Sediment yield and erosion rate estimation in the mountain catchments of the Camastra artificial...

17
Sediment yield and erosion rate estimation in the mountain catchments of the Camastra articial reservoir (Southern Italy): A comparison be- tween different empirical methods M. Lazzari a, , D. Gioia a , M. Piccarreta a,c , M. Danese a , A. Lanorte b a Consiglio Nazionale delle Ricerche, IBAM, C/da S. Loja, Tito Scalo, PZ 85050, Italy b Consiglio Nazionale delle Ricerche, IMAA, C/da S. Loja, Tito Scalo, PZ 85050, Italy c Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR), Italy abstract article info Article history: Received 3 September 2014 Received in revised form 5 November 2014 Accepted 7 November 2014 Available online xxxx Keywords: Mountain catchments Articial reservoir RUSLE USPED Tu index Basilicata The sedimentary budget assessment is an important topic for the scientic and social community, because it is crucial to understand the dynamics of orogenic belts and cope with a number of practical problems, such as soil conservation and sediment accumulation in reservoir. Sediment yield or denudation rate estimates in southern-central Italy are generally obtained either by applying simple empirical relationships based on the sta- tistical regression between the geomorphic parameters of the drainage network and the suspended sediment yield measured at the drainage basins outlets, or by using qualitativequantitative sediment delivery ratio or ero- sion models. In this work, we carry out a study of catchment dynamics and a sedimentary yield computation of several mountain catchments of the central-western sector of the Basilicata region (southern Italy), which are lo- cated upstream of an articial reservoir. The sediment yield and erosion rate have been computed through both an indirect assessment of the suspended sediment yield, based on the Tu index (mean annual suspension sedi- ment yield), and the application of the RUSLE and USPED empirical methods. The results obtained have been compared with the historical data of sediment accumulation, measured in the articial reservoir of the Camastra dam where a detailed evaluation of the volumes of historical (i.e. about 40 years) sediment storage was collected. The collected dataset represents a basic tool both for the investigation of the morpho-dynamics of a typical mountain catchment of the Mediterranean area and the evaluation of sediment budget related to uvial and hill- slope processes. The 38-year-long record of the sediment storage in the Camastra articial reservoir located at the outlet of the studied mountain catchments permits the validation of empirical relationships based on the geo- morphic and climatic parameters of the drainage basin. Among the three different methods of sediment yield evaluation (Tu index, RUSLE and USPED), the Tu index showed the best prediction ability, although USPED erosion model also furnished a good estimation. As also conrmed by other works, the indirect estimation of sediment yield based on Ciccacci's empirical relationships can represent a good proxy of short-term denudation rates in the Mediterranean areas with geological and geomorphological features similar to the study area but the use of empirical models with increasing complexity such as the USPED erosion model can help to explore the spatial distribution of the sediment yield sectors of the drainage basin, the erosion hot-spots, and the role of landslides in sediment mobilization and hillslope-channel connectivity processes. © 2015 Elsevier B.V. All rights reserved. 1. Introduction The sedimentary budget estimate is one of the major topics in ap- plied geomorphology, especially when it is crucial to understand catch- ment dynamics as well as study and work out practical problems, such as the assessment of soil erosion and the estimation of sediment depo- sition in articial reservoirs. The theoretical concepts used for assessing the sediment budget are relatively simple and based on mass estima- tions in source and sink areas (de Vente et al., 2006; Brown et al., 2009). Nevertheless, the complex relationships between erosion and accumulation systems and the introduction of several simplications can inuence the quality of such estimates (Brown et al., 2009). In re- cent years, advances in computer technology have led to the prolifera- tion of physically-based predictive models aimed to the quantitative investigation of both soil erosion and sediment ux evaluations in catchments (Merritt et al., 2003). The renement of estimation models of catchment-scale denudation rates often requires an increase in the number of parameters, which are difcult to assess and assume as spa- tially and temporally homogeneous (Boardman, 2006; Vanmaercke et al., 2011). Another important limitation is the evaluation of the com- plex interactions and feedback mechanisms between the different Catena 127 (2015) 323339 Corresponding author. E-mail address: [email protected] (M. Lazzari). http://dx.doi.org/10.1016/j.catena.2014.11.021 0341-8162/© 2015 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Catena journal homepage: www.elsevier.com/locate/catena

Transcript of Sediment yield and erosion rate estimation in the mountain catchments of the Camastra artificial...

Catena 127 (2015) 323–339

Contents lists available at ScienceDirect

Catena

j ourna l homepage: www.e lsev ie r .com/ locate /catena

Sediment yield and erosion rate estimation in the mountain catchmentsof the Camastra artificial reservoir (Southern Italy): A comparison be-tween different empirical methods

M. Lazzari a,⁎, D. Gioia a, M. Piccarreta a,c, M. Danese a, A. Lanorte b

a Consiglio Nazionale delle Ricerche, IBAM, C/da S. Loja, Tito Scalo, PZ 85050, Italyb Consiglio Nazionale delle Ricerche, IMAA, C/da S. Loja, Tito Scalo, PZ 85050, Italyc Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR), Italy

⁎ Corresponding author.E-mail address: [email protected] (M. Lazzari).

http://dx.doi.org/10.1016/j.catena.2014.11.0210341-8162/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 3 September 2014Received in revised form 5 November 2014Accepted 7 November 2014Available online xxxx

Keywords:Mountain catchmentsArtificial reservoirRUSLEUSPEDTu indexBasilicata

The sedimentary budget assessment is an important topic for the scientific and social community, because it iscrucial to understand the dynamics of orogenic belts and cope with a number of practical problems, such assoil conservation and sediment accumulation in reservoir. Sediment yield or denudation rate estimates insouthern-central Italy are generally obtained either by applying simple empirical relationships based on the sta-tistical regression between the geomorphic parameters of the drainage network and the suspended sedimentyieldmeasured at the drainage basins outlets, or by using qualitative–quantitative sediment delivery ratio or ero-sion models. In this work, we carry out a study of catchment dynamics and a sedimentary yield computation ofseveralmountain catchments of the central-western sector of the Basilicata region (southern Italy), which are lo-cated upstream of an artificial reservoir. The sediment yield and erosion rate have been computed through bothan indirect assessment of the suspended sediment yield, based on the Tu index (mean annual suspension sedi-ment yield), and the application of the RUSLE and USPED empirical methods. The results obtained have beencomparedwith the historical data of sediment accumulation, measured in the artificial reservoir of the Camastradamwhere a detailed evaluation of the volumes of historical (i.e. about 40 years) sediment storagewas collected.The collected dataset represents a basic tool both for the investigation of the morpho-dynamics of a typicalmountain catchment of theMediterranean area and the evaluation of sediment budget related to fluvial and hill-slopeprocesses. The 38-year-long record of the sediment storage in the Camastra artificial reservoir located at theoutlet of the studied mountain catchments permits the validation of empirical relationships based on the geo-morphic and climatic parameters of the drainage basin. Among the three different methods of sediment yieldevaluation (Tu index, RUSLE andUSPED), the Tu index showed the best prediction ability, althoughUSPED erosionmodel also furnished a good estimation. As also confirmed by other works, the indirect estimation of sedimentyield based on Ciccacci's empirical relationships can represent a good proxy of short-term denudation rates inthe Mediterranean areas with geological and geomorphological features similar to the study area but the useof empirical models with increasing complexity such as the USPED erosionmodel can help to explore the spatialdistribution of the sediment yield sectors of the drainage basin, the erosion hot-spots, and the role of landslides insediment mobilization and hillslope-channel connectivity processes.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

The sedimentary budget estimate is one of the major topics in ap-plied geomorphology, especially when it is crucial to understand catch-ment dynamics as well as study and work out practical problems, suchas the assessment of soil erosion and the estimation of sediment depo-sition in artificial reservoirs. The theoretical concepts used for assessingthe sediment budget are relatively simple and based on mass estima-tions in source and sink areas (de Vente et al., 2006; Brown et al.,

2009). Nevertheless, the complex relationships between erosion andaccumulation systems and the introduction of several simplificationscan influence the quality of such estimates (Brown et al., 2009). In re-cent years, advances in computer technology have led to the prolifera-tion of physically-based predictive models aimed to the quantitativeinvestigation of both soil erosion and sediment flux evaluations incatchments (Merritt et al., 2003). The refinement of estimation modelsof catchment-scale denudation rates often requires an increase in thenumber of parameters, which are difficult to assess and assume as spa-tially and temporally homogeneous (Boardman, 2006; Vanmaerckeet al., 2011). Another important limitation is the evaluation of the com-plex interactions and feedback mechanisms between the different

324 M. Lazzari et al. / Catena 127 (2015) 323–339

parameters used to describe erosion and sedimentation (de Vente andPoesen, 2005). Due to all these limitations, the physically-based predic-tion models of soil erosion showed controversial results and poorreliability (de Vente and Poesen, 2005; Boardman, 2006; De vente etal., 2006) and their application is limited to small areas where a largeamount of data are available (de Vente and Poesen, 2005).

The difficulty in applying the physically-based erosion models tonatural landscapes lies in the fact that sediment yield predictions arestill widely based on very simple empirical models developed bymultiple regression methods between morpho-climate parametersand limited measurements of sediment yield and/or sediment fluxes(Jansen and Painter, 1974; Ciccacci et al., 1980; Mulder and Syvitski,1996; Poesen et al., 2003).

In Italy, sediment yield and denudation rate evaluation are quite rareand are greatly affected by most of the above-mentioned problematicaspects. Short-term denudation rates are generally obtained eitherfrom simple regression equations which involve hydro-geomorphic pa-rameters and the suspended sediment yield estimates of several Italiandrainage basins (Ciccacci et al., 1980; Della Seta et al., 2007, 2009;Santangelo et al., 2013; Vergari et al., 2013; Borrelli et al., 2014; Gioiaet al., 2014) or from empirical sediment delivery ratio and/or erosionmodels (Onori et al., 2006; Capolongo et al., 2008a; Fagnano et al.,2012; Conoscenti et al., 2013). The validation of sediment yield assess-ment is frequently difficult to achieve and the predictive ability ofthese equations is limited to the specific regions for which they havebeen developed (de Vente and Poesen, 2005). Another limit is the im-possibility to fully consider the spatial variation of sediment yieldwithinthe catchment. Moreover, direct measurements of sediment flux instreams are extremely rare and limited to a few years and one of themost implemented approaches to estimate sediment yield is based onsediment storage in closed systems, such as lakes and artificial reser-voirs (Van Rompaey et al., 2005; de Vente et al., 2006).

In this work, we have estimated sediment yield in amountain catch-ment of the central-western sector of the Basilicata region, southernItaly, (Fig. 1) by draining an artificial reservoir (i.e. the Camastra dam,Fig. 2). Specifically, the sediment yield evaluation has been obtainedthrough both an indirect estimation of suspended sediment yieldbased on the Tu index (mean annual suspended sediment yield,Ciccacci et al., 1980) and the application of the RUSLE (Renard et al.,1997) and the USPED(Mitasova et al., 1996) empirical methods.

Themain purpose of thework is to compare the results coming fromthese different empirical models of sediment yield estimation with themid-term (i.e. about 40 years) record of sediment accumulation in theartificial reservoir. The validation of such estimations of sedimentyield at the scale of large catchments using sediment storage in reser-voir allows testing the reliability and the predictive ability of differentempirical estimations based on drainage network morphometricproperties. Furthermore, we performed an in-depth geomorphologicalanalysis of the area in order to outline the main geomorphological pro-cesses acting within the studied catchment and their role on the spatialdistribution of the sediment yield. This kind of approach allows us tocompare the results of erosion models with the erosional processes ofthe study area.

Fig. 1. (a) Geological setting of the southern Apennines. The study area is represented in the bsyntectonic deposits; 3) Cretaceous to Oligocene ophiolite-bearing internal units; 4) Mesozoic–toMiocene shallow-water and deep-sea successions of the Lagonegro-typeMonte Arioso unit; 6Cretaceous to Miocene deep-sea successions of the Lagonegro-type Campomaggiore unit; 8) Mthrust front of the chain. (b). Geological sketch map of the Camastra catchment. Legend: 1) alPleistocene toHolocenematrix-supported breccias); 3)Upper Pliocene to Early Pleistocenematsilty clays and gravels; 5) Lower-Upper Pliocene gray-blue silty clays rarely interbedded with cogravels levels; 7) Gorgoglione Flysch Fm (upper Miocene syntectonic siliciclastic turbidites); 8)9) Argille Varicolori and Corleto Perticara Fms (middle Cretaceous to Oligocene varicoloured clasiliceousmarls and shales); 11) Scisti silicei Fm (Jurassic radiolarites and cherts); 12) Calcari conshallow-water siliciclastic sediments, organogenic limestones and, toward the top, siliciclastplatform.

2. Geological and climatic setting

The test area is located in the axial-outer belt of the southernApennines (Fig. 1), featured by an alternation of tectonic basins andmorphostructural ridges related to Pliocene and Early Quaternaryphases of thrusting and folding (Amato and Cinque, 1992; Bonini andSani, 2000).

The southern Apennines (Fig. 1) are a north-east verging fold-and-thrust belt, derived from the deformation of the western border of theApulian plate from late Oligocene to Pleistocene time (Pescatore et al.,1999). The belt ismainlymadeof shallow- and deep-water sedimentarydeposits derived from the deformation of Mesozoic–Cenozoic circum-Tethian domains and Neogene–Pleistocene syntectonic and foredeepdeposits (Pescatore et al., 1999; Menardi Noguera and Rea, 2000;Lazzari, 2008). Starting from Langhian–Tortonian times, the Apenninethrust front moved progressively toward the east and was followedby a back-arc extension, responsible for synchronous extensionalcollapse of the inner domains of the thrust belt and Tyrrhenian seaopening (Malinverno and Ryan, 1986). The Pliocene to Pleistocenepost-collisional history of the southern Apennines is characterized bystrike-slip and extensional tectonics, which promoted the creation oflongitudinal and transversal fault-bounded basins (Cinque et al., 1993;Schiattarella et al., 2006; Gioia et al., 2011b).

The study area is located at about 10 km to the south of the Potenzatown and it is mainly characterized by strongly deformed geologicalunits of the Lagonegro basin and flysch deposits of Miocene syntectonicbasins. Themiddle Triassic-to-Miocene Lagonegro units are characterizedby shallow-water, basinal and shelf-margin facies and limestone andsiliciclastic deposits of pelagic environment, affected by dome-and-basin folds (Pescatore et al., 1999). Middle Cretaceous to Oligocene grayand reddish clays and marls (Argille Varicolori and Corleto PerticaraFms) and upper Oligocene to lower Miocene marls and volcaniclasticsandstones (Tufiti di Tusa Fm) also outcrop widely in the study area.The upper Miocene deposits are mainly constituted by deep-sea con-glomerates, sandstones andpelites (Gorgoglione Flysch Fm.), unconform-ably overlying the Lagonegro units (Pescatore et al., 1999).

The Lower Pliocene–Early Pleistocene clastic deposits of the Calvellopiggy-back basin, mainly made of a 300m-thick succession of gray siltyclays with rare shell fragments, interbedded with fine sands, uncon-formably overlay the deformed pre-Pliocene bedrock (Amato andCinque, 1992; Bonini and Sani, 2000). A 100 m-thick succession com-posed of Early Pleistocene marine/transitional gravels with intercala-tions of sandy levels unconformably overlies the upper Pliocene–EarlyPleistocene marine sediments (Amato and Cinque, 1992).

From a geomorphological point of view the main streams of thestudy sector cut across the NW–SE trending contractional structures,with minor channels arranged in a dendritic pattern (Fig. 2). The belttops are frequently characterized by remnants of ancient polycyclic ero-sional land-surfaces, elevated by the Quaternary regional uplift anddisplaced by fault activity (Amato and Cinque, 1999; Schiattarellaet al., 2013).

The climatic context has been defined using temperature data fromPotenza and rainfall data record of the Calvello and Laurenzana stations.

ox. Legend: 1) Pliocene to Quaternary clastic deposits and volcanic products; 2) MioceneCenozoic shallow-water carbonates of the Apennines platform; 5) lower–middle Triassic)Mesozoic toMiocene deep-sea successions of the Lagonegro-type Groppa d'Anzi unit; 7)esozoic–Cenozoic shallow-water carbonates of the Apulian platform; 9) volcanoes; 10)

luvial deposits (Upper Pleistocene to Holocene sand and gravel) 2) slope deposits (Upperrix-supported gravelswith intercalations of sandy levels; 4)Upper Pliocene sand, gray-bluearse sand and gravels; 6) Lower Pliocene sandwith intercalations of clay, calcarenites andTufiti di Tusa Fm (upper Oligocene to lower Miocenemarls and volcaniclastic sandstones;ys andmarlswith calcarenites and calcilutites); 10) Galestri Fm (lower–middle Cretaceousselce Fm (upper Triassic cherty limestones); 13)Monte Facito Fm (lower–middle Triassicic basinal deposits; 14) Mesozoic–Cenozoic shallow-water carbonates of the Apenninic

325M. Lazzari et al. / Catena 127 (2015) 323–339

The yearly average temperature ranges from 11.6 °C to 14.5 °C, with anaverage maximum between 18.6 °C and 24.5 °C during summer and anaverage minimum ranging between 3.2 °C and 8.0 °C during winter(Piccarreta et al., in press). The pluviometric stations furnish an averageannual rainfall of about 621 mm yr−1 (Calvello) and 735 mm yr−1

(Laurenzana). The annual and seasonal total precipitation indicated ageneral downward trend over the 1951–2010 period, mainly due to adecrease in precipitation during the autumn–winter season, whichalso occurs in other areas of the Mediterranean basin. As observed inother sectors of the Basilicata region, analyses of multi-day sequences

326 M. Lazzari et al. / Catena 127 (2015) 323–339

of moderate to heavy rainfall indicated an increase in their frequencyand intensity, especially during the last decade (Piccarreta et al.,2013). The occurrence of multi-day extreme events with higher intensi-ty/frequency is probably the main triggering factor of the increase ofsevere flooding and landslide events, not only in autumn and winter,but also in early spring (Lazzari et al., 2006).

3. Source to sink: approaches and methods

In order to calculate the sediment budget in the studied mountaincatchments, three different empiricalmethods of sediment yield assess-ment have been used (Tu index, Ciccacci et al., 1980; RUSLE,Renard et al.,1991; USPED,Mitasova et al., 1996). Themost reliable erosionmodel hasbeen chosen on the basis of the comparisonwith the sediment accumu-lation in the artificial reservoir located at the outlet of the catchmentstudied (i.e. the Camastra dam). In order to investigate the spatial distri-bution of sediment yield and its relationships with fluvial and slope

Fig. 2. DEM of the study area with Strahler's hierarchical scheme of t

processes, a classical geomorphological analysis has been carried out(i.e. aerial-photo interpretation and field survey). The validation ofsuch sediment yield estimations at the scale of large catchments usingthe sediment storage in reservoir permitted to test the reliability andthe predictive ability of different empirical estimations based on themorphometric properties of drainage network. This approach is basedon the assumption that the amount of sediment storage in catchmentsis only a limited amount of the total sediment yield (i.e. the erosionalprocesses are dominant with regard to the depositional ones withinthe catchments studied). This assumption has been verified through ageomorphological field-survey(Fig. 3), the analysis of the distributionof sedimentary bodies along the main catchment channels and thegrain-size analysis of samples cored in the soft-bottom damsediments. Finally, a GIS-supported statistical analysis of the areasfeatured by landslides and accelerated erosion processes has been car-ried out and compared with the spatial distribution of sediment yieldcoming from the erosion models.

he catchments. Numeration of sub-basins is shown in the frame.

Fig. 3. Geomorphological and landslide density maps of the Camastra catchment.

327M. Lazzari et al. / Catena 127 (2015) 323–339

3.1. Historical sediment storage in the Camastra artificial reservoir

The artificial reservoir considered in this study is the Camastra dam,built between 1965 and 1967, located in the central sector of theBasilicata region (southern Italy), at the outlet of several drainage basins

Fig. 4. 38-year record of sediment storage in the Camastra dam. Dot

(i.e. the Inferno, Camastra and Carpino basins), which drain an area of330 km2(Fig. 2). Considering also the small catchments located on thedam slopes, the total drainage area of the artificial reservoir is equal to350 km2. During the activity period of the dam, five bathymetric andseismo-acoustic surveys integrated with geotechnical core drilling

ted line represents the mean annual sediment yield since 1967.

328 M. Lazzari et al. / Catena 127 (2015) 323–339

programs were performed in 1988, 1993, 1995, 1997 and 2005 (Molinoet al., 2007). The amount of the sediment volumes stored in the damwas reconstructed on the basis of the comparison between the pre-existingland-surface and the sediment volumes estimated by thedepth of the reservoir infill. Then, the reservoir sedimentation basedon bathymetric and seismo-acoustic surveys was used in order to esti-mate the mean annual sediment deposition for the following periods:1967–1988, 1989–1993, 1994–1995, 1996–1997 and 1998–2005 (Fig. 4).

Reservoir sedimentation data can provide useful information onthe catchment sediment yield, although some possible sources of uncer-tainty should be considered, such as trap efficiency, wave erosion of thereservoir shoreline.On thebasis of an analysis of thedammanagementhis-tory, VanRompaey et al. (2005) assumed a trap efficiency close to the100%for theCamastra reservoir. Another causeof uncertainty in reservoir data toestimate sediment yield is the additional source of sediments related towave erosion of the reservoir shoreline, although it has generally been ar-gued that this source of sediment is only a limited contribution to the totalamount of sediment storage in the reservoir (de Vente et al., 2006).

Historical data of sedimentation in theCamastra reservoir (i.e. 38 years)allowed an estimation of the total sedimentation in the reservoir in about15,000,000 m3. Apart from the sediment stored in the 1989–1993 and1997–2005 time span (Fig. 4), temporal variation of annual sedimentyield exhibits a low variability. It is worthy of note that the 1989–1993 pe-riod is characterized by several landslide events affecting the eastern slopeof the dam,which brought about a significant increase in sediment storagewithin the reservoir (Molino et al., 2007). This hypothesis is confirmed bythe analysis of multitemporal isopach maps, showing a strong increase ofsediment thickness along the eastern sector of the dam during the 1988to 2005 period. It is worth noting that, at the end of 1980, drought periodsoccurred and the reservoir was completely drained several times due to avariation of the strategies of water use. According to Molino et al. (2007),the associated rapid drawdown could have promoted the landslide activi-ties of the reservoir slope.

Since 1997,management plans have been carried out along the east-ern slope of the reservoir and themost important tributaries of the lake.The significant reduction of sediment storage in reservoir during the1997–2005 time span can be related to these management practices(Molino et al., 2007).

In order to obtain a sediment yield estimation more representative ofthefluvial system, themean annual sediment storage has been estimatedexcluding the anomalous data from the 1988–1993 and 1997–2005years. The mean annual sediment yield is equal to 390,000 m3.

The core drilling and seismic and bathymetric surveys indicate thatthe reservoir sediment infill is constituted by silt and mud deposits,thus implying that the suspended load is themost relevant contributionto the sedimentation in the Camastra dam.

Grain-size analysis indicates that the samples acquired on theCamastra reservoir bottom (Fig. 2) can be classified as sandy silt. Thebulk density of these sediments is 1.3–1.4 g cm−3. Using this bulkdensity value, it has been possible to convert the sediment volumesinto a mean annual sediment yield, which is directly comparable withmean annual sediment yield coming from the different empiricalmethods used in this work. The resulting reservoir sediment yield is1615 Mg km−2 yr−1.

3.2. RUSLE

The Revised Universal Soil Loss Equation (RUSLE) is a detachment-limited model which predicts the average soil loss (A) by followingthe equation (Renard et al., 1997)

A ¼ R � K � L � S � C � P ð1Þ

where, A [Mg ha−1 yr−1] is the annual average soil loss, R[MJ mm h−1 ha−1 yr−1] is the rainfall intensity factor, K[Mg h MJ−1 mm−1] is the soil erodibility factor, L [dimensionless]

is the slope length factor, S [dimensionless] is the slope steepness factor,C [dimensionless] is the land cover factor, and P [dimensionless] is thesoil conservation or prevention practice factors (Renard et al., 1991).

3.2.1. Model parametersThe R factor depends on rainfall–runoff characteristics, in turn influ-

enced by geographic location. In the original formulation of the USLE,the R factor was calculated as the product of the total kinetic energy ofthe storm (E) and its maximum 30-minute intensity (I30). Since thisinformation is not readily available, different replacement equationshave been proposed over time for the calculation of this factor.

The equation used in the present study is based on the power-lawequation developed for Basilicata by Capolongo et al. (2008b), derivedfrom 20-minute and hourly precipitation data. It takes into account adaily precipitation of ≥10.0 mm:

EI30 ¼ 0:1087 P24ð Þ1:86 ð2Þ

where EI30 is the rainfall erosivity in MJ mm h−1 ha−1 yr−1 and P24isthe daily rainfall amount in mm. A raster layer (cell size 10 × 10 m),containing a spatially distributed annual averaged R-factor value, hasbeen created for the period from January 1951 to December 2010.

The soil erodibility factor (K) is the rate of soil loss per rainfall ero-sion index unit, Mg h MJ−1 mm−1 as measured on a standardWischmeier erosion plot. The standard plot is 22.13 m long and has a9% slope. Practically, the K factor is the average, long-term soil and soilprofile response to the erosive power of rainstorms. Specifically, it rep-resents soil detachment and transport by raindrop impact and surfaceflow, localized deposition due to topography and roughness inducedby tillage, as well as rain water infiltration into the soil profile. In thisstudy we have used the following equation:

K ¼ 2:1 � 10−4 � 12‐Mð Þ � Siþ fSð Þ � 100‐Cð Þ 1:14þ 3:25 A‐2ð Þ þ 2:5 P‐3ð Þh i

=100

ð3Þ

where, M is the organicmatter content (%), Si is the silt content (%), 2 to50 μm, fS is thefine sand content (%) 50 μ to100 μm, C is the clay content(%) less than 2 μm, S is the sand content (%) 50 μm to 2 mm, A is thestructure and P is the permeability class (within the top 0.60 m).

The A structure consists of four classes (classes 1–4):

1 = very fine granular (aggregates 1–2 mm dia)2 = fine granular (aggregates 2–3 mm dia)3 = medium or coarse granular (aggregates 3–5 mm dia)4 = blocky, platy, or massive (usually construction sites).

The permeability consists of six classes:

1 = rapid (sandy to gravelly soils, no layering)2 = moderate to rapid (sandy loam)3 = moderate (loam, silty loam soils)4 = slow to moderate (clay loam, silt soils)5 = slow (high clay content, or compacted soils of other texturalgroups)6 = very slow (high clay content, poor aggregation).

Texture, structure and permeability data have been extractedfrom the regional pedological map of Basilicata in scale 1:250,000(http://www.basilicatanet.it/suoli/index.htm),while the organicmatterdata have been derived from the European Soil datasets (Jones et al.,2005).

In order to incorporate the impact of flow convergence in the slope-length factor (LS) [dimensionless] estimation, the RUSLE equation pro-posed by Renard et al.(1991) has been replaced by the ones proposedin RUSLE-3D by Mitasova et al.(1996). The topographic parameterslope and specific catchment area have been derived from a 10 m

329M. Lazzari et al. / Catena 127 (2015) 323–339

DEM, delineated through the Topogrid sub-module of ArcInfo 9.3(Hutchinson, 1989), where contour lines and elevation points of a1:10,000-scale topographic map formed the basis of the interpolation.

The modified LS factor is given by the following relation:

LS dimensionless½ � ¼ mþ 1ð Þ � Aca0

� �m� sin β

β0

� �nð4Þ

where, Ac(m) is the upslope contributing area per unit of width; β is theslope angle (degree); α0 is the length (72.6 ft, equal to 22.13 m) and β0

is the angle of the standard terrain parcel (9%, equal to 5.16° slope) ofUSLE. Finally, m (0.6) and n (1) are exponent parameters dependingon the specific prevailing type of flow.

The C-factor [dimensionless], which reflects the effects of croppingandmanagement practices on soil erosion rates, has been calculated fol-lowing the authors' suggestions (Renard et al., 1997; Wischmeier andSmith, 1978), with values assigned on the basis of the vegetativecover, its density and the monthly rainfall runoff erosivity.

As concerns the characterization andmapping of the vegetation types,we started from the Landsat-TM data processing and the subsequent ap-plication of techniques of supervised classification (Maximum LikelihoodClassification) and methods of spectral analysis at the sub-pixels level(Mixture TunedMatching Filtering, Lasaponara et al., 2006). The prelimi-nary activity in the mapping process was the ground-truth retrieval data,obtained through an accurate recognition of the study area developedover several years and aimed at the identification of vegetation types.

Fig. 5. Landsat image (Landsat-TM imageRGB3-2-1, August 2, 2011) and vegetationmap (vegetchestnut woods; 5 conifer woods; 6 beech woods; 7 alder woods; 8 ostrya-carpinifolia woods;forestations; 13 exotic broadleaves reforestations; 14 broom shrublands; 15 other shrublands; 1woods (Quercus frainetto); 19 mesophytic grasslands; 20 xerophytic grasslands; 21 arable landrocks and debris; 27 badlands; and 28 bare soils.

ALANDSAT-TM image acquired on August 2, 2011 (Fig. 5a) was selectedand processed. In order to take into account the phenological changes ofvegetation, related to the seasonal differences, the image was comparedwith a LANDSAT-TM spring image (June 15, 2011). The images are totallycloud-free and have a good illumination (large solar elevation angle). Theimages were georeferenced using a set of approximately 30 ground-truthpoints. In the next step, a supervised classification (Maximum LikelihoodClassification) was applied and subsequently also the spectral analysis atthe sub-pixel level (Mixture TunedMatch Filtering) in order to obtain thevegetation-type map.

Theuse of spectral analysis at the sub-pixel level can reduce theuncer-tainty in hard classification techniques since it is able to capture, ratherthan ignore, sub-pixel heterogeneity (Shimabukuro and Smith, 1991). Fi-nally, we identified 20 classes (Fig. 5) corresponding to the spectral rela-tively homogeneous vegetation types identified during the ground-truthretrieval and the selection of an appropriate number of regions of interest(ROI) to define the classification training points (Riaño et al., 2002). Theclassification also included some non-vegetated classes.

The C-values (Table 1) were assigned to the resulting 12 land useclasses according to the C-values proposed in the literature (Chisci andBazzoffi, 1995; Zanchi and Giordani, 1995).

3.3. USPED method

The Unit Stream Power Erosion and Deposition model (USPED) hasthe basic model structure of the RUSLE Eq. (1)(Renard et al., 1991).

ationmap fromLANDSAT-TM spectral analysis). Legend: 1Water; 2 floodplains; 3 urban; 49 mixed broadleaves woods; 10 hygrophiles formations; 11 garrigue; 12 exotic conifer re-6 Turkey oakwoods (Quercus cerris); 17 oakwoods (Quercus pubescens); 18Hungarian oaks; 22 olive groves; 23 conglomerates outcrops; 24 sand outcrops; 25 cultivated areas; 26

Table 1C-values assigned to the vegetation classes.

Class name C-factor value

Broad-leaved trees 0.042Built-up areas 0.001Conifer trees 0.009Degraded areas 0.067Farming 0.003Garrigue 0.025Grasslands 0.003Olive-groves 0.0007Reforestation with both exotic conifer and broad-leaved trees 0.0007Sown ground 0.028

330 M. Lazzari et al. / Catena 127 (2015) 323–339

In order to estimate the erosion and deposition (ED), which areboth computed as a change in sediment transport capacity across aGIS grid cell, the following equation is used (Mitasova et al., 1996):

ED ¼ ∂ A cosαð Þ=∂xþ ∂ A sinαð Þ=∂y ð5Þ

where,α is the slope aspect of the terrain (in degrees). Net erosion areascoincidedwith areas of profile convexity and tangential concavity (flowacceleration and convergence), and net deposition areas coincided withareas of profile concavity (decreasing flow velocity).

3.4. Morphometric analysis of drainage network of the study area andempirical estimation of sediment yield

Aquantitative analysis of the drainage basin has been performedwiththe support of a high-resolution DEM to investigate the morphometric

Table 2Morphometric indexes and Tu index values calculated for each hydrographic basin.

R Ga Da ga Θ (%) Emax basin (m) Emin basin

1.01 197.00 0.90 3.72 6.28 1244 5400.48 29.00 0.29 1.88 10.87 1365 5400.60 2437.00 1.09 9.38 4.61 1833 5400.77 150.00 1.00 17.43 10.41 1287 6090.80 185.00 1.00 6.98 10.81 1725 7930.50 36.00 0.51 4.53 14.14 1833 7930.68 556.00 0.73 6.04 4.76 1742 5920.70 234.00 0.74 5.97 6.52 1391 5670.66 78.00 0.57 6.00 9.50 1433 6430.68 82.00 0.72 5.79 12.59 1725 8420.25 5.00 0.20 0.71 13.43 1594 8420.17 5.00 0.36 13.51 18.31 1092 8420.56 28.00 0.53 3.85 14.78 1833 8380.99 140.00 0.51 3.50 5.15 1805 6540.37 18.00 0.50 13.04 15.64 1231 7120.29 7.00 0.25 3.60 13.64 1028 6651.02 83.00 0.87 22.68 13.88 1287 6620.39 83.00 0.44 3.13 7.22 1476 6880.00 0.00 0.00 0.00 11.79 1144 6150.17 3.00 0.17 0.60 11.82 1177 7390.46 22.00 0.42 1.71 7.08 1244 7370.70 70.00 0.72 4.62 6.05 1237 6520.08 1.00 0.08 0.62 15.70 1316 8800.63 26.00 0.35 3.69 10.32 1391 8850.06 3.00 0.05 0.71 17.38 1336 7200.42 10.00 0.42 4.48 15.13 1214 6910.60 29.00 0.57 3.53 10.16 1334 6250.83 15.00 0.36 10.34 20.21 1113 6500.43 10.00 0.48 6.99 16.72 1104 6681.00 30.00 0.88 18.89 12.59 1093 6910.33 17.00 0.61 12.69 7.86 960 7200.46 33.00 0.33 1.58 8.87 1712 7480.59 46.00 0.49 5.26 10.51 1742 7741.19 68.00 0.70 11.24 11.27 1472 774

Ltot: total length of channels; D: drainage density; Rb: bifurcation ratio; Rbd: direct bifurcationga: hierarchic anomaly density; θ: gradient of the main stream; Qmax,Qmean,Qmin: maximum, msuspended sediment yield estimated by Eq. (1)(Tu index).

properties of physical landscape and drainage basin. The DEM with aspatial resolution of 10 m has been extracted through the interpolationof contours and spot heights deriving from 1:5000 and 1:10,000 scaletopographic maps.

The drainage network of the study area has been derived from theI.G.M.I. (Italian Military Geographical Institute) 1:25,000-scale topo-graphic map and photo-aerial interpretation. Then, the fluvial networkhas been digitized by using the software ArcGIS on geo-referred topo-graphic maps and the channels have been classified according to theStrahler's hierarchic scheme.

Several morphometric parameters related to the recent/activegeomorphological system have been evaluated for each sub-basin(Table 2) in order to perform the indirect estimation of the suspendedsediment load (Tu, mean annual sediment yield transported in suspensionper unity area within reservoir basins, Mg km−2 yr−1) of the drainagenetwork of the study area. Empirical equations are estimated bymultipleregressions between the measured amount of suspended sediment yieldand a large number of geomorphic and climatic parameters in fourteenItalian catchments (Ciccacci et al., 1980).

Specifically, the parameters are:

• Drainage density (D= Lst/Atot), where Lst is the sum of all the streamlengths within a sub-basin and Atot is the drainage basin area;

• bifurcation ratio (Rb=Nu/Nu + 1), where Nu and Nu + 1 are the num-ber of streams per order u and u + 1, respectively (Strahler, 1957);

• direct bifurcation ratio (Rbd= Ndu/Nu + 1), where Ndu is the numberof streams of u order flowing in u+ 1 order and Nu + 1 is the numberof streams per order u + 1 (Avena et al., 1967);

• bifurcation index (R=Rb–Rbd), where Rb andRbd are the bifurcationratio and the direct bifurcation ratio, respectively (Avena et al., 1967);

(m) Emean basin (m) Ip # Mean Slope (°) Tu * (T km2 yr−1)

919 0.540 13.6 261.40894 0.432 16.6 396.60995 0.355 15.1 720.70860 0.370 12.8 1712.82

1193 0.431 15.0 692.321221 0.407 17.1 856.21995 0.353 15.9 683.73980 0.503 14.0 906.97

1019 0.479 12.4 1469.111243 0.454 16.1 766.641233 0.521 14.3 464.69982 0.563 15.4 7323.81

1247 0.408 17.2 702.561042 0.339 15.3 457.31935 0.432 13.4 5350.18837 0.481 13.4 1316.06978 0.506 13.4 3530.03

1011 0.409 18.3 642.86914 0.567 10.7 419.22

1011 0.619 11.2 310.331025 0.568 15.0 290.25944 0.498 13.6 539.06

1117 0.545 15.0 1709.401172 0.567 12.9 1103.521072 0.573 14.1 2973.18922 0.441 14.0 1264.69995 0.521 14.8 598.81953 0.627 24.1 4312.98895 0.524 18.7 3213.50884 0.483 15.8 1629.99858 0.584 14.9 974.39

1141 0.408 12.7 402.861158 0.394 14.0 1578.461123 0.501 23.4 1479.14

ratio; R: bifurcation index; Ga: hierarchic anomaly number; Da: hierarchic anomaly index;ean and minimum elevation of the sub-basin; Ip: hipsometric integral; Tu: mean annual

331M. Lazzari et al. / Catena 127 (2015) 323–339

• hierarchical anomaly number (Ga), the number of I order streams toadd for making the drainage network perfectly hierarchized, i.e.with a value of R = 0 (Avena et al., 1967); the index is a measure ofthe number and the type of anomalous confluences (i.e. channels ofu order not flowing in channels of order u + 1) of the drainagebasin (Avena et al., 1967; Capolongo et al., 2005; Della Seta et al.,2007; Gioia et al., 2011a; Santangelo et al., 2013). More specifically,Ga is the total sum of the number of anomalous confluences multi-plied by a weight factor which depends to the hierarchical differencebetween confluent channels. The higher the difference betweenanomalous channel and higher-order stream, the higher the weightfactor (see Avena et al., 1967 for more details on the values of theweight factor).

• hierarchical anomaly index (Da = Ga/N1), where Ga is the hierarchi-cal anomaly number and N1 is the number of 1st order streams(Avena et al., 1967);

• hierarchical anomaly density (ga = Ga/A), where Ga is the hierarchi-cal anomaly number and A is the area of the sub-basins (Avena et al.,1967).

The empirical equations of Ciccacci et al.(1980) with high statisticalcorrelations established a relation between the predicted suspendedsediment yield and these geomorphological parameters. Several worksdemonstrated that the indirect estimation of mean annual suspendedsediment yield (i.e. Tu values) obtained by this approach are consistentwith short-term denudation rates estimated by reservoir sedimentation(Gioia et al., 2011a) and field monitoring data of erosion (Della Setaet al., 2007, 2009). In the first analysis, this is also confirmed by thecomparison between the predicted sediment yield based on empiricalequations and specific sediment yield resulting from sediment storagein Camastra dam. The indirect estimation of sediment yield relatingto the study area has been performed using the Eq. (6), which isbased on spatially-distributed geomorphological parameters (D and ϑ;Ciccacci et al., 1980):

log10Tu ¼ 2:7454 log10 Dþ 0:0639 ϑþ 1:15548 ð6Þ

where, Tu is the mean suspended sediment yield (Tu index), D is thedrainage density and ϑ is the mean slope of the main stream. Eq. (6)suggests that sediment yield increaseswith the increase of the drainagedensity and the slope of the channels. The indirect estimation of meanannual suspended sediment yield (i.e. Tu, expressed as Mg Km−2 yr1)has been made for each sub-basin within the hydrographic catchmentof the Torrente Camastra.

3.5. Geomorphological analysis

In order to outline the main geomorphological processes actingwithin of the catchment studied and their role on the spatial distribu-tion of the sediment yield, a geomorphological map (Fig. 4) has beendrawn by field survey, multi-temporal aerial photointerpretetionand a critical analysis of literature data (Conforti et al., 2013; Lazzari,2011). The photo-aerial interpretation has been made by analyzingthe aerial photographs at a 1:33,000 scale (1955, 1974 and 1991years). This approach allows us to identify fluvial- and massmovement-related landforms and investigate the spatial distributionof denudation processes of the study area. Landslides have been classi-fied based on the type of movement and the state of activity followingthe scheme proposed by Cruden and Varnes (1996). Geomorphologicalanalysis has also been focused on the delimitation of badlands and areasaffected by severe rill and gully erosion. The mapping of fluvial-relatedprocesses include alluvial terraces and floodplain depositionallandforms such as gravel- and sand-bed river bars.

4. Results

4.1. Geomorphological analysis and drainage basin morphometry

The study area is a medium to high relief landscape, featured byelongated catchments with steep slopes and narrow valleys (Fig. 3).The main streams of the area (e.g. Torrente Camastra and Fiumarad'Anzi rivers) flow within highly-gradient and roughly rectilinear val-leys. Geomorphological analysis highlights that the study area is alandslide-dominated landscape. The slopes are largely affected by land-slides of different types, sizes, ages and distributions. We mapped 653landslides, classified as: 516 earth-flows, 43 rotational slides, 3 transla-tional slides and 91 shallow landslides (creep, coalescent mudflowswith a depth of the surface slip lower than 2 m), whereas the slopeshaped on the Pliocene clayey deposits outcropping along the lowervalley of the Fiumara La Terra river are largely affected by badlandareas. The active phenomena represent 25% (165 events) of the total,68% (464 events) is due to the dormant landslides, whereas the residual7% is represented by inactive landslide areas.

The overall analysis shows that most of the landscape is affected bylow-velocity landslides, where the higher landslide density values areconcentrated in the north-eastern (Torrente Inferno drainage basin)and central sectors (4, 16, 17 and 26 drainage basins; Table 2) of thestudy area. Here, the lithological factor, due to the outcropping of ArgilleVaricolori and Corleto Perticara Formations (varicoloured clays andmarls with calcarenites and calcilutites), seems to be the main predis-posing factor of mass movement processes.

The fluvial landforms of the study area are mainly featured by deepincised V-shaped valleys, where incision and fluvial deepening are pre-vailing on the depositional ones. Eluvial, colluvial and slope deposits(breccias and debris fan) are rather rare in the study area and deposi-tional landforms related to recent/present-day geomorphological evo-lution featured only the lower trunks of the thalweg of the mainstreams. In fact, gravel-bed bars occurred along the lower reaches ofTorrente Inferno, Torrente Carpino, Torrente Serrapotamo and FiumaraLa Terra braided rivers, whereas the lowest trunk of the TorrenteCamastra River is characterized by a narrow alluvial plain withsandy bars. Intense processes of gully and rill erosion aremainly locatedin the upper reach of catchments where pelitic rocks crop out andalong the right-side valley of the lower trunk of the Fiumara La TerraRiver.

The drainage basin is composed of the 6th order basin of theTorrente Camastra River and theminor sub-basins of the Torrente Infer-no and Torrente Carpino rivers. It covers an area of about 330 km2. Thestudy area exhibits a medium- to high-drainage density (D, valuesranging from 2.79 to 8.8, average values of 4.6) and its pattern can beclassified as dendritic. The lowest values of D (b3) were found onlyfor the slopes of the Torrente Inferno catchment, in correspondence ofoutcrops of Oligocene–Early Miocene sandstone and marls. The highestvalues of D (N4.5) were observed in the catchments located to the rightside of the Torrente Camastra river, where varicoloured clays widelycrop out, and in small sub-basins (i.e. 28, 29, 30, 31 sub-basins) mainlydraining the fine clastic deposits of the Calvello infill. Moreover, Dvalues higher than 4 were also observed in sub-basins (i.e. 33 and 34),characterized by outcrops of pelitic facies of Lagonegro units (i.e. MtFacito and Galestri Fms).

The analysis of the spatial distribution of D values suggests a strongcorrelation with lithology, since higher values are typical of sub-basinswhere more erodible lithological complexes are present. Anomalousconfluences (i.e. channels of u order not flowing in channels oforder u + 1) are widely diffused in drainage basins of the study area,as inferred by the general high values of the hierarchic anomaly density(i.e. ga, Avena et al., 1967). In fact, ga values span between 1.2 and 24,with an average value of 6.4 (Table 2). As previously described, sub-basins with higher values of ga are featured by a high number of anom-alous confluences and/or by low-order channels flowing into main

Fig. 6. Spatial distribution of the RUSLE/USPED factors in the Camastra drainage basin: A) rainfall erosivity, ‘R-factor’ B) soil erodibility, ‘K-factor’, C) cover-management factor, ‘C-factor’,D) topographic factor, ‘LS-factor’.

332 M. Lazzari et al. / Catena 127 (2015) 323–339

streams (see Avena et al., 1967). Such a hierarchical organization ofcatchments can indicate perturbations disturbing the regular devel-opment of the drainage network (Capolongo et al., 2005; Della Setaet al., 2007; Ribolini and Spagnolo, 2008; Santangelo et al., 2013).In the study area, ga values higher than 10 were observed in severalsub-basins largely affected by mass movements and/or badlanddevelopment. For example, the highest value of ga (about 22, Table 2)was found for the sub-basin 17, which is characterized by a landslidedensity equal to 15.6%. Then, the development of a highly-anomalous(i.e. far with respect to a well-hierarchized drainage network) drainagenetwork could be related to the perturbations induced by mass move-ment processes and/or badland development, as demonstrated bysome authors in other sectors of the Apennine belt (Della Seta et al.,2007).

4.2. RUSLE model

4.2.1. Input parametersThe annual average erosivity factor (Fig. 6a) computed for the

Camastra basin is equal to 793.30 MJ mm ha−1 h−1 yr−1 for the1951–2010 period. The spatial distribution of the annual average

rainfall erosivity varies greatly within the drainage basin area with thelowest value of 512 MJ mm ha−1 h−1 yr−1 and the highest value of1068 MJ mm ha−1 h−1 yr−1; the spatial distribution of the R factorvalues exhibits a gradual decrease moving toward the north-easternsector of the study area.

The soil erodibility factor (K-factor) (Fig. 6b) ranges in theCamastra drainage basin between 0.010 Mg h MJ−1 mm−1 and0.31 Mg h MJ−1 mm−1.

The land use and land management factor (C-factor) in thedrainage basin (Fig. 6c) varies between maximum values assignedto plantations (0.5) andminimum values allocated to areas coveredby densely vegetated forests (0.003). A zero C-factor valuewas assigned to each sealed areas such as paved roads andsettlements.

The terrain analysis shows a prevalence of areas with moderateslopes characterized by average inclinations of 15°. Accordingly,the average topographic LS-factor of the Camastra watershed(Fig. 6d) is low, with an average of 3.49 and a standard deviationof σ = 6.75. Minimum LS-factor values occur in flat areas withsmall catchments and are typically assigned to areas along thedivides.

Fig. 7. Predicted average annual soil erosion for the Camastra watershed (1951–2010) based on RUSLE model. Legend shows annual soil erosion rates in Mg ha−1 yr−1.

333M. Lazzari et al. / Catena 127 (2015) 323–339

4.2.2. RUSLE soil loss predictionThe average soil loss predicted by RUSLE for the Camastra drainage

basin, calculated for a 10 × 10m cell size is equal to 1,677,377.94 Mg yr−1,with an average area-specific soil loss of 51.1 Mg ha−1 yr−1. Thepredicted spatial pattern of gross soil erosion (Fig. 7) is subdividedinto five classes.

Approximately 53.1% of the Camastra drainage basin shows only verylow soil erosion rates (less than 1 Mg ha−1 yr−1), whereas 24.1% ofthe area experiences extreme soil erosion (higher than 40 Mg ha−1 yr−1).

By using a bulk density value of 1.3 g cm−3, the average volume ofsoil loss is 1,290,290.72 m3 yr−1. This value is much higher than themean annual sediment yield of the Camastra dam.

4.2.3. USPEDThe results of the application of theUSPEDmodel are shown in Fig. 8.

Each pixel has been classified using the following nine classes oferosion/deposition:

Extreme erosion (N−40 Mg ha−1 yr−1)High erosion (−20/−40 Mg ha−1 yr−1)Moderate erosion (−10/−20 Mg ha−1 yr−1)Low erosion (−1/−10 Mg ha−1 yr−1)Stable (−1/1 Mg ha−1 yr−1)Low deposition (1/10 Mg ha−1 yr−1)Moderate deposition (10/20 Mg ha−1 yr−1)

High deposition (20/40 Mg ha−1 yr−1)Extreme deposition (N40 Mg ha−1 yr−1).

Analysis of the USPED model indicate that stable and low erosion/deposition areas of the catchments cover 42.6% of the entire territory,corresponding to a surface of about 140 km2. The results also indicatethe diffuse presence of predicted areas affected by extreme erosionand deposition. As a matter of fact, the surface area with values of ±40 Mg ha−1 yr−1 covers, respectively, 19.3 and 19.1 km2 of the totalcatchment area (Fig. 9). The overall erosion predicted by the USPEDmodel is of 325,956.22 Mg yr−1, with an average area-specific soil lossof 9.93 Mg ha−1 yr−1.

By using a bulk density value of 1.3 g cm−3, the average volume ofsoil loss is m3 yr−1. This value is lesser than the mean annual sedimentyield of the Camastra dam of 390000 m3 yr−1, but it could be morereliable than the value predicted with the RUSLE model.

4.3. Tu index

Fig. 10 shows the spatial distribution of the Tu index values withinthe catchments studied. The values of Tu index for the catchmentsrange from 261Mg km−2 yr−1 to 7323Mg km−2 yr−1, with an averagevalue of 1501 Mg km−2 yr−1. Although some small sub-basins exhibit-ed very high or very low values of Tu index, the range interquartile ofthe statistical population is represented by a more narrow interval of539–1630 Mg km−2 yr−1. The low correlation (R2 = 0.08) between

Fig. 8. Predicted average annual soil erosion/deposition for the Camastra watershed (1951–2010) based on USPEDmodel. Legend shows annual soil erosion/deposition rates in Mg ha−1 yr−1.

334 M. Lazzari et al. / Catena 127 (2015) 323–339

Tu index values and the 5th- and 6th-orders catchment area suggest adominant influence of local factors such as lithology, land cover, topog-raphy and landslide occurrence (de Vente et al., 2006; Vanmaerckeet al., 2011) in controlling sediment yield for the catchments of the

Fig. 9. The three histograms show, respectively, the total distribution of the areas for each USPEclasses in the areas interested by linear erosion (rills, gullies and badlands).

study area. Higher values of Tu index were observed in some small andhigh-relief sub-basins (e.g. 12, 15, 28 and 17, Table 2) featured by claydeposits,which are largely affected bymassmovements or the develop-ment of badlands areas. The sum of the Tu index values for the three

D class of reference, the distribution of classes in the landslide areas and the distribution of

Fig. 10.Map showing the spatial distribution of the indirect estimation of sediment yield based on the Ciccacci's empirical relationship (Eq. (1)). a) Box andwhisker plot of the Tu index forthe 34 catchments of the study area.

335M. Lazzari et al. / Catena 127 (2015) 323–339

catchments draining to the Camastra dam is 1392 Mg km−2 yr−1. Thisvalue of the Tu index can be considered as a reasonably proxy forshort-term sediment yield in the drainage basin of the Camastra damif the following assumptions are valid: i) sediment accumulation inthe Camastra dam represents the most part of the sediment yield andii) bedload is only a small amount of the total solid load of the catch-ments studied.

Indeed, geomorphological analysis and field survey suggest that theerosional processes prevail over the depositional ones; moreover, thefine-grained deposits of the dam bottom indicate that bedload is onlya minimum fraction of the total solid load. It goes without saving thatit is difficult to obtain an accurate estimation of the contribution ofbedload to sediment deposition volumes in the Camastra reservoir.The comparison between empirical estimations of Tu index and annualsediment yield coming from reservoir sedimentation can give an indica-tion about the amount of sediments accumulated in the reservoirthrough the bedload processes. The ratio between the total value ofthe Tu index for the catchment (1379 Mg km−2 yr−1) and the meanannual sediment storage in reservoir (1615 Mg km−2 yr−1) is 0.86and it could suggest that the bedload is below 20% of the solid load.

5. Discussion

Empirical estimation of soil erosion and sediment yield of a drainagebasin draining an artificial reservoir have been performed using theRUSLE and USPED models and an indirect estimation of suspendedsediment yield based on geomorphological parameters.

The validation of the predictive ability of these different empiricalmodels has been verified on the basis of a long-term analysis of thesedimentation rates in reservoir. Of course, this estimate includes anumber of uncertainties and approximations, such as trapping efficien-cy, lateral variation of the grain-size reservoir sediments, possible errorsin bathymetric surveys and interpolation procedures of raw data.Snyder et al. (2004) provided a valuable pilot study of the Englebrightdam, California aimed to the quantification of error sources in evalua-tion of reservoir sediment storage. Although the sediment amount ofthe Englebright dam is extremely well-constrained, the Authors dem-onstrated that the basic data useful to the estimation of sedimentreservoir are affected by significant uncertainties. Nevertheless, theseuncertainties introduce relativelyminor errors on the overall estimationof sediment storage, especially when one considers a long-term history

Table 3Table of comparison between the values of sediment yield calculated from Tu, USPED and RUSLE methods and the landslide density values.

Basin Order Area Landslide area (km2) Landslide density (%) SSY from Rusle (T ha−2 yr−1) SSY from USPED (T ha−2 yr−1) Tu index (T ha−2 yr−1)

1 5 53.00 12.18 22.98 25.66 11.65 2.692 4 15.39 1.03 6.68 19.57 8.39 3.153 6 259.86 24.22 9.32 21.85 9.66 8.244 5 8.65 1.61 18.61 33.46 19.25 17.735 5 26.5 1.84 6.94 13.01 6.06 7.006 5 7.95 0.71 8.94 18.70 7.34 7.067 5 92.12 8.61 9.34 24.38 9.95 6.948 5 39.17 3.894 9.94 18.68 8.77 9.049 4 12.99 1.91 14.70 19.30 9.18 13.3710 4 14.16 0.98 6.95 12.79 5.37 6.9111 4 7.05 0.02 0.29 9.21 7.49 3.4712 4 0.37 0 0 11.01 2.45 56.2713 4 7.27 0.71 9.77 19.38 7.80 5.7514 4 39.98 3.00 7.51 18.18 7.73 4.2715 4 1.38 0.23 16.37 26.39 24.28 44.5416 4 1.94 0.80 41.02 39.44 21.54 10.1517 4 3.66 0.57 15.57 25.37 16.54 33.7718 4 26.56 1.43 5.38 22.47 7.86 5.7219 4 3.65 0.89 24.49 33.51 15.17 3.0020 4 5.04 1.55 30.73 20.76 6.17 2.3321 4 12.84 1.98 15.44 21.83 9.71 2.5322 4 15.17 4.69 30.93 23.30 11.18 5.3323 4 1.61 0.11 7.09 12.11 8.31 12.1724 4 7.05 0.35 4.92 13.30 4.10 9.1925 4 4.22 0.26 6.21 10.10 6.01 20.6526 4 2.23 0.34 15.20 24.68 15.64 10.0527 4 8.21 0.61 7.42 19.22 12.50 5.3128 4 1.45 0.01 0.69 27.99 10.57 31.8929 4 1.43 0 0 23.73 11.85 25.8730 4 1.59 0 0 13.80 6.99 15.7131 4 1.34 0 0 12.15 6.95 9.1332 4 20.91 2.68 12.83 12.94 6.81 3.3433 4 8.75 1.89 21.55 24.45 8.03 13.8234 4 6.05 0.06 1.02 47.28 17.41 13.69

336 M. Lazzari et al. / Catena 127 (2015) 323–339

of the dam (Snyder et al., 2004). In the Camastra dam, the data comingfrom extensive bathymetric surveys and the associated coring cam-paigns are certainly affected by similar analytical errors which arehard to quantify, but we can reasonably argue that their introduce alittle error in the mean annual estimation of sediment rate based on a38-year long history of the reservoir.

Mean value of soil loss estimated by RUSLE model is1,290,290 m3 yr−1. Adopting an averaged bulk density of therocks outcropping within the catchment of 1.3 g cm−3, specific sedi-ment yield has an unlikely high value of 5110 Mg km−2 yr−1(Table 3).In first analysis, the USPED model provided a more reasonable estima-tion of erosion/denudation processes (993 Mg km−2 yr−1), althoughthe reservoir sedimentation data suggests that themean annual deposi-tion is of 1615 Mg km−2 yr−1. On the other hand, the sum of the Tu

Fig. 11. Histogram representing the comparison a

index values for the three catchments draining to the Camastra dam is1392 Mg km−2 yr−1. The comparison of sediment storage in theCamastra reservoir and the empirical estimation of the mean annualsuspended sediment yield from the Tu index (1615 Mg km−2 yr−1

and 1379 Mg km−2 yr−1, respectively) suggests that the Tu empiricalmethod is the erosion model with the best predictive ability in thestudy area. The good prediction ability of this kind of approach is alsobased on the assumption of high values of the Sediment DeliveryRatio (SDR). The spatial distribution of fluvial and mass movementlandforms, the analysis of the distribution of sedimentary bodies alongthe main channels of the catchment and the grain-size analysis of de-posits accumulated in the reservoir suggest that the erosionalprocesses appear to be dominant with regard to the depositional oneswithin the catchments studied. The estimation of sediment yield from

mong the three methods for each subbasin.

Fig. 12. Graphs show the correlation between the Tu index and USPED values withlandslide density.

337M. Lazzari et al. / Catena 127 (2015) 323–339

Tu index does not include the amount of bedload either. On the otherhand, the evaluation of geomorphological and depositional processesacting in the catchments and grain-size analysis of sediments accumu-lated into the reservoir suggest that the channels bedload is only a littleamount of the sediment load.

What is the reason of such a good prediction ability of the verysimple Ciccacci's empiricalmethod?Despite it being limitedly represen-tative in terms of spatial and temporal distribution of the datasets, themethods are still now the only ones available with systematic measure-ment carried out in a good number of Italian drainage basins. In otherwords, the good efficiency of this simple empirical model may be dueto the fact that these equations are calibrated in the Mediterraneanareas characterized by fluvial processes mainly acting on catchmentswith geological and geomorphological features similar to the studyarea, such as medium to high drainage density and outcrops of terrige-nous deposits.

The good agreement between the empirical estimations of Tu indexand the annual sediment yield coming from reservoir sedimentationsuggests that: i) the empirical estimation of Tu index values can beconsidered as a reasonable proxy for short-term sediment yield in thedrainage basin of the Camastra dam; ii) the suspended load is themost relevant part of the sediment accumulated in the Camastrareservoir.

Our analyses suggest that the USPEDmethod can reasonably furnisha valuable quantitative estimation of the erosion–deposition processesat catchment and sub-basin scale. In particular, the USPED erosionmodel can provide additional information on the spatial distributionof sediment yield sectors of the drainage basin where erosion is moredeveloped and the role of landslide processes on sediment yield ismore significant.

Fig. 11 shows the average values of RUSLE, USPED and Tu index foreach sub-basin of 4th, 5th and 6th order; it can be noted that in thebasins of 5th and 6th order, as well in the basins of 4th order of greatersurface extension, themeanvalues from Tu index andUSPEDmethod forthe subbasins of the study area are very close and comparable, whereasthe sediment yield coming from RUSLE method is much higher. On theother hand, several small sub-basins located to the north-eastern sectorsof the study area (i.e. Torrente Inferno basin, sub-basins 1, 19, 20, 21, 22,

Fig. 2), shows a huge difference between the mean values of thesediment yield from Tu index and USPED method. These sub-basinsare characterized by the highest landslide density values (Figs. 3and 11) and by a prevailing areal erosion than a linear one.

According to other studies of sediment yield based on the compari-son between RUSLE and WATER/SEDEM empirical models and sedi-ment storage in artificial reservoir (Van Rompaey et al., 2005; deVente et al., 2006), the Italian drainage basins did not exhibit a negativerelation of the sediment delivery ratio with the catchment area. Indeed,the study area does not show the expected negative correlationbetween the specific sediment yield coming from. Tu index and USPEDmodel and the drainage basin area, because of other factors, such astopography, lithology, land use and landslides, already invoked bydifferent authors as the local-scale factor controls of sediment yield(de Vente et al., 2006; Vanmaercke et al., 2011). In particular, differentAuthors point out the strong contribution of mass movements, rill andgullies as point sources of sediment (Casali et al., 1999; Osterkampand Toy, 1997; Poesen et al., 2003; Van Rompaey et al., 2005; deVente et al., 2006) and their important role to the total sediment yield.

Relationships between sediment yield based on Tu index and RUSLEmethod and topographic factors do not show significant statisticaltrends for the catchments of the study area. Tu index values are higherfor small sub-basins (b4 km2) with a low degree of hierarchical organi-zation (ga N 13, Table 2) and largely affected by mass movement pro-cesses. Higher values of Tu index have been observed in some smallsub-basins (e.g. 12, 15, 28 and 17, Table 2) featured by clay deposits,which are largely affected by mass movements or the development ofbadlands areas. Fig. 12b shows the statistical relationships betweenlandslide density and Tu index values for some catchments of thestudy area, characterized by homogenous lithology (i.e. clay deposits,see Fig. 1) and similar geomorphological features. A linear correlationbetween Tu index values and landslide density (r2 = 0.54, Fig. 12b)can be observed in these sub-basins mainly characterized by outcropsof clay deposits. By analyzing the wider subbasins of the study area(i.e. subbasins with an area larger than 10 km2), a direct correlationbetween the sediment yield from USPED method and landslide densityis also observed (Fig. 12a). These observations point out an importantrole of landslides as local factors in controlling the catchment-scalesediment yield of the study area.

The analysis of the distribution of erosion/deposition classes in land-slides areas (Fig. 9) can furnish additional information about the areaswhere an interaction of drainage network and landslide processes ex-ists, thus permitting to investigate the problem of hillslope-channelconnectivity and the role of landslides as point of sources of sediments.Almost all the landslide areas fall in the low-erosion and extreme depo-sition classes; this observation could suggest that landslides have an im-portant role in sediment mobilization but did not produce a significantincrease in sediment yield of the study area. In other words, landslidesdid not move the sediment outside the catchments but only provideda source of sediments for fluvial processes.

Apart from the inferred important contribution of landslides as sed-iment sources, the geomorphological analysis, the spatial distribution ofhigher classes of soil erosion of the USPED method and the Tu indexpoint out that the erosion hot-spots of the study area are located incatchments with a high drainage density, steep channels as well asintense gully and rill erosion.

6. Final remarks

The collected dataset represents a basic tool both for the investiga-tion of the morpho-dynamics of a typical mountain catchment of theMediterranean area and the evaluation of sediment budget related tofluvial and hillslope processes. The 38-year-long record of the sedimentstorage in the Camastra artificial reservoir located at the outlet of thestudied mountain catchments permits the validation of empirical rela-tionships based on the geomorphic and climatic parameters of the

338 M. Lazzari et al. / Catena 127 (2015) 323–339

drainage basin. Among the three different methods of sediment yieldevaluation (Tu index, RUSLE and USPED), Tu index showed the bestprediction ability, although USPED erosion model also furnished agood estimation (1615 Mg km−2 yr−1 and 995 Mg km−2 yr−1,respectively).

As also confirmed by other works (Massaro et al., 1997; Della Setaet al., 2007, 2009; Gioia et al., 2011a,2014; Borrelli et al., 2014), theindirect estimation of sediment yield based on the Ciccacci's empiricalrelationships can represent a good proxy of short-term denudationrates in the Mediterranean areas with geological and geomorphologicalfeatures similar to the study area but the use of empirical models withincreasing complexity such as the USPED erosion model can help toexplore the spatial distribution of the sediment yield sectors of drainagebasin, the erosion hot-spots, and the role of landslides in sedimentmobilization and hillslope-channel connectivity processes.

Our results emphasize the role of landslides as point sources ofsediment whereas the sectors of higher sediment yields are generallydistributed along slopes with a high drainage density, steep channelsand intense gully and rill erosion.

Finally, the results here presented can also be useful to solve practi-cal problems, such as burial velocity of the artificial reservoir located atthe outlet of the catchments analyzed aswell as implement strategies tocontrol sediment dynamics at the catchment scale. The validation ofsuch evaluation of sediment yield at the scale of large catchments,using sediment storage in reservoirs, can provide also a good opportuni-ty: i) to test the reliability of the empirical methods used to estimate thesediment yield; ii) to investigate the catchment dynamics and its spatialand temporal evolution in terms of erosion, transport and deposition;and iii) to improve the prevision of the dam activity in function of theerosion/deposition rates.

Acknowledgments

We are grateful to Prof. M. Del Monte and an anonymous referee fortheir useful and constructive suggestions to improve the manuscriptand native speaker Marcella D'Amico for the accurate English revision.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.catena.2014.11.021.

References

Amato, A., Cinque, A., 1992. Il bacino plio-pleistocenico di Calvello (Potenza): evoluzionegeologica e geomorfologica. Studi Geol. Camerti 1, 181–189.

Amato, A., Cinque, A., 1999. Erosional land surfaces of the Campano–Lucano Apennines (S.Italy): genesis, evolution, and tectonic implications. Tectonophysics 315, 251–267.

Avena, G.C., Giuliano, G., Lupia Palmieri, E., 1967. Sulla valutazione quantitativa dellagerarchizzazione ed evoluzione dei reticoli fluviali. Boll. Soc. Geol. Ital. 86, 781–796.

Boardman, J., 2006. Soil erosion science: reflections on the limitations of currentapproaches. Catena 68, 73–86.

Bonini, M., Sani, F., 2000. Pliocene-quaternary transpressional evolution of the Anzi-Calvello and Northern S. Arcangelo basins (Basilicata, Southern Apennines, Italy) asa consequence of deep-seated fault reactivation. Mar. Pet. Geol. 17, 909–927.

Borrelli, P., Märker, M., Panagos, P., Schütt, B., 2014. Modeling soil erosion and riversediment yield for an intermountain drainage basin of the Central Apennines, Italy.Catena 114, 45–58.

Brown, A.G., et al., 2009. From sedimentary records to sediment budgets: multipleapproaches to catchment sediment flux. Geomorphology 108, 35–47.

Capolongo, D., Cecaro, G., Giano, S.I., Lazzari, M., Schiattarella, M., 2005. Structural controlon drainage network of the south-western side of the Agri River upper valley (South-ern Apennines, Italy). Geogr. Fis. Din. Quat. 28, 169–180.

Capolongo, D., Pennetta, L., Piccarreta, M., Fallacara, G., Boenzi, F., 2008a. Spatial andtemporal variations in soil erosion and deposition due to land-levelling in a semi-arid area of Basilicata (Southern Italy). Earth Surf. Process. Landf. 33, 364–379.

Capolongo, D., Diodato, N., Mannaerts, C.M., Piccarreta, M., Strobl, R.O., 2008b. Analyzingtemporal changes in climate erosivity using a simplified rainfall erosivity model inBasilicata (southern Italy). J. Hydrol. 356, 119–130.

Casali, J., Laburu, A., Lopez, J.J., Garcia, R., 1999. Digital terrain modelling of drainagechannel erosion. J. Agric. Eng. Res. 74, 421–426.

Chisci, G., Bazzoffi, P., 1995. Fruttiviticoltura di collina, limitazione dell'erosione edell'inquinamento. Interventi agrobiologici. Agricoltura 10, 41–45.

Ciccacci, S., Fredi, F., Lupia Palmieri, E., Pugliese, F., 1980. Contributo dell'analisigeomorfica quantitativa alla valutazione dell'entita dell'erosione nei bacini fluviali.Boll. Soc. Geol. Ital. 99, 455–516.

Cinque, A., Patacca, E., Scandone, P., Tozzi, M., 1993. Quaternary kinematic evolution ofthe southern Apennines. Relationships between surface geological features anddeep lithospheric structures. Ann. Geofis. 36, 249–260.

Conforti, M., Pascale, S., Pepe, M., Sdao, F., Sole, A., 2013. Denudation processes andlandforms map of the Camastra River catchment (Basilicata — South Italy). J. Maps9, 444–455.

Conoscenti, C., et al., 2013. A GIS-based approach for gully erosion susceptibility model-ling: a test in Sicily, Italy. Environ. Earth Sci. 70, 1179–1195.

Cruden, D.M., Varnes, D.J., 1996. Landslide types and processes. In: Turner, A.K., Shuster,R.L. (Eds.), Landslides: Investigation and Mitigation. National Academies Press,Washington DC, pp. 36–75.

de Vente, J., Poesen, J., 2005. Predicting soil erosion and sediment yield at the basin scale:scale issues and semi-quantitative models. Earth Sci. Rev. 71, 95–125.

de Vente, J., Poesen, J., Bazzoffi, P., Van Rompaey, A., Verstraeten, G., 2006. Predictingcatchment sediment yield in Mediterranean environments: the importance ofsediment sources and connectivity in Italian drainage basins. Earth Surf. Process.Landf. 31, 1017–1034.

Della Seta, M., Del Monte, M., Fredi, P., Palmieri, E.L., 2007. Direct and indirect evaluationof denudation rates in Central Italy. Catena 71, 21–30.

Della Seta, M., Del Monte, M., Fredi, P., Lupia Palmieri, E., 2009. Space-time variability ofdenudation rates at the catchment and hillslope scales on the Tyrrhenian side ofCentral Italy. Geomorphology 107, 161–177.

Fagnano, M., Diodato, N., Alberico, I., Fiorentino, N., 2012. An overview of soil erosionmodelling compatible with RUSLE approach. Rend. Lincei 23, 69–80.

Gioia, D., Martino, C., Schiattarella, M., 2011a. Long- to short-term denudation rates in thesouthern Apennines: geomorphological markers and chronological constraints. Geol.Carpath. 62, 27–41.

Gioia, D., Schiattarella, M., Mattei, M., Nico, G., 2011b. Quantitativemorphotectonics of thePliocene to Quaternary Auletta basin, southern Italy. Geomorphology 134, 326–343.

Gioia, D., Gallicchio, S., Moretti, M., Schiattarella, M., 2014. Landscape response to tectonicand climatic forcing in the foredeep of the southern Apennines, Italy: insights fromQuaternary stratigraphy, quantitative geomorphic analysis, and denudation rateproxies. Earth Surf. Process. Landf. 39, 814–835.

Hutchinson, M.F., 1989. A new procedure for gridding elevation and stream line data withautomatic removal of spurious pits. J. Hydrol. 106, 211–232.

Jansen, I.M.L., Painter, R.B., 1974. Predicting sediment yield from climate and topography.J. Hydrol. 21, 371–380.

Jones, R.J.A., Hiederer, R., Rusco, E., Montanarella, L., 2005. Estimating organic carbon inthe soils of Europe for policy support. Eur. J. Soil Sci. 56, 655–671.

Lasaponara, R., Lanorte, A., Pignatti, S., 2006. Multiscale fuel type mapping in fragmentedecosystems: preliminary results from hyperspectral MIVIS and multispectral LandsatTM data. Int. J. Remote Sens. 27, 587–593.

Lazzari, M., 2008. Il comportamento tettonico e sedimentario del bacino d'avanfossaBradanica durante il Pleistocene inferiore. Mem. Descr. Carta Geol. LXXVII, 61–76.

Lazzari, M., 2011. Note illustrative della Carta Inventario delle Frane della Basilicatacentroccidentale. Editore Zaccara, Lagonegro, Potenza, Italy.

Lazzari, M., Geraldi, E., Lapenna, V., Loperte, A., 2006. Natural hazards vs human impact:an integrated methodological approach in geomorphological risk assessment on theTursi historical site, Southern Italy. Landslides 3, 275–287.

Malinverno, A., Ryan, W.B.F., 1986. Extension in the Tyrrhenian Sea and shortening in theApennines as result of arc migration driven by sinking of the lithosphere. Tectonics 5,227–245.

Massaro, M.E., Russo, M., Zuppetta, A., 1997. Analisi indiretta dell'entita dell'erosionenel bacino del Fiume Tammaro (Appennino Campano). Geogr. Fis. Din. Quat. 19,381–394.

Menardi Noguera, A., Rea, G., 2000. Deep structure of the Campanian–Lucanian Arc(Southern Apennine, Italy). Tectonophysics 324, 239–265.

Merritt, W.S., Letcher, R.A., Jakeman, A.J., 2003. A review of erosion and sediment trans-port models. Environ. Model Softw. 18, 761–799.

Mitasova, H., Hofierka, J., Zlocha, M., Iverson, L.R., 1996. Modelling topographic potentialfor erosion and deposition using GIS. Int. J. Geogr. Inf. Syst. 10, 629–641.

Molino, B., Viparelli, R., De Vincenzo, A., 2007. Effects of river network works and soilconservation measures on reservoir siltation. Int. J. Sediment Res. 22, 273–281.

Mulder, T., Syvitski, J.P.M., 1996. Climatic and morphologic relationships of rivers:implications of sea-level fluctuations on river loads. J. Geol. 104, 509–523.

Onori, F., De Bonis, P., Grauso, S., 2006. Soil erosion prediction at the basin scale using therevised universal soil loss equation (RUSLE) in a catchment of Sicily (southern Italy).Environ. Geol. 50, 1129–1140.

Osterkamp, W.R., Toy, T.J., 1997. Geomorphic considerations for erosion prediction.Environ. Geol. 29, 152–157.

Pescatore, T., Renda, P., Schiattarella, M., Tramutoli, M., 1999. Stratigraphic and structuralrelationships between Meso-Cenozoic Lagonegro basin and coeval carbonateplatforms in southern Apennines, Italy. Tectonophysics 315, 269–286.

Piccarreta, M., Pasini, A., Capolongo, D., Lazzari, M., 2013. Changes in daily precipitationextremes in the Mediterranean from 1951 to 2010: the Basilicata region, southernItaly. Int. J. Climatol. 33, 3229–3248.

Piccarreta, M., Lazzari, M., Pasini, A., 2014. Trends in daily temperature extremes over theBasilicata region (southern Italy) from 1951 to 2010 in a Mediterranean climaticcontext. Int. J. Climatol. http://dx.doi.org/10.1002/joc.4101 (in press).

Poesen, J., Nachtergaele, J., Verstraeten, G., Valentin, C., 2003. Gully erosion and environ-mental change: importance and research needs. Catena 50, 91–133.

339M. Lazzari et al. / Catena 127 (2015) 323–339

Renard, K.G., Foster, G.R.,Weesies, G.A., Porter, J.P., 1991. RUSLE: revised universal soil lossequation. J. Soil Water Conserv. 46, 30–33.

Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., Yoder, D.C., 1997. Predicting SoilErosion by Water: A Guide to Conservation Planning with the Revised UniversalSoil Loss Equation (RUSLE) USDA Agr Handb. No 703.

Riaño, D., Chuvieco, E., Salas, J., Palacios-Orueta, A., Bastarrika, A., 2002. Generation of fueltype maps from Landsat TM images and ancillary data in Mediterranean ecosystems.Can. J. For. Res. 32, 1301–1315.

Ribolini, A., Spagnolo, M., 2008. Drainage network geometry versus tectonics in theArgentera Massif (French-Italian Alps). Geomorphology 93, 253–266.

Santangelo, M., Gioia, D., Cardinali, M., Guzzetti, F., Schiattarella, M., 2013. Interplay be-tween mass movement and fluvial network organization: an example from southernApennines, Italy. Geomorphology 188, 54–67.

Schiattarella, M., Leo, P.D., Beneduce, P., Giano, S.I., Martino, C., 2006. Tectonically drivenexhumation of a young orogen: an example from the southern Apennines, Italy. In:Willett, S.D., Hovius, N., Brandon, M.T., Fisher, D. (Eds.), Tectonics, Climate, andLandscape Evolution. Special Paper of the Geological Society of America 398,pp. 371–385.

Schiattarella, M., Giano, S.I., Gioia, D., Martino, C., Nico, G., 2013. Age and statistical prop-erties of the summit palaeosurface of southern Italy. Geogr. Fis. Din. Quat. 36,289–302.

Shimabukuro, Y.E., Smith, A., 1991. The least-squares mixing models to generate fractionimages derived from remote sensing multispectral data. IEEE Trans. Geosci. RemoteSens. 29, 16–20.

Snyder, N.P., et al., 2004. Estimating accumulation rates and physical properties of sedi-ment behind a dam: Englebright Lake, Yuba River, northern California. Water Resour.Res. 40, W11301.

Strahler, A.N., 1957. Quantitative analysis of watershed geomorphology. EOS Trans. Am.Geophys. Union 38, 913–920.

Van Rompaey, A., Bazzoffi, P., Jones, R.J.A., Montanarella, L., 2005. Modeling sedimentyields in Italian catchments. Geomorphology 65, 157–169.

Vanmaercke, M., Poesen, J., Verstraeten, G., de Vente, J., Ocakoglu, F., 2011. Sediment yieldin Europe: spatial patterns and scale dependency. Geomorphology 130, 142–161.

Vergari, F., Della Seta, M., Del Monte, M., Fredi, P., Palmieri, E.L., 2013. Long- and short-term evolution of several Mediterranean denudation hot spots: the role of rainfallvariations and human impact. Geomorphology 183, 14–27.

Wischmeier, W.H., Smith, D.D., 1978. Predicting Rainfall Erosion Losses—A Guide toConservation Planning.

Zanchi, C., Giordani, C., 1995. Elementi di conservazione del suolo. Pàtron Editore,Bologna, Italy.