Desalination 213 (2007) 263–272

Presented at the International Conference on New Water Culture of South East European Countries-AQUA 2005,21–23 October 2005, Athens, Greece.

0011-9164/06/$– See front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.desal.0000.00.000

An approach to mapping soil erosion by water with application to Albania

Spiro Grazhdania*, Spase Shumkab aAgriculture University of Tirana, Faculty of Forestry, Tirana, Albania

Tel./Fax +355 4243-812; email: sgrazhdani@yahoo.combAgriculture University of Tirana, Faculty of Agriculture, Tirana, Albania

Received 30 January 2006; revised accepted 5 March 2006

Abstract

The combination of the mountainous terrain of Albania and a seasonally wet Mediterranean climate has led tosome of the most extreme erosion in Europe. New methodology is used in this work to estimate soil erosion riskand provide a framework for soil conservation that can effectively and economically control soil loss. A one-dimensional hydrology, vegetation and erosion model (RDI/CSEP) is used to calculate accumulative soil erosion.This model provides an integrating the climatic, topography and associated vegetation components into soilerosion modeling. The forecast runoff, accumulated across the frequency distribution of storms, is used to give aclimatic potential, which is then appropriately combined with measures of topography and soil erodibility toestimate the expected rate of soil erosion at a resolution of 1 km.

Annual and monthly soil erosion maps of Albania are also developed in this research. The proposed approachfor evolution and mapping soil erosion by water uses existing soil maps, land use maps, a digital elevation model(1 km), and interpolated climate data. The erosion maps clearly show that Albania is a country where erosion ispotentially severe. The annual erosion rates are estimated at 10 t ha–1y–1 or more, especially in the south and centerpart of the country. In three areas (two in Gjirokastër, and one in Sarandë) the annual erosion rate is more than100 t ha–1 y–1. Erosion rates are highest in October, November, February and December and lowest in June and July.

Keywords: Soil erosion; Mapping; Land use; Land cover; Albania

1. Introduction

Soil erosion is the consequence of a set ofimportant processes, which are most active in theabsence of vegetation, on both cropland, and for

uncultivated arid and semi-arid environments,where water is the main limiting factor in plantgrowth [1,2]. It is known to be directly con-trolled by a number of factors climate, vegeta-tion, soil properties and topography. Each factoris itself complex and the various factors interactwith one another. *Corresponding author.

264 S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272

The high erosion rates occurring in Mediter-ranean type areas are attributed to the climaticregime and to the generally poor vegetationcover [3]. Although Albania is a small countryrainfall is highly variable in time and space. Therainfall regime is of the Mediterranean type witha dry summer season and a rainy season inautumn and winter. Kosmas et al. [4] had foundhigh rates of runoff and sediment loss underrainfall higher than 280 mm per year for Medi-terranean conditions. Rondelli et al. [5] has pre-viously identified the high value of the erosivityindex for Albania (annual mean value 269.7),which presents a severe risk in generating runoffand sediment loss.

Soil erosion and land degradation constitutea severe problem in Albania. Due to poor man-agement in Albania over the last 50 years,unsuitable land has been brought into cultivationand exposed to severe overgrazing. Climatologi-cally, Albania provides some of the mostextreme conditions for erosion in Europe, due tothe combination of mountainous country andsevere land mis-management and may lead to anirreversible loss soil in some areas. Soil lossesexceed 100 t ha−1y−1 [6] in some places and thisrate very far outstrips any possible recovery bysoil weathering. As well as on-site soil loss thereare substantial off-site impacts through pollutionand siltation of reservoirs and coastal areas. Theaverage annual total nitrogen, potassium and phos-phorous losses of 4.9; 28.64; and 3.56 kg ha−1,respectively, are estimated for experimental siteslocated in southeastern Albania [7]. The prob-lems found in much at the arable land ofAlbania includes a decline in soil organic matterand soil structure, an increase in crust formationconsequent on a reduction in infiltration, anincrease in runoff, and a reduction in crop emer-gence and crop yield. Little is known about theextent of these processes and their variationwith landform and land use soil type and climatein Albania.

The regional degradation index (RDI) withclimatic soil erosion potential (CSEP) concept[1,2,8–12] is also used to calculate accumulativesoil erosion in this work. A 1 km DEM (digitalelevation model) is used in conjunction with theAlbanian soil types and land use maps. A model(RDI/CSEP) is proposed in this study for eleva-tion of soil erosion. The proposed approachconsists of taking a very simple and conserva-tive erosion model, and showing that it can bedisaggregated into components that dependrespectively on climate, vegetation, topographyand soil factors. These factors can then beassessed separately using available data sources,and used to make regional forecasts. Because thecomponents are explicit, the impact of changesin land use or climate can also be clearly identi-fied, so that sensitivity to changed conditionscan be estimated.

This study is the first attempt to integrateclimate, topography, soil type and associatedvegetation into a soil erosion model, and to pro-duce soil erosion by water maps for Albania.

2. Material and methods

The rate of sediment transport is estimated asa mean soil loss in t/ha, obtained as a product ofterms, which are primarily dependent on soil,climate/vegetation and topography. This approachoriginates from early work by the USDA [13],leading to expressions of the general form:

(1)

where k is the soil erodibility; q is overlandflow discharge per unit width; is local slopegradient and m, n are empirical exponents in therange 0–2.

The particular form proposed for rill erosion is

(2)

S kqm n= Λ

S k q= −( )2Λ Θ

S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272 265

where Θ is a flow power threshold. The dis-charge per unit width, q, is estimated as

q = j x (3)

where j is the runoff (per unit area) and x is thedistance downslope (from the divide). The termqΛ = jxΛ then explicitly breaks down into a cli-matically driven term j, and a topographic termxΛ. The climatic term is summed over the fre-quency distribution of storm rainfalls and thisapproach provides a rationale for combining theeffects of topography, soils and climate into asingle integrated erosion forecast.

High frequency rainfall events are clearlycrucial for generating overland flow. The sim-plest effective tool for estimating runoff is thenotion of a threshold storm size. Beneath thethreshold there is little or no runoff. Above it or ahigh proportion of the additional rainfall gener-ates overland flow. The average annual overlandflow runoff (per unit area) is given by

(4)

where the summation is made over all storms r,which exceeds the runoff threshold h and p isthe proportion of runoff above the threshold.The runoff is estimated by integration as a math-ematically defined distribution function. The dis-tribution of storm (or daily) rainfalls is expressedon an exponential form. If the density of dayswith rainfall >r is estimated as:

N (r) = N0 exp (–r/r0) (5)

Then J = pR exp (–h/r0) (6)

where N(r) is the number of days with rainfall>r; N0 is the number of rain days; r0 isthe mean rain per rain day and R is the totalrainfall = r0N0.

In practice a sum of two exponential terms[10] is more appropriate in most cases, and the

parameters are calculated for each month sepa-rately. Summation over storms is achieved byfitting the distribution of daily rain amounts tothe sum of two exponential distributions, usingthe monthly values for number of rain days,mean rain-day and its standard deviation. Thusthe cumulative frequency of a daily rainfall inexcess of r is

(7)

where μ is a fraction between 0 and 1, and r1, r2

are rainfall intensity parameters. The mean and variance s 2 of this distribution are

(8)

It has been found convenient in practice tosolve these equations iteratively, assuming afixed ratio r1:r2 = 5:1 [9].

In the expression for sediment transportused above (Eq. (2)), sediment transport isrelated to (discharge)2. Integration overthe frequency distribution is used for terms

. The effect

of the threshold term makes additional changesto the integrated sum, leading to terms of theform:

(9)

for each of the two exponential distributions. Lopez-Bermudez et al. [14] and Kirkby et al.

[2] have confirmed that vegetation exerts anextremely strong effect on runoff and erosion.Vegetation and land use are clearly importantfactors controlling the intensity and frequencyof overland flow and surface wash erosion [15,16].

J p r hr h

= −>∑ ( )

N r Nr

r

r r

r

r( ) exp

1exp0

1 1 2 2

= −⎛⎝⎜

⎞⎠⎟

+ − −⎛⎝⎜

⎞⎠⎟

⎡

⎣⎢

⎤

⎦⎥

μ μ

r

r r r

r r r r

= + −

= − − − + −

μ μ

μ μ μ μ μ1 2

212

1 2 22 2

(1 )

(2 ) 2 (1 ) (1 )σ

p r h p R r h rr h

2 2 2( ) 2 exp ( / )0 0− = −>∑

2 exp( / )exp20 0

0

p R r h rr H

− −⎛⎝⎜

⎞⎠⎟

Θ

266 S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272

Vegetation is taken from the map of land use ofyear 1999, which has been digitized at School ofGeography (University of Leeds, UK) from non-digitized form of this map provided by Ministryof Agriculture and Food.

It is concluded that vegetation and associatedsoil properties, in particular soil organic mattercontent, have a major impact on the runoffthreshold h, which takes values about 10 mmfor bare soil up to 100 mm or more for forestedareas. Vegetation acts on several ways, whichmay be dominant under different conditions, firstby protecting the soil from rainsplash and crusting,second by intercepting rainfall, which is lost toevaporation and third by building up organic mat-ter in the soil which greatly enhances the shorttime dynamic storage and release of soil moisture.The combined effect of these processes is toincrease the runoff threshold h.

The runoff threshold is calculated by equa-tion [17]:

(10)

where h is runoff threshold; α is the proportionof the surface covered by vegetation; hs is thestorage for the un-crusted soil; hR is the storageassociated with surface roughness; hI is theinterception storage on vegetation, and hc isthe storage for a crusted surface. For bare soilsthis is simply a crusted surface storage term hc.

The value of hc for each relative crusting classis given in Table 1. For soil under plant coverthe threshold h is the sum of three terms. Theun-crusted soil surface storage term hs (Table 2)is derived from the soil texture class, and a termis added for depression storage within surfaceroughness hR (Table 3), which depends onwhether the land use classification indicatestillage [17].

These terms are fixed through out the year.The actual threshold varies throughout the yearreflecting the change in the percentage of plantcover. The annual plant cover cycle throughoutthe year is estimated according to the land usetype. The land use classification of Albania doesnot correspond to the CORINE classification soa new set of monthly reclassification tables hadto be developed (Table 4). For each month theappropriate proportions of uncovered and coveredthreshold are mixed. The actual rainfall thresh-old was modified by the canopy interception hi

h h h h hs R I c= + + + −α α( ) (1 )

Table 1Storage for a crusted surface

Relative crusting classes

Crust storage hc (mm)

Level of confidence

RMS error (%)

None n/a Weak 12 Weak ±80 Moderate 6 Moderate ±50 Strong 2 Strong ±50

Table 2Un-crusted soil water retention

Soil texture class Water retention at saturation (mm)

Water retention (mm) at field capacity (50 cm)

Soil storage hs (mm)

Course 403 294 109 Medium 439 379 60 Medium fine 430 406 24 Fine 520 472 48 Very fine 614 567 47 Organic 766 708 58

S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272 267

(Table 5) for each land use type, then the runoffthreshold was applied and the square wasaccumulated.

Vegetation also resists erosion by adding tosurface roughness, which reduces overland flowvelocity, and binds the soil together with shallowroot mats, particularly in grasses. This secondgroup of processes tends to increase the erosionthreshold Θ, which determines the minimum flowpower for erosion to occur and the soil erodibilityk. These two parameters are distinguished fromthe hydraulic properties of the soil, which aresubsumed within the runoff threshold. Soil erod-ibility is more variable parameter, and appearsto take value in the range 10–5–10–2 when mea-sured in t/ha eroded for 1 mm of runoff on unitgradient. This parameter is thought to depend

most strongly on soil properties, with a maximumfor silt and sand textures, leading to approximatevalues for each soil textural class. The erodibilityfactor is reclassified from the soil material type,which is divided into weak, medium and strongclasses (Table 6).

Relief is estimated as the standard deviationof relief at each point in DEM, based on all pointswithin a 2 km radius [2,8,9]. A method of inte-gration of topographic data was developed [1] toincorporate the effects of relative relief into theCSEP. The incorporation of the topography takesthe CSEP stage further towards the creation ofthe integrated RDI, which combines soil, bio-climatic and topographic terms.

Combining the terms described above, anestimate is obtained for the average rate of soilerosion, averaged over an area. This expressionis grouped into three sets of terms [2,9]:

(11)

Table 3Surface roughness storage

Land cover Initial surface storage hR (mm)

Reduction per month (%)

Arable 10 50Other 5 0

Table 4Vegetation cover

Land cover type Monthly cover (%)

J F M A M J J A S O N D

Arable land 1 8 12 25 46 88 96 84 50 30 5 6 7 Arable land 2 7 11 24 45 87 95 95 51 32 6 6 7 Arable land 3 10 15 30 54 98 99 79 40 24 7 7 9 Arable land 4 9 14 28 52 95 98 82 44 27 7 7 9 Arable land 5 8 12 25 47 96 95 84 49 20 5 6 7 Arable land 6 8 12 26 48 90 96 85 50 31 6 7 8 Arable land 7 10 26 31 56 100 100 77 47 24 6 7 10 Pasture 60 70 80 90 100 100 100 90 80 80 70 60 Vineyards and tree crops 30 30 30 40 50 60 60 60 60 40 30 30 Forest 100 100 100 100 100 100 100 100 100 100 100 100 Citrus/olive 30 30 30 40 50 60 60 60 60 40 30 30 Inland water n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a

Y k Hr H

L

p r R h r k

= −⎛⎝⎜

⎞⎠⎟

−⎡⎣ ⎤⎦ =∑

2

months

exp /

2 exp( / )

0

0 0

Θ

ψ Ω

268 S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272

where Y is sediment loss (mean per unit area);k is soil erodibility; ψ, topographic erosionindicator; Ω, bio-climatic erosion indicator; His mean slope relief; L is mean slope length.The equation dynamically simulates the effectsof climate, vegetation biomass and topographyon soil erosion and resulting sediment yield.

All the necessary data for this model areintroduced to a regional geographical informa-tion system (GIS), and overlaid in accordancewith the developed algorithm. Maps of annualand monthly-predicted soil erosion rate are thencompiled. In outline, the methodology proposedfor maps compilation is as follows: • Each pixel is treated as a mixture of bare soil

area and plant cover. Two runoff thresholdfigures for each pixel are calculated;

• From land use classification estimate a vege-tation coverage percentage for each pixel isestimated for each month of the year;

• This cover is used to mix the two thresholdvalues to give a composite threshold for eachpixel and each month;

• The daily rainfall data is interpolated to calcu-late the total squared daily runoff for eachmonth of the year;

• The erodibility for each pixel is estimatedfrom the soil type data;

• Local relief is calculated from the DEM as thestandard deviation of elevation within a 2-kmradius around each pixel;

• For each month sediment loss is calculated asproduct of erodibility, runoff and local relief.

Based on the above methodology, in thisresearch, maps have been produced for pre-dicted annual and monthly soil erosion rate forAlbania. This approach includes data, which canbe found easy for any area.

3. Results and discussion

The daily rainfall sequence for four years,1980–1981 and 1990–1991 are obtained for 23climate stations distributed throughout the coun-try. For the study period the mean rainfall was1400 mm/year. The precipitation fell especiallyin autumn (558 mm or 40% of total) and winter(362 mm or 26%), with a maximum in October(271.50 mm or 17.50%), (Table 7). The highestdaily rainfall was on 23 October 1981 (Lushnjë467,8 mm), on 24 October 1981 (Gjirokastër427.8 mm), and on 8 November 1981 (Malësi eMadhe 310.4 mm). The lowest annual rainfall is572.1 mm (for Pogradec, 1990) and the highestare 3777.2 mm (for Malësi e Madhe, 1980) and3144 mm (for Gjirokastër, 1980). These highrainfalls present a severe risk in generatingrunoff and sediment loss.

Albania has a sparse land cover (Table 4).Land cover is lowest in autumn and winter. Inthe autumn, when rainfall peaks at an average of910 mm (66% of total) the surface has a very

Table 5Canopy interception storage

Land cover type Interception storage hi of storm rainfall (%)

Arable 5 Pasture, vineyards and

tree crops 10

Forest 20 Heterogeneous 10 Natural degraded land 5 Urban, rock, water lands n/a

Table 6Relative erodibility classes

Relative erodibility classes

Value Level of confidence

RMS error

Weak 1 Weak ±80 Moderate 3 Moderate ±50 Strong 10 Strong ±50

S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272 269

low land cover. In autumn and winter 34% ofthe country has a land cover of <13 and <10%,respectively. Minimum land cover is in Octoberand November. In October and November about32% of the country has a land cover less than10%. There is a land cover of 90–95% for 90%of the country for only two months (June andJuly). Vegetation does not play an essential rolein soil erosion at this season, because duringJune and July rainfall reaches only 85 mm(6% of total).

The soil storage is very low in October andNovember. Histogram analyses show that thesoil storage in October is <40 mm for 80% ofthe country, 41–80 mm for12%, and >80 mm for8%. In November, soil storage is <40 mm for76% of the country, 41–80 mm for 14% and>80 mm for 10%. A contrasting situation ispresented in June and July within storage.Storage is <40 mm for 13% of the country,41–80 mm for 60% and >81 mm for 27%.

A considerable number of maps are producedalong the mapping process in this research. Suchmaps are annual and monthly soil storage, soil,

land use, soil texture class, and erodibility. Inthis set of maps we point out the final ones. Thepredicted annual erosion map is shown in Fig. 1and the predicted monthly erosion maps inFig. 2. These figures demonstrate the erosionrate estimated for Albania. Annual erosion ratesare especially high in the southern and centralareas of the country. There are three areaswhere the annual erosion rate is more than100 t ha–1 y–1: two areas in Gjirokasër and onein Sarandë. The highest annual erosion rates arein Gjirokastër, Sarandë, Vlorë, Durrës, Kavajë,Tiranë, Lezhë, Shkodër and Skrapar. Annualerosion rates are more than 10 t ha–1y–1 in mostof these areas. The lowest annual erosion ratesare in Tropojë, Pukë, Kukës, Korçë, Gramsh,Fier and Lushnjë. These annual erosion ratesare generally less than 1 t ha–1y–1. Erosion ratesare very high especially in October, November,February, and December and low in June andJuly. In October, soil erosion is more than100 t ha–1 y–1 in a considerable area in south andthe centre of the country and in November therate is more than 100 t ha–1y–1 in some southernparts of the country.

4. Conclusions

Figs. 1 and 2 show that Albania is clearly acountry where soil erosion by water is a poten-tial problem. The erosion conditions (climate,land use and topography) are extreme and con-servation measures are urgently needed.

Analysing the erosion factors shows lowvalues of soil storage and sparse land cover,especially in October and November.

Erosion rates are high. For a considerablearea estimated erosion rates are more than10 t ha–1y–1, especially in the south and centerof the country. There are three areas where theannual erosion rate is more than 100 t ha–1y–1

(two in Gjirokastër and one in Sarandë). Erosionrates are highest in October, November, February

Table 7Mean monthly rainfall, and mean and standard deviationof daily rainfall

Months Mean monthly rainfall (mm)

Mean daily rainfall (mm)

Standard deviation of daily rainfall

January 65.61 18.77 11.50 February 115.52 17.43 16.28 March 86.94 14.63 15.64 April 107.93 16.19 14.09 May 145.20 10.32 12.63 June 50.23 12.79 9.23 July 35.10 11.82 6.46 August 53.51 18.58 17.46 September 81.42 11.43 8.88 October 271.50 17.45 1.43 November 205.20 12.19 11.73 December 180.93 30.29 36.08

270 S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272

and December and lowest in June and July(<1 t ha–1y–1). These rates show that erosion inmost cases is not well controlled and the mostappropriate soil management practices are notbeing applied. This evidently will create not onlyenvironmental but socio-economic costs. The

removal of vegetative cover especially by anthro-pogenically-induced change requires carefulmonitoring and management.

More research is requiring in Albania withregards to soil erosion and the use of sustainablesoil conservation practices.

Fig. 1. Predicted annual erosion rate for Albania.

S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272 271

Acknowledgements

We would like to thank the Royal Society(UK), which funded this work through contractNATO/98A/nvb. Also we thank Hydrometeoro-logical and Soil Science Institute (Albania) forproviding weather data, soil types and land use

maps and School of Geography, University ofLeeds for providing software facilities.

References

[1] M.J. Kirkby, R. Abrahat, M.L. McMahon, J. Shaoand J.B. Thornes, MEDALUS soil erosion

Fig. 2. Predicted monthly erosion rate for Albania.

272 S. Grazhdani, S. Shumka / Desalination 213 (2007) 263–272

models for global change, Geography, 24 (1998)35–49.

[2] M.J. Kirkby, M.L. McMahon, M.D. Agnewand S. Dhima, Modem (ITT 97/12STD2/0501),Co-ordinator: SAI-EMAP on behalf of the ECDirectorate General XII-D2, Sub-contractor: Prof.M.J. Kirkby, Annual Report, 1999.

[3] W.B. Langbein and S.A. Schumm, Yield of sedi-ment in relation to mean annual precipitation, Am.Geophys. Union Trans., 39 (1958) 1076–1084.

[4] C. Kosmas, N. Danalatos and L.H. Cammerratet al., The effect of land use on runoff and soilerosion rates under Mediterranean conditions,Catena, 29 (1997) 45–49.

[5] F. Rondelli, S. Grazhdani and S. Dhima, Eleva-tion of the water laminar erosion in Albania byuniversal soil loss equation, Anali della Faculta’di Agraria dell’Universita’ di Perugia, VolXLVIII, 1995, pp. 509–523.

[6] S. Dhima and S. Grazhdani, Vlerësimi i riskuttë erosionit potencial në kushtet e Shqipërisë,Buletini i Shkencave Bujqësore Nr., 2 (1997)15–23.

[7] S. Grazhdani, S. Dhima and V. Kovaçi, Subsur-face drainage effects on soil erosion and on nitro-gen, phosphorous and potassium losses insoutheastern Albania, in: Proceeding of Confer-ence on Erosion and Land Degradation in theMediterranean, Portugal University of Aveiro,14–18 June, 1995, 161–168.

[8] M.J. Kirkby, An erosion-limited hillslope evalua-tion model, Catena Suppl., 23 (1992) 157–187.

[9] M.J. Kirkby, Modeling the links between vegeta-tion and landforms, Geomorphology, 13 (1995)319–335.

[10] M.J. Kirkby and N.J. Cox, A climatic index for soilerosion potential (CSEP) including seasonal andvegetation factors, Catena, 25 (1995) 333–352.

[11] M.J. Kirkby, TOPMODEL: a personal view,Hydrological Proc., 11 (1997) 1087–1097.

[12] C. Kosmas, M.J. Kirkby and N. Geeson, Manualon key indicators of desertification and mappingenvironmentally sensitive areas to desertification,2000, p. 87.

[13] G.W. Musgrave, Quantitative evaluation of fac-tors in water erosion – a first approximation,J. Soil Water Conserv., 2 (1947) 133–138.

[14] F. Lopez-Bermudez, A. Romero-Diaz and J.Martinez-Fernandez, Vegetation and soil erosionunder a semi-arid Mediterranean climate: a casestudy from Murcia (Spain), Geography, 24(1998) 51–58.

[15] R.B. Bryan and I.A. Campell, Runoff and sedi-ment discharge in a semi-arid drainage basin,Geomorphology 58 (1986) 121–143.

[16] D.J. Mitchell, The use of vegetation and land useparameters in modeling catchment sedimentyields, in: J.B. Thornes (Ed.), Vegetation andErosion, Processes and Environments, Wiley,Chishester, 1990, pp. 289–314.

[17] D. King, Y.L.C. Biossonais, C.L. Bas, J. Daroussin,M.J. Kirkby, T.L. Montanarella and B. Jones,Work plan for preliminary erosion map ofFrance, 1998.