Estimation of C factor for soil erosion modeling using ...

Ozean Journal of Applied Sciences 3(1), 2010 Ozean Journal of Applied Sciences 3(1), 2010 ISSN 1943-2429 © 2010 Ozean Publication

Estimation of C factor for soil erosion modeling using NDVI in Buyukcekmece watershed Ahmet Karaburun Fatih University, Department of Geography, Buyukcekmece, Istanbul, 34500, Turkey E-mail address for correspondence: [email protected]

__________________________________________________________________________________________ Abstract: In order to take measures in controlling soil erosion it is required to estimate soil loss over area of interest. Soil loss due to soil erosion can be estimated using predictive models such as Universal Soil Loss Equation (USLE) and Revised Universal Soil Loss Equation (RUSLE). The accuracy of these models depends on parameters that are used in equations. One of the most important parameters in equations used in both of models is C factor that represents effects of vegetation and other land covers. Estimating land cover by interpretation of remote sensing imagery involves Normalized Difference Vegetation Index (NDVI), an indicator that shows vegetation cover. The aim of this study is estimate C factor values for Buyukcekmece watershed using NDVI derived from 2007 Landsat 5 TM Image. The final C factor map was generated using the regression equation in Spatial Analyst tool of ArcGIS 9.3 software. It is found that north part of watershed has higher C factor values and almost 60% of watershed area has C factor classes between 0.2 and 0.4 Keywords: Erosion, RUSLE, USLE, C factor, Landsat, NDVI,

___________________________________________________________________________

INTRODUCTION Sediment yield studies play key role for various soil and water conservation planning processes including reservoir sedimentation analysis, studies on river morphology changes and river bed siltation, and agricultural project planning. Erosion process result in soil loss from a watershed and it is difficult to estimate soil loss as it arises from a complex interaction of various hydro-geological processes (Singh et al., 2008). Estimating the soil loss risk and its spatial distribution are the one of the key factors for successful erosion assessment. Thus it can be possible to develop and implement policies to reduce the effect of soil loss under varied geographical conditions (Colombo et al., 2005). The accuracy of estimating soil risk depends on model and its factors. Researchers have developed many predictive models that estimate soil loss and identify areas where conservation measures will have the greatest impact on reducing soil loss for soil erosion assessments (Angima et al., 2003). Those models can be classified into three main categories as empirical, conceptual and physical based models (Merrit et al.,2003). USLE and its modifications are the examples of empirical models and ANSWER, CREAMS, and MODANSW are the samples of conceptual models. Examples for the first two groups comprise the empirical USLE and its modifications, and some of the more comprehensive models like ANSWERS, CREAMS, and MODANSW. ANSWERS and CREAMS are basically conceptual and eventbased. European Soil Erosion Models, EUROSEM/KINEROS, EUROSEM/MIKE SHE and SHESED-UK are the physically-based models that have been developed at catchment or small subbasin scales (Fistikoglu and Harmancioglu, 2002). Universal Soil Loss Equation (USLE) was designed to predict longtime average soil losses in runoff from specific field areas in specified cropping and management systems. The USLE (Wischmeier and Smith, 1978) estimates the average annual soil loss from: A = R.K.LS.C.P

77

Ozean Journal of Applied Sciences 3(1), 2010

where A is the estimated soil loss per year R is the runoff factor, K is the soil erodibility factor, LS is the slope length and steepness factor, C is the cover and management factor and P is the support practice factor (Wischmeier and Smith, 1978). The R factor expresses the erosivity occurring from rainfall and runoff at a particular location. An increase in the intensity and amount of rainfall results in an increase in the value of R. The K factor expresses inherent erodibility of the soil or surface material. The value of "K" is defined as a function of the particle-size distribution, organic-matter content, structure, and permeability of the soil or surface material. The LS factor expresses the effect of topography, specifically hillslope length and steepness, on soil erosion. An increase in hillslope length and steepness results in an increase in the LS factor. The C covermanagement factor is used to express the effect of plants and soil cover. Plants can reduce the runoff velocity and protect surface pores. The C-factor measures the combined effect of all interrelated cover and management variables, and it is the factor that is most readily changed by human activities. The P factor is the support practice factor. It expresses the effects of supporting conservation practices, such as contouring, buffer strips of close-growing vegetation, and terracing on soil loss at a particular site. A good conservation practice will result in reduced runoff volume, velocity and less soil erosion. The USLE concept has more recently been modified and adapted by a large number of researchers by including additional data and incorporating research results. Revised Universal Soil Loss Equation (RUSLE) was developed by integrating several recent techniques and additional data that improves the accuracy of factors of USLE model (Renard and Freimund, 1994; Renard et al. 1997; Yoder and Lown, 1995). Thus RUSLE was extended to include forest, rangelands and disturbed areas compared to USLE. The Revised Soil Loss Equation (RUSLE) followed the same formula as the USLE, but it has a subfactor for evaluating the cover-management factor, a new equation for slope length and steepness, and new conservation practice values. It is also applicable to non-agricultural conditions such as construction sites. The RUSLE model is widely used as a predictive model for estimating soil erosion potential and effects of different management practices for over 40 years (Renard et al., 1997). One of the most important parameters in USLE and RUSLE equations is the cover management factor (C) that represents effects of vegetation and other land covers. The C factor reflects the effect of cropping and management practices on the soil erosion rate. The C factor indicates how conservation plans will affect the average annual soil loss and how that soil-loss potential will be distributed in time during construction activities, crop rotations, or other management schemes (Van der Knijff et al., 2000). Vegetation cover protects the soil by dissipating the raindrop energy before reaching the soil surface. As such, soil erosion can be effectively limited with proper management of vegetation, plant residue, and tillage (Lee, 2004). In both of USLE and RUSLE, the C factor is computed using empirical equations that contain field measurements of ground cover. (Wischmeier and Smith, 1978; Renard et al., 1997). Since the satellite image data provide up to date information on land cover, the use of satellite images in the preparation of land cover maps is widely applied in natural resource surveys (Deng et al., 2008; Serra et al.,2008; Yuan,2008). The traditional method for spatial estimation of C factor is assigning values to land cover classes using classified remotely sensed images of study areas. At the end of supervised or unsupervised classification, land cover classes are derived from image for study area and then C factors that are obtained from USLE/RUSLE guide tables or computed using field observation for each land cover classes are assigned to each pixel in land cover class (Karaburun, 2009; Efe et al., 2008; Morgan, 1995; Folly et al., 1996; Juergens and Fander, 1993). Since all pixels in a vegetation class have the same C factor value, those pixels can not represent variation of this vegetation class over the study area (Wang et al., 2002). Researchers developed many methods to estimate C factor using NDVI for soil loss assessment with USLE/RUSLE (De Jong 1994; De Jong et al., 1999; De Jong and Riezebos, 1997; Wang et al., 2002; Lin et al. 2002). These methods employ regression model to make correlation analysis between C factor values measured in field or obtained from guide tables and NDVI values derived from remotely sensed images. The unknown C factor values of land cover classes can be estimated using equation obtained from linear regression analyses. The aim of this study is to estimate C factor values of land cover classes using NDVI values by regression analysis for erosion modeling in Buyukcekmece Watershed. Study Area Buyukcekmece watershed is located in the west of Istanbul and adjacent to the Marmara Sea (Fig.1) and it contains most important water sources and a dam provides drinking water for Istanbul. The Buyukcekmece watershed is one of the largest watersheds in Istanbul having an area of about 63000 ha and located 50 kilometers west of the center of Istanbul. Agriculture is one of the most dominant land use patterns in Buyukcekmece watershed, occupying about 42000 hectares of land which makes around 67 % of the total surface of watershed. Forest area of watershed is located in the west side and covers about 8000 hectares. The upper side of watershed is forest and receives annual average rainfall about 750 mm with annual average temperature 170C. The climate of watershed is influenced by Mediterranean climate and Black Sea climate.

78

Ozean Journal of Applied Sciences 3(1), 2010

Fig.1 Study area

METHODOLOGY

Normalized Difference Vegetation Index (NDVI) Remote sensing techniques are employed for monitoring and mapping condition of ecosystems of any part of earth. Vegetation cover is the one of most important biophysical indicator to soil erosion. Vegetation cover can be estimated using vegetation indices derived from satellite images. Vegetation indices allow us to delineate the distribution of vegetation and soil based on the characteristic reflectance patterns of green vegetation. The Normalized Difference Vegetation Index (NDVI), one of the vegetation indices, measures the amount of green vegetation. The spectral reflectance difference between Near Infrared (NIR) and red is used to calculate NDVI. The formula can be expressed as (Jensen, 2000); NDVI = (NIR – red) / (NIR + red) The NDVI has been used widely in remote sensing studies since its development (Jensen, 2005). NDVI values range from -1.0 to 1.0, where higher values are for green vegetation and low values for other common surface materials. Bare soil is represented with NDVI values which are closest to 0 and water bodies are represented with negative NDVI values (Lillesand et al., 2004: Jasinski, 1990; Sader and Winne, 1992). More than 20 vegetation indices have been proposed and used at present. Since NDVI provides useful information for detecting and interpreting vegetation land cover it has been widely used in remote sensing studies (Gao, 1996: Myneni and Asrar, 1994; Sesnie et al., 2008). A time-series of NDVI were derived from Landsat 5 TM images acquired on April, May, June and August 2007. An average NDVI of watershed area was calculated using those NDVI images (Fig.2).

79

Ozean Journal of Applied Sciences 3(1), 2010

Fig.2 Average NDVI map of Buyukcekmece watershed C Factor Estimation Soil loss is very sensitive to vegetation cover with slope steepness and length factor (Renard and Ferreira 1993; Benkobi et al., 1994; Biesemans et al., 2000). Vegetation cover protects the soil by dissipating the raindrop energy before reaching soil surface. The value of C depends on vegetation type, stage of growth and cover percentage (Gitas et al., 2009). The C factor values vary between 0 and 1 based on types of land covers. Since NDVI values have correlation with C factor (De Jong, 1994; Tweddales et al., 2000; De Jong et al., 1999; De Jong and Riezebos, 1997). Many researchers used regression analysis to estimate C factor values for land cover classes in erosion assessment (Lin et al., 2002; 2006; Symeonakis and Drake, 2004; Van der Knijff et al., 2002). The goal of regression analysis is to estimate the unknown values of dependent variable based upon values of an independent variable using a mathematical model. The linear or non-linear regression equations are constructed using correlation analysis between NDVI values obtained from remotely sensed image and corresponding C factor values obtained from USLE/RUSLE guide tables or computed using field observation.

Fig.3 Workflow of C factor estimating using NDVI The study assumes that there exists a linear correlation between NDVI and C factor and uses bare soil and forest NDVI values as reference values (Fig.3). Sample NDVI values were collected for bare soil and forest land cover classes from average NDVI image. Since C factor values range from 0 for well-protected soil to 1 for bare soil (Pierce et al.,1986; Vicenta et al., 2007) the C factor values for bare soil and forest land cover were set to 1 and 0, respectively in the regression analysis. Fig.4 shows the graph of the regression equation. The line in Fig.4 is

80

Ozean Journal of Applied Sciences 3(1), 2010

the regression line that describes relationship between C and NDVI values and R shows the correlation coefficient of regression analysis. The regression equation was found as; C factor = 1.02 – 1.21 * NDVI The final C factor map was generated using the regression equation in Spatial Analyst tool of ArcGIS 9.3 software. The graphs of regression analysis and C factor are given in Fig. 4 and Fig.5 respectively.

RESULTS As can be seen from Table 1, Buyukcekmece experienced lowest mean NDVI values in August. Since watershed consists of agricultural areas the mean NDVI values of April, May and June are close to each other. Mean NVDI values of August are lowest because there is no vegetation on agricultural areas.

Table 1 NDVI values of Landsat images Image Date

Max NDVI

Min NDVI

Mean NDVI

April 2007 May 2007 June 2007 August 2007

1 1 1 0,66

-1 -1 -1 -0,43

0,395 0,40 0,31 0,09

Fig.4 Linear regression of NDVI and C factor values As seen from Fig.5 the north part of watershed is represented by 0-0.1 C factor class since it is occupied by forest. Water bodies like Buyukcekmece Lake are represented by 0.9-1.0 C factor class. The agricultural areas of watershed are represented by C factor classes from 0.2 to 0.4

81

Ozean Journal of Applied Sciences 3(1), 2010

Fig.5 C factor map of Buyukcekmece Watershed The estimated C factor values through regression equation were divided into ten categorical classes and those classes vary between 0-0.1 and 1.0. The pixel numbers of those classes are shown in Fig.6. Since agricultural areas cover almost 60% of watershed area the C factor classes 0.2-0.3 and 0.3-0.4 contain the highest pixel number respectively. As shown in Fig.6, 0.2-0.3 class has about 28% of total pixels while 0.3-0.4 has about 15% of total pixels. The 0.9-1.0 class has the lowest pixel number.

Fig.6 Pixel distribution of the C factor map based on NDVI

CONCLUSION An attempt has been made to estimate C factor values of land cover classes using NDVI values for modeling soil erosion using ArcGIS 9.3 software. A regression analysis was performed between NDVI and C factor using an assumption. C factor values were assigned to pixels of NDVI image through regression equation. Based on an assumption, the C factor map of Buyukcekmece watershed was produced to use in soil erosion methods such as USLE and RUSLE based on an assumption that NDVI and C factor values are correlated with each other. The results revealed that large parts of areas were assigned to C factor classes that vary between 0.2 and 0.4. It should be noted that C factor values can be precisely estimated using empirical equations that contain field measurements of land cover classes. However, this study shows that NDVI based regression method offers an optimal method to estimate C factor values of land cover classes of large areas in a short time.

82

Ozean Journal of Applied Sciences 3(1), 2010

REFERENCES Angima, S.D., Stott, D.E., O’Neill, M.K., Ong, C.K. and Weesies, G.A. (2003). ―Soil erosion prediction using RUSLE for central Kenyan highland conditions.‖ Agriculture, Ecosystems and Environment 97, 295– 308 Biesemans J., Meirvenne, M. V., and Gabriels, D. (2000). ―Extending the RUSLE with the Monte Carlo error propagation technique to predict long-term average oV-site sediment accumulation.‖ Journal of Soil and Water Conservation, 55, 35–42. Benkobi, L., Trlica, M. J., and Smith, J. L. (1994). ―Evaluation of a re. ned surface cover subfactor for use in RUSLE.‖ Journal of Range Management, 47, 74–78 Colombo, S., Hanley, N. and Calatrava, J. (2005). ―Designing policy for reducing the off-farm effect of soil erosion using choice experiments.‖ Journal of Agricultural Economics, 56, 81–95. De Jong, S. M. (1994). ―Application of Reflective Remote Sensing for Land Degradation Studies in a Mediterranean Environment‖ (Utrecht: Netherlands Geographical Studies, University of Utrecht) De Jong, S.M., Paracchini, M.L., Bertolo, F., Folving, S., Megier, J., De Roo, A.P.J. (1999). ―Regional assessment of soil erosion using the distributed model SEMMED and remotely sensed data.‖ Catena 37 (3–4), 291–308. De Jong, S.M., Riezebos, H.T.(1997). ―SEMMED: a distributed approach to soil erosion modelling. In: Spiteri, A. (Ed.), RemoteSensing '96: Integrated Applications for Risk Assessment and Disaster Prevention for the Mediterranean. Balkema, Rotterdam, pp. 199–204. Deng, J.S., Wang, K., Deng, Y.H., and Qi, G.J. (2008). ―PCA-based land-use change detection and analysis using multitemporal and multisensor satellite data.‖ International Journal of Remote Sensing, 29(16), 4823-4838. Efe, R., Ekinci, D., Curebel, I. (2008). ―Erosion Analysis of Fındıklı Creek Catchment (NW of Turkey) Using GIS Based on RUSLE (3d) Method.‖ Fresenius Environmental Bulletin, 17 (5), 586-576 Fistikoglu, O. and Harmancioglu, N.B. (2002). Integration of GIS with USLE in Assessment of Soil Erosion. Water Resources Management 16, 447–467, Folly, A., Bronsveld, M. C., and Clavaux, M. (1996). ―A knowledge-based approach for C-factor mapping in Spain using Landsat TM and GIS.‖ International Journal of Remote Sensing, 17 (12), 2401-2415. Gao, B. C. (1996). ―NDWI—a normalized difference water index for remote sensing of vegetation liquid water from space.‖ Remote Sensing of Environment, 58, 257–266. Gitas, I.Z., Douros, K., Minakou1, C., Silleos, G.N., and Karydas, C.G. (2009). ―Multi-Temporal Soil Erosion Risk Assessment In N. Chalkidiki Using A Modified Usle Raster Model.‖ Earsel Eproceedings 8, 1/2009 Pp.40-52 Jasinski, M. F. (1990). ―Sensitivity of the normalized difference vegetation index to subpixel canopy cover, soil albedo, and pixel scale.‖ Remote Sensing of Environment, 32, 169– 187. Jensen, J.R. (2000). ―Remote Sensing of the Environment: An Earth Resource Perspective‖ Prentice Hall, New Jersey. Jensen, J.R. (2005). ―Introductory Digital Image Processing: A Remote Sensing Perspective‖ Pearson Prentice Hall, New Jersey. Juergens, C. and Fander, M. (1993). ―Soil erosion assessment by means of Landsat TM and ancillary digital data in relation to water quality.‖ Soil Technology, 6, 215-223 Karaburun, A., 2009, Estimating Potential Erosion Risks in Corlu Using the GIS-Based RUSLE Method, Fresenius Environmental Bulletin, 18(9a), 1692-1700.

83

Ozean Journal of Applied Sciences 3(1), 2010

Lee, S. (2004). ―Soil erosion assessment and its verification using the universal soil loss equation and geographic information system: A case study at Boun, Korea.‖ Environmental Geology, 45, 457–465 Lillesand, T.M, Kiefer, R.W. and Chipman, J.W. (2004). ―Remote Sensing and Image Interpretation.‖ 5th ed. New York: John Wiley & Sons, Inc. Lin, C.-Y., Lin, W.-T., Chou, W.-C. (2002). ―Soil erosion prediction and sediment yield estimation: the Taiwan experience.‖ Soil and Tillage Research 68 (2), 143–152 Lin, W.-T., Lin, C.-Y., Chou, W.-C. (2006). ―Assessment of vegetation recovery and soil erosion at landslides caused by a catastrophic earthquake: a case study in Central Taiwan.‖ Ecological Engineering 28 (1), 79–89. Merritt, W. S., Letcher, R. A., and Jakeman, A. J. (2003). A review of erosion and sediment transport models. Environmental Modelling & Software, 18, 761–799. Myneni, R. B. and Asrar, G. (1994). ―Atmospheric effects and spectral vegetation indices.‖ Remote Sensing of Environment, 47, 390– 402. Morgan, R.P.C. (1995). ―Soil Erosion and Conservation.‖ Addison-Wesley Longman, Edinburgh Pierce, F.J., Larson, W.E., Dowdy, R.H., 1986. Soil Conservation: An Assessment of the National Resources Inventory, vol. 2. National Academy Press, Washington, D.C. Renard K.G and Ferreira V.A. (1993). ―RUSLE model description and database sensitivity.‖ Journal of Environmental Quality 22(3), 458–466. Renard K.G. and Freimund, J.R. (1994). ―Using monthly precipitation data to estimate the R factor in the revised USLE.‖ J Hydrol 157, 287–306 Renard, K.G., Foster, G.R., Weesies, G.A. and McCool, D.K. (1997). ―Predicting soil erosion by water: a guide to conservation planning with the Revised Universal Soil Loss Equation (RUSLE)‖, Agric. Handb., US Department of Agriculture, Washington, DC, vol.703, Sader, S.A. and Winne, J.C. (1992). ―RGB-NDVI colour composites for visualizing forest change dynamics.‖ International Journal of Remote Sensing, 13, 3055-3067. Serra, P., Pons, X., Sauri, D. (2008) ―Land-cover and land-use change in a Mediterranean landscape: A spatial analysis of driving forces integrating biophysical and human factors.‖ Applied Geography, 28, 189-209 Sesnie, S. E., Gessler, P.E. and Finegan, B. (2008). ―Integrating Landsat TM and SRTM-DEM derived variables with decision trees for habitat classification and change detection in complex neotropical environments.‖ Remote Sensing of Environment, 112, 2145-2159. Singh PK, Bhunya PK, Mishra SK, Chaube UC (2008) A sediment graph model based on SCS-CN method. J Hydrol 349, 244-255. Symeonakis, E., Drake, N. (2004).‖ Monitoring desertification and land degradation over sub-Saharan Africa.‖ International Journal of Remote Sensing 25 (3), 573–592 Tweddales, S.C., Eschlaeger, C.R., Seybold, W.F. (2000). ―An Improved Method for Spatial Extrapolation of Vegetative Cover Estimates (USLE/RUSLE C factor) using LCTA and Remotely Sensed Imagery.‖ USAEC report No. SFIM-AEC-EQ-TR-200011, ERDC/CERL TR-00-7, US Army of Engineer Research and Development Center, CERL, Champaign, Illinois Van der Knijff, J.M., Jones, R.J.A., Montanarella, L. (2000). Soil Erosion risk Assessment in Europe, European Commission, European Soil Bureau, http://eusoils.jrc.ec.europa.eu/ESDB_Archive/pesera/pesera_cd/pdf/ereurnew2.pdf (June. 5, 2009).

84

Ozean Journal of Applied Sciences 3(1), 2010 Wang, G.,Wente, S., Gertner, G.Z., Anderson, A. (2002). ―Improvement in mapping vegetation cover factor for the universal soil loss equation by geostatistical methods with Landsat Thematic Mapper images.‖ International Journal of Remote Sensing 23 (18), 3649–3667. Wischmeier, W. H., and D.D. Smith. (1978). ―Predicting rainfall-erosion losses - a guide to conservation planning.‖ AH-537. U.S. Dept. Agr., Washington, D.C., Vicente, M.L., Navas, A., and Machin, J. (2007). ―Identifying erosive periods by using RUSLE factors in mountain fields of the Central Spanish Pyrenees.‖ Hydrol. Earth Syst. Sci. Discuss., (4), 2111-2142 Yoder D. and Lown, J. (1995). ―The Future of RUSLE: inside the new revised universal soil loss equation.‖ J Soil Water Conservation 50(5), 484–489 Yuan, F. (2008). ―Land-cover change and environmental impact analysis in the Greater Mankato area of Minnesota using remote sensing and GIS modeling.‖ International Journal of Remote Sensing, 29(4), 1169-1184.

85