): 425-442.
Ferro V, Porto P(2000) Sediment delivery distributed (SEDD) model. Journal of Hydrologic Engineering 5(4): 411-422.
Ferro V, Di Stefano C, Minacapilli M(2003) Calibrating the SEDD model for Sicilian ungauged basins. In: Boer D de, Froehlich W, Mizuyama T, Pietroniro A (eds) Erosion prediction in ungauged basins: integrating methods and techniques. Proceedings of an international symposium, Sapporo, Japan, 8-9 July 2003 pp. 151-161.
Haan CT, Barfield BJ and Hayes JC (1994) Design Hydrology and Sedimentology forSmall Catchments. Academic Press. 588pp.
Haregeweyn N, Poesen J, Nyssen J, De Wit J, Haile M, Govers G, Deckers S(2005) Reservoirs in Tigray (northern Ethiopia): characteristics and sediment deposition problems. Land Degrad. Dev. 17,211–30
Hurni H (1985) Erosion - Productivity - Conservation Systems in Ethiopia. Proceedings 4th International Conference on Soil Conservation, Maracay, Venezuela: 654- 674
Hutchinson MF (1989) A new procedure for gridding elevation and stream line datawith automatic removal of spurious pits. Journal of Hydrol. 106, 211-232.
Jain MK, Kothyari UC (2000) Estimation of soil erosion and sediment yield using GIS.Hydrological Sciences Journal45 (5): 771-786.
Lane LJ, Renard KG, Foster GR, and Laflen JM (1997) Development and application of modern soil erosion prediction technology: The USDA experience. Eurasian soil science. Volume 30, No. 5: 531-540.
Loureiro NS and Couthino MA(2001) A new procedure to estimate the RUSLE EI30 index, based on monthly rainfall data applied to the Algarve region, Portugal. Journal of Hydrology 12-18, 12-18.
Liu BY, Nearing MA, Shi PJ, and Jia ZW (2000) Slope Length Effects on Soil Loss for Steep Slope. Published in Soil Sci. Soc. Am. J. 64: 1759-1763.
Machado MJ, Perez-Gonzalez A, Benito G (1996a) Geomorphology and soils. In:Feoli E (ed) Rehabilitation of degrading and degraded areas of Tigray,northern Ethiopia. Department of Biology, University of Trieste, Trieste, pp.65-106.
Machado MJ, Perez-Gonzalez A, Benito G (1996b) Assessment of soil erosionusing a predictive model. In: Feoli E (ed) Rehabilitation of degrading anddegraded areas of Tigray, northern Ethiopia. Department of Biology,University of Trieste, Trieste, pp. 237-248