AN APPROACH OF GIS BASED ASSESSMENT OF SOIL EROSION RATE ON COUNTRY LEVEL IN THE CASE OF MACEDONIA

Abstract

One of the major environmental problems in the Republic of Macedonia is accelerated soil erosion. Steep slopes of the terrain combined with soft rocks (schists on the mountains; sands and sandstones in depressions), erodible soils, semi-arid continental climate and sparse vegetation cover, give high potential for soil erosion. For that reasons, different approaches and methods are used for assessment of soil erosion intensity in the country. Aside of field analyses and hydrological measurements, estimation of average soil erosion potential and sediment yield is generally achieved with Erosion Potential Model (EPM) of Gavrilovic[5]as one of the most frequently used in the region. However, unlike the traditional approach with high subjectivity, in latest decade GIS approach of EPM is introduced, with most of the model parameters derived from digital elevation model and satellite imagery. In this paper, one such approach for the entire country is presented, after numerous evaluations of smaller catchments. The results in general are similar with Soil Erosion Map of the Republic of Macedonia, but there are some significant local differences.

Full text

Physical Geography; Cartography; Geographic Information Systems & Spatial Planing
97
AN APPROACH OF GIS BASED ASSESSMENT OF SOIL EROSION RATE ON
COUNTRY LEVEL IN THE CASE OF MACEDONIA
DOI: http://dx.doi.org/10.18509/GBP.2015.13
UDC: 007:004]:[528.8:551.311 (497.7)
Assoc. Prof. Dr. Ivica Milevski
Faculty of Sciences, Ss. Cyril and Methodius University, Republic of Macedonia
ABSTRACT
One of the major environmental problems in the Republic of Macedonia is accelerated
soil erosion. Steep slopes of the terrain combined with soft rocks (schists on the
mountains; sands and sandstones in depressions), erodible soils, semi-arid continental
climate and sparse vegetation cover, give high potential for soil erosion. For that reasons,
different approaches and methods are used for assessment of soil erosion intensity in the
country. Aside of field analyses and hydrological measurements, estimation of average
soil erosion potential and sediment yield is generally achieved with Erosion Potential
Model (EPM) of Gavrilovic [5] as one of the most frequently used in the region. However,
unlike the traditional approach with high subjectivity, in latest decade GIS approach of
EPM is introduced, with most of the model parameters derived from digital elevation
model and satellite imagery. In this paper, one such approach for the entire country is
presented, after numerous evaluations of smaller catchments. The results in general are
similar with Soil Erosion Map of the Republic of Macedonia, but there are some
significant local differences.
Keywords: GIS, soil erosion, natural hazards, EPM
INTRODUCTION
Soil erosion by water is a widespread problem throughout Europe. Erosion rate is very
sensitive to both climate and land use, therefore southern Europe and the Mediterranean
region is particularly prone to erosion because it is subject to long dry periods followed
by heavy bursts of erosive rain, falling on steep slopes with fragile soils [6]. In this regard,
Republic of Macedonia is highly exposed and devastated by soil erosion. One of the key
factors for that is uncontrolled deforestation in the past as well as inappropriate land use
and high human impact on the landscape. Some of the most devastating effects are large
areas with excess erosion, flash floods, landscape degradation and significant damages to
the local and national economy. Because of that, assessment of soil erosion intensity and
identification of endangered areas is very significant for better prevention and protection
of landscape and population [3]. For assessment of soil erosion risk, generally, three types
of approaches exist in Europe [4]: qualitative approach, quantitative approach, and model
approach, all of which vary in their characteristics and applicability. In last two decades,
already developed models and approaches are improved through the use of geospatial
databases developed using GIS technology. In practical applications, advantage usually
is given to the most known, well-evaluated and tested empirical models USLE, PESERA,
KINEROS, WEP, WEPP etc. However, there are few locally used, not so widely known
models which were shown as pretty accurate and easy to use [5]. One of these is Erosion
Potential Model (EPM) of Gavrilović (1972) which has been successfully applied and
tested in various regions and entire countries on the Balkan Peninsula and wider, showing

International Scientific Conference GEOBALCANICA 2015
98
very good results [5]. For that reasons, there was several attempts to adopt EPM to GIS-
based approach [8],[6], and in this paper is represented last update which is tested in
details.
METHODOLOGY
In Macedonia, as well as in other countries in the region, estimation of average soil
erosion potential and sediment yield is generally achieved with Erosion Potential Model
(EPM) of Gavrilović (1972). The equation is Wy = T ∙ H ∙ 3.14 ∙ sqrtZ3 ∙ f, where: W is
average annual soil erosion in m3; T is temperature coefficient in form: T = (0.1 ∙ t +
0.1)0.5, where t is mean annual air temperature; H is mean annual precipitation in mm; Z
is erosion coefficient ranging from 0.1 to 1.5 and over; and f is study area in km2. Among
these factors, coefficient Z has the highest importance combining: soil erodibility (Y),
land cover index (Xa), index of visible erosion processes (φ), and mean catchment slope
(J) in relation: Z = Y ∙ Xa ∙ (φ + sqrtJ0.5). Values of Z usually ranged between 0 (no
erosion) and 1.5 (excess erosion). Based on this traditional catchment oriented approach,
erosion map of the Republic of Macedonia is made [2]. However, determination of some
significant factors in the model as coefficient φ (visible erosion processes) and X*a (land
cover index) were subjective (expert-related) in nature since of visual estimation. Because
of that, in GIS approach of the model most of the parameters are derived from digital
elevation model (DEM), satellite imagery and other digital thematic maps [9]. Thus, for
the GIS-based EPM assessment of the Republic of Macedonia, set of very detailed digital
models and grids with 20 m resolution are prepared, evaluated and used.
For erodibility coefficient Y, digitalized and rasterized lithology map is prepared with
corresponding values for the rock erodibility proposed by Gavrilović [5]. In general,
values range from 0.1 (very solid rocks like marble, limestone quartzite etc.) to 2.0 (very
soft rocks like sand, sandstone, tuff etc.). However, because it is very difficult to estimate
exact erodibility, value fitting is made with double square rooting in form: Y =
sqrt(sqrt(Y1))(Fig. 1).
Land cover index X*a is prepared from CORINE Land cover model (CLC2006) with
values proposed by original model ranging from 0.1 (dense forests) to 1.0 (bare soils).
For the value of coefficient φ of visible erosion processes, instead of very subjective
estimation in traditional model, Landsat ETM+ band 3 (b3-red) or Landsat 8 (b4-red) is
used in such way that grayscale values (0-255) are divided by 255. That is because this
channel has 255 tones of gray where low values correspond to areas without visible
erosion processes, and values near 255 usually correspond to areas with excess erosion
(Fig. 1). However, high values may also represent light anthropogenic objects, uncovered
soils and rocks, deposition sites etc. For that reason, correction with slope gradient (a) in
form: φ = ((b3/255)*log(a+1)) is made, resulting in much more accurate values for
coefficient φ [11].
Slope factor J is calculated from high quality 20 m DEM as a raster layer for slope angle
in radians (a=a/57.3). Finally, GIS-calibrated coefficient Z is calculated according to the
equation: Z = sqrt(sqrt(Y))*φ*((X*a+φ)*log(a+1)+sqrt(a/57.3))
Climate parameters in the model e.g. temperature coefficient (T) and mean annual sum
of precipitations (H) are prepared with combination of real data from meteorological
stations in the Republic of Macedonia and corresponding vertical gradients. Firstly,
position and mean temperature/precipitation values (1961-1990) for 30 meteorological

Physical Geography; Cartography; Geographic Information Systems & Spatial Planing
99
stations across the country are inserted on the map as a vector points with attributes. The
next step was gridding (rasterize) of these values using Modified Quadratic Shepard
module in SAGA GIS v2. The produced grids of temperatures and precipitations are
checked well for consistency and accuracy after which altitude regression scatterplots in
regard to the 20 m DEM are calculated. With these regressions, raster grids of mean
temperature and precipitation based on vertical gradients were generated. Final models
are produced with averaging of grids of interpolated real data and vertical gradient grids
[11].
Figure 1. Digital models of erodibility coefficient "Y" and coefficient "φ".
Having available temperature and precipitation grids, as well as the erosion coefficient
grid (Z), final mean annual erosion loss (Wy) for the entire country is calculated. This
GIS-based methodology is previously tested with researches in different regions and
catchments in the Republic of Macedonia showing good results when comparing with
measured and observed data [10]. Also, because of exactly defined approach, real
comparisons of different areas or different periods are possible, without the issue of
subjectivity.
RESULTS
With GIS implementation of EPM through the SAGA GIS v.2.1 software, two most
significant models are prepared: model of erosion (risk) coefficient Z; and model of mean
annual sediment yield W. Both grid models are in 20 m resolution, corresponding to
resolution of used input layers (datasets).
GIS-based model of the erosion coefficient Z, shows values from 0 (deposition areas) to
more than 3 (areas with excess erosion), with mean value of 0.4 (low to moderate
erosion). However, there are large areas with high and very high erosion risk (values
greater than 0.8). On these areas, even modest rainfall causes high production, transport
and accumulation of eroded material. This is especially the case during heavy rains of
more than 0.5 mm/min and duration of more than 30 min [9].

International Scientific Conference GEOBALCANICA 2015
100
Figure 2. Maps of mean annual erosion production in the Republic of Macedonia according to the GIS-
based approach and traditional EPM-model (in m3/km2/y-1).
From the second output: GIS-based model of the sediment yield follows that average
annual production of sediment yields in Macedonia is very high: 691 m3/km2 or 0.69

Physical Geography; Cartography; Geographic Information Systems & Spatial Planing
101
mm/y-1. Regardless of different approaches, this value coincide to that obtained in
traditional Erosion map of the Republic of Macedonia [2], but the spatial structure of the
erosion classes is apparently different (Fig. 2). However, there is a large spatial difference
from the plains into depressions bottom to its step uncover sides and further toward the
mountain hillslopes. In dense forest mountainous areas erosion rate is usually in range of
100-300 m3/km2/y-1, while some areas with high human impact have over 2000 m3/km2/y-
1. Highest erosion rate have areas in the Bregalnica catchment, the lower Crna catchment,
the Pčinja catchment in the Tikveš Basin and many others. There actually appears severe
or excessive erosion causing numerous destructive landforms, losing valuable fertile land
and filling of river beds with a large amount of sediment material.
Table 1. Structure of areas according to the soil erosion classes with mean elevation for each class.
Type
Int. m3/km2/y
In km2
In %
Mean Elev.
Excess
>2000
1625.6
6.3
815.8
Severe
1500-2000
1351.7
5.3
815.6
Very High
1000-1500
2699.4
10.5
825.5
Medium-high
750-1000
2364.5
9.2
848.0
Medium
500-750
3807.5
14.8
881.5
Low
250-500
6048.4
23.5
931.0
Very low
50-250
7229.2
28.1
859.5
Deposition
<50
586.7
2.3
527.9
Total
25713.0
100.0
829.2
Because of more detailed analyses, erosion intensity is divided into 8 ranges (classes):
from deposition, to excess erosion one. Spatial area and mean elevation for each class is
calculated (Table 1). The results show dominance of areas with low to very low erosion
rate with total of 51.6%. They are related with forested mid-altitude mountain areas and
plains in depressions which cause their higher mean elevation (Fig. 3-left). Deposition
sites are connected with lakes, reservoirs and downstream valley parts of larger rivers,
thus their low mean elevation in respect to other classes.
Figure 3. Graph of soil erosion intensity vs. elevation (left) and vs. slopes (right).
Table 2. Altitude vs. mean erosion rate (ER) in m3/km2/y-1, calculated from the GIS-model.
Altitude
Area km2
Area in %
Mean ER
Total ER
ER in %
2000-2753
354.0
1.4
1285.0
454890
2.6
1500-2000
1872.1
7.3
806.0
1,508913
8.5
1000-1500
5718.0
22.2
575.0
3,287850
18.5
500-1000
11297.3
43.9
712.2
8,045937
45.3
250-500
4875.0
19.0
780.6
3,805425
21.4

International Scientific Conference GEOBALCANICA 2015
102
0-250
1596.3
6.2
436.2
696306
3.9
Total/Avrg.
25712.7
100.0
690.9
17,764904
100.0
According to the model, significant percentages of the country cover areas with very high
to severe and excess erosion rate with total of 22.1%. They are mostly extended between
depression bottoms and mid-altitude mountain slopes. For that reason, height range of
these erosion classes is around 400-1000 m of altitude, and toward south-slope areas (Fig.
3-rigth) where human impact is the largest. Because of the high erosion rate, the landscape
is often dissected into rills, gullies, badlands, earth pyramids etc. Here, instead of normal
river catchments, torrential watersheds with high overload of sediments during the heavy
rain occur.
CONCLUSION
In the Republic of Macedonia, large areas are under accelerated erosion caused by
intensive human impact on suitable environment during centuries. Such areas are mostly
steep, south inclined slopes of depressions and valleys, generally below 1000 m elevation,
which were most appropriate for early settling as well. Because of accelerated erosion,
the “original” landscape is very often dissected into rills, gullies, badlands, earth pyramids
etc. Moreover, there are landscapes extremely devastated by soil erosion like the
Bregalnica catchment, the lower Crna catchment, the Pčinja catchment and many others.
On the other hand, lower parts of these catchments and valleys suffer severe deposition
of eroded material with notable impact on fluvial processes. The lower sections of the
Vardar, Pčinja, Bregalnica, and Crna River are bordered by large alluvial plains with fresh
deposits, where lateral erosion, meandering and even channel accretion prevail. As a
result of excessive deposition, many downstream riverbeds are now more than 10-15 m
uplifted. Large tracts of agricultural land were abandoned with significant negative social-
economic impacts on the rural environment.
Because of the high environmental and overall impact, accurate assessment of soil erosion
intensity and identification of endangered areas is very significant issue [3]. In
Macedonia, as well as in other countries in this region, estimation of average soil erosion
potential and sediment yield is generally achieved with Erosion Potential Model (EPM)
which show better results compared to other tested models as USLE [1]. However,
determination of some significant factors in the model was subjective in nature since of
visual estimation. Because of that, in implemented GIS approach, most of the parameters
are derived from digital elevation model (DEM), satellite imagery and other digital
datasets [9]. This approach has many advantages over traditional one:
 Preparing of input layers, calculations and final modelling is much faster than with
traditional field-based approach, taking years to complete (Erosion map of
Macedonia which is in use yet was prepared during the period of almost 20 years).
 Grid-based approach is far more spatially detailed (20 m resolution in our case) and
usable than catchment based (Fig. 2).
 Entire GIS-based procedure is far less expensive, because almost all data layers
(Landsat imagery, CLC model, digital elevation model etc.) as well as SAGA GIS
software (or similar one) are free of charge.
 The subjectivity of individual expert estimations of visual erosion processes is
drastically reduced and replaced with corresponding values from corresponding
satellite imagery.

Physical Geography; Cartography; Geographic Information Systems & Spatial Planing
103
 The ability of temporal monitoring with GIS-procedure, because it is easy to change
some of input parameters (layers) with newer one, especially those based on
frequently updated free satellite imagery (coefficient φ or X*a).
 Finally, original EPM model is limited in soil loss estimation during one season,
month or one rain event. However, our detailed calculations show that it is possible
even that with introduction of daily rain value instead of average annual
precipitations (H), in form of: H=(Hy*(Hd/(Hy/6))). Here Hd is daily value of rain
and Hy is average yearly sum of precipitations in that area. This approach shows
good observation field-fitting but it is not tested enough with field measurements
which are necessary for evaluation. Even with that, usable rain-intensity scenarios
can be produced showing erosion risk areas vs. daily rain amount/intensity.
However, there are several drawbacks and uncertainties of this approach which must be
taken into account in further applications and developments.
 First of all is the problem with coefficient of rock and soil erodibility "Y". Values for
this coefficient are relative and given by the author of the original model based
generally on its estimations. This is rather problematic when numerous different
types of rocks and soils must be considered. Also, it is very difficult to include and
combine these two types of erodibility (rock and soil) in only one value, or to
determine where the greater influences of each are. Because the Republic of
Macedonia do not have high-detail soil properties map yet, only 100k digitalized
geological map is used for now. Even that map is rather old, produced mostly in
1970-ties almost without updates afterward (despite that lithology is relatively
"static" factor).
 The next is the problem with coefficient X*a referring to the land use type and pattern
closely related to Corine Land Cover data. For the extent of Macedonia, the last
available model is CLC2006 which is used in our procedure, but it is rather old and
the newer CLC2012 is expected soon. Some events like large forest fires in 2007 and
2010 make significant changes in land cover which are not represented by CLC2006.
Thus, alteration with newest satellite imagery is necessary to obtain real
representation.
 Using the red spectral band from Landsat imagery as a basis for the coefficient of
visible erosion processes φ, maybe is not the best choice in some cases. Namely, in
this spectral band the white-colored resistant rocks like marbles and limestones
correspond to high value or heavily eroded terrains which is not real. Thus, a better
combination of spectral bands may provide more accurate values. Also, it is difficult
to acquire the newest possible Landsat scenes of Macedonia (6 scenes of entire
country) for the same season, with same quality and very low cloud cover.
REFERENCES
[1] Blinkov, I. and S. Kostadinov 2010. Applicability of various erosion risk
assessment methods for engineering purposes, BALWOIS 2010 Conference -
Ohrid, Republic of Macedonia. 25 - 29 May 2010.
[2] Djordjevic M, Trendafilov A, Jelic D, Georgievski S, A. Popovski 1993. Erosion
map of the Republic of Macedonia. Memoir. Water Development Institute, Skopje.
89 p. (in Macedonian)

International Scientific Conference GEOBALCANICA 2015
104
[3] Dragicević S., D. Filipović, S. Kostadinov, R. Ristić, I. Novković, N. Zivković, G.
Andjelković, B. Abolmasov, V. Secerov, S. Djurdjić. Natural hazard Assessment
for Land-use Planning in Serbia. Int J Environ Res., 5(2), 2011a, pp. 371–380.
[4] Eckelmann, W., Baritz, R., Bialousz, S., Bielek, P., Carré, F., Houšková, B., Jones,
R.J.A., Kibblewhite, M.G., Kozak, J., Le Bas, C., Tóth, G., Tóth, T., Várallyay, G.,
Yli Halla, M., & Zupan, M. (2006). Common Criteria for Risk Area Identification
according to Soil Threats. EUR 22185 EN. Luxembourg: Office for Official
Publications of the European Communities.
[5] Gavrilović, S. 1972. Engineering of Torrents and Erosion, Journal of Construction
(Special Issue), Belgrade, Yugoslavia. (in Serbian)
[6] Globevnik L., Holjevic D., Petkovsek G. & Rubinic J. (2003). Applicability of the
Gavrilovic method in erosion calculation using spatial data manipulation. Erosion
Prediction in Ungauged Basins: Integrating Methods and Techniques (Proceedings
of symposium HS01 held during IUGG2003 at Sapporo, July 2003). IAHS Publ.
No. 279, 224-233.
[7] Grimm, M. R. Jones, L. Montanarella 2001. Soil Erosion Risk in Europe. European
Soil Bureau Institute for Environment & Sustainability JRC Ispra, pp. 1-40
[8] Milevski I. (2001). Modeling of soil erosion intensity with software tools, in the
example of Kumanovo Basin. Proceedings of II Congress of Macedonian
Geographic Society, Ohrid, 49-57. (in Macedonian)
[9] Milevski I. 2011. Factors, Forms, Assessment and Human Impact on Excess
Erosion and Deposition in Upper Bregalnica Watershed (Republic of
Macedonia). In: Human Impact on Landscape, Ed. S. Harnischmachter and D.
Loczy. Zeitschrift für Geomorphologie, Vol. 55, Suplementary issue 1, Stuttgart,
77-94.
[10] Milevski I. 2015. Digital Model of the Mean Annual Temperature and Precipitation
of the Republic of Macedonia. Geographical Reviews No. 48, Skopje. pp. 33-38.
[11] Milevski I., Dragicevic S., A. Georgievska 2013. GIS and RS-based modelling of
potential natural hazard areas in Pehchevo municipality, Republic of Macedonia.
Journal of the Geographical Institute „Jovan Cvijić” SASA, 63(3): 95-107.