Disaster Risk Reduction Based on a GIS Case Study of the Čađavica River Watershed

(1) University of Belgrade, Faculty of Forestry, Department of Ecological Engineering for Soil and Water Resources Protection, Kneza Višeslava 1, RS-11000 Belgrade, Serbia; (2) University of Belgrade, Faculty of Forestry, Department of Landscape Architecture and Horticulture, Kneza Višeslava 1, RS-11000 Belgrade, Serbia; (3) Institute for Nature Conservation of Serbia, Dr Ivana Ribara 91, RS-11070 Belgrade, Serbia

100 SEEFOR 8 (2): 99-106 The torrential (flash) flood represents a sudden appearance of maximal discharge in a torrent bed with a high concentration of sediment.The torrential watershed is a hydrographic entity which involves the bed of the mainstream and its tributaries, and the gravitating surfaces with erosion processes of a certain intensity.The attribute "torrential" refers to any watershed with a sudden appearance of maximal discharge with a high concentration of sediment, regardless of the size and category of the stream [7].Climate, specific relief characteristics, distinctions of the soil and vegetation cover and social and economic conditions cause the occurrence of torrential floods as one of the consequences of the existing erosion processes.
It is very important to raise public awareness of the threats of flooding and promote a wise use of watersheds [8], combining environmental protection and flood management as factors of similar importance [9].Destructive erosion processes [10][11][12] and torrential floods cannot be prevented.However, a better understanding of the processes and scientific methodologies for their prediction can help mitigate their impact [13].In most cases, torrential floods are caused by natural incidents (such as climatic and morphohydrographic particularities of watersheds), but the human factor contributes significantly to the effects of disasters (the mismanagement of forest and agricultural surfaces, uncontrolled urbanization and the absence of erosion control and flood protection structures).Inadequate dimensions of protective structures are commonly the initial cause of their damage or destruction, which significantly increases the intensity of torrential floods.Therefore, hydraulic and hydrological computations should be based on reliable input data (precipitation, land use, hydrographic characteristics and runoff curve number).
Representative examples are the torrential floods in Western Serbia, particularly in the Municipality of Krupanj, covering a territory of 342 km².Local watersheds received a three-day rainfall ranging from 180 to 420 mm, while the absolute daily maximal precipitation amounted to 218 mm.A few settlements were struck by floods on local torrents on May 15 th 2014, causing the deaths of two people, almost 900 hectares of flooded arable land or damage by landslides, 333 flooded buildings (of which 40 severely damaged or destroyed), 120 km of destroyed or damaged roads, 14 destroyed and 8 damaged bridges, 5 km of destroyed river regulations and 300 evacuated inhabitants.In addition, 269 landslides were activated during the propagation of heavy precipitation and flood waves.The estimated material damage amounted to over 30 million €.A total of four protected surfaces with areas ranging from 0.03 to 6.73 ha were endangered (three monuments of nature and one nature reserve).
The values of the maximal discharge, area sediment yields and sediment transport, are the basic input data for the design and dimensioning of ETCS (Erosion and Torrent Control Structures) such as check-dams, overflows, regulations, contour ditches and channels, silt-filtering stripes and wattle works.In May 2014, during the torrential floods, numerous river regulations, check-dams, cascades, and culverts did not have a sufficient capacity for maximal discharge and sediment, which caused their obturation, damaging and destruction.A GIS-based flood reconstruction was carried out, with a recalculation of maximal discharges (using data on the maximal daily precipitation in May 2014), area sediment yields and sediment transport.The corrected results of the calculations will be used as the basic input data for ETCS dimensioning, both for the reconstructed structures and the new ones.
This paper presents the results of calculations of the maximal discharge, area sediment yields and sediment transport in the experimental watershed of the Čađavica River, using GIS processing of two digital elevation models (DEMs) with different spatial resolution.

StUDY aRea
The experimental watershed of the Čađavica River is located in Western Serbia, in the Municipality of Krupanj, with the outlet profile in the center of the city (Figure 1).The watershed is built from schists and sandstones, with layers of phyllite and argillaceous schist [14].The dominant soil is Dystric Cambisol with a light mechanical composition, medium porosity and good aeration [15].The soil profile is shallow, with good infiltration and poor retention capacity, due to the high percentage of sand.

MetHODOlOGY
Spatial analysis was carried out by processing of the DEM of 20 m (hereinafter referred to as DEM 20 ) and 90 m (hereinafter referred to as DEM 90 ) resolutions using software ArcMap 10.3 and its extension 3D Analyst.In addition, analyses concerning watershed and stream network delineation were performed using ArcHydro Tools in Arc Map.DEM 20 was generated using scanned topographic maps (scale 1:25000) and vectorized isolines as primary spatial elements for the triangulated irregular network (TIN) database creation and later conversion to a 20 m raster resolution.DEM 90 was derived from Shuttle Radar Topography Mission (SRTM).The land use analysis for DEM 20 was performed using 2014 orthophoto with a 1 m resolution.The land use analysis for DEM 90 resolution was performed using the CORINE database [16].The determination of hydrographic characteristics was performed with the ArcHydro® model [17], which is often used for creating hydrological information systems on the basis of geospatial and temporal information about water resources [18].ArcHydro ® was developed as an extension of ArcGIS software, which is suitable for the delineation of watershed boundaries [17].DEM is a necessary input data for spatial analysis and could be generated using different techniques such as photogrammetry [19,20], interferometry [21], laser scanning [22] and topographic surveys [23].
The factors dominating the formation of torrential floods were analyzed, such as natural characteristics (hydrographic characteristics, soil and geological conditions) and human impact (land use structure, the relation between surfaces with low and high water infiltration-retention capacity).Land use analysis was based on the field investigations, orthophoto, the CORINE (COoRdination of INformation on the Environment) database, topographic, geological and soil maps.Land use classification was based of the CORINE methodology [16].Area sediment yields and the intensity of erosion processes were calculated using the "Erosion Potential Method" (EPM).This method was created, developed and calibrated at the Faculty of Forestry of the University of Belgrade and at The Jaroslav Černi Institute for the Development of Water Resources in Belgrade [24,25].The method is still in use in all countries that originate from former Yugoslavia.The application of this method is based on the calculation of the basic parameters: the coefficient of erosion Z, sediment yields and sediment transport: R u -sediment delivery ratio, P -perimeter of the watershed (km) L -the length of the watershed (km) A md -medium altitude difference of the watershed (km) The method is based on the analytical processing of data on factors affecting erosion.The erosion spatial phenomenon appears on the map according to the classification based on the analytically calculated erosion coefficient (Z), which does not depend on climate, but on soil characteristics, vegetation cover, relief and visible representation of erosion.The coefficient of erosion (Z) is obtained from the following expression [24]: Y -coefficient of soil resistance to erosion X•a -the land use coefficient, ϕ -coefficient of the observed erosion process (takes into consideration clearly visible erosion processes), I m -mean slope of terrain The computations of maximal discharges (for control profile CP, Figure 1) were performed using a method combining the synthetic unit hydrograph (maximum ordinate of unit runoff q max ) and Soil Conservation Service [26] methodologies (deriving effective rainfall Pe from total precipitation Pb).This combined method is the most frequently used procedure for the computation of maximal discharges in unstudied watersheds in Serbia.The computations were performed for AMC III (Antecedent Moisture Conditions III-high content of water in the soil and significantly reduced infiltration capacity).Synthetic triangular unit hydrographs were transformed to synthetic (computed) curvilinear hydrographs using the SCS basic dimensionless hydrograph [27].The computations of maximal discharges were performed using the regional analysis of lag time [28], the internal daily distribution of precipitation [29] and the classification of soil hydrologic groups for CN-runoff curve number determination [30].

ReSUltS
The main hydrographic characteristics of the experimental watershed are presented in Table 1.

land Use
Land use was determined using DEM 20 and DEM 90 with a structure presented in Table 2 and Figure 2.

erosion and Sediment transport
The result of the area sediment yields and sediment transport calculations based on using different DEMs resolutions (DEM 20 , DEM 90 ) are presented in Table 3, as well as the representative values of the coefficient of erosion Z.
W a -annual yields of erosive material; W asp -specific annual yields of erosive material; W at -annual transport of sediment through the hydrographic network; W atsp -specific annual transport of sediment through the hydrographic network; W abls -annual amount of bed load sediment; W ass -annual amount of suspended sediment.
The spatial distribution of the erosion coefficient Z is presented in Figure 3

Hydrological Conditions
Maximal discharges (Q max1% ) were computed using a combined method based on designed precipitation Pbr 24h(1%) =113.8 mm.The hydrographs of maximal discharges (Q maxDEM20_1% ; Q maxDEM90_1% ) are presented in Figure 4. Some characteristic outputs of hydrologic computations are presented in Table 5 (unit runoff q max ; CN -runoff curve number; Pbr -total precipitation; Pe -effective rain).

DISCUSSIOn
Destructive erosion processes and torrential floods endanger the life security of the population and material goods, while they also have environmental and social impacts.Current climate fluctuations (precipitation, air temperature extremes, droughts) associated with anthropogenic impacts (urbanization, forest fires, land degradation) provoke intensive erosion processes and a frequent occurrence of torrential floods.
The experimental watershed of the Čađavica River was analyzed using GIS processing of two DEMs with different spatial resolution (20 m (DEM 20 ) and 90 m (DEM 90 ) resolution), which produced differences in hydrographic characteristics, land use, and the runoff curve number.It also affected the values of maximal discharges, area sediment yields, and sediment transport.Among the hydrographic characteristics, the most expressive one is the difference in Smt (mean slope of terrain): S mtDEM20 =33.24% and S mtDEM90 =21.14%.Unlike DEM 90 , DEM 20 m recognized some specific land uses such as degraded areas and degraded forests.
The actual state of erosion processes is marked with the representative Z values of Z DEM20 =0.31 (dominant weak erosion -deep processes) and Z DEM90 =0.25 (dominant weak erosionmixed surface and deep processes).Consequently, the annual yields of the erosive material amount to W aDEM20 =12367.5 m 3 and W aDEM90 =9005.1 m 3 , with a specific annual transport of sediment through the hydrographic network of W atspDEM20 =292.5 m 3 •km -2 •year -1 and W atspDEM90 =219.3 m 3 •km - 2 •year -1 .DEM 20 registered excessive erosion and larger surfaces under medium and strong erosion than DEM 90 .
The runoff curve number values CN DEM20 =84 and CN DEM90 =79 have an impact on the computed maximal discharges Q maxDEM20_1% =75.06 m 3 •s -1 and Q maxDEM90_1% =63.84 m 3 •s -1 .In addition to that, the volume of the computed hydrograph of direct runoff W DEM90_1% =1.125 .10 6 m 3 is significantly reduced in comparison to the volume of direct runoff W DEM20_1% =1.33 .10 6 m 3 .
A decrease in the DEM resolution (DEM 90 in comparison to DEM 20 ) leads to a loss of detailed topographic characteristics such as mean altitude, slope steepness and area [31,32].
Field work was carried out to determine the accuracy of the spatial analysis using different DEMs resolutions (DEM 20 and DEM 90 ), especially for land use and the erosion map.DEM 20 m recognized degraded areas and degraded forests, as well as surfaces under excessive erosion processes, which was not possible when DEM 90 was used.The higher accuracy of DEM 20 enabled a more precise identification of the zones which were the sources of erosive material production and generation of surface runoff.Consequently, the results of the computations of area sediment yields and transport and maximal discharge on the basis of DEM 20 were significantly higher.Since they are the basic input data for the dimensioning of ETCS in the torrent bed and on watershed slopes, these higher results caused the design of structures with larger dimensions and higher construction costs, but also an elevated level of security.In addition, DEM 20 recognized small protected areas (0.03-6.73 ha), which were almost "invisible" when DEM 90 was used.

COnClUSIOn
The values of the maximal discharge, area sediment yields, and sediment transport are the basic input data for the design and dimensioning of protective structures in torrential beds and on watershed slopes.GIS applications and their tools offer an effective spatial analysis of the watershed with a precise determination of hydrographic characteristics, land use, land use changes and runoff curve number, as parameters of great importance for the final values of the maximal discharge, area sediment yields, and sediment transport.This requires a careful approach in accordance with some specific conditions at torrential watersheds, including the steepness of slopes of the terrain and the torrent bed, intensive erosion processes, favorable conditions for fast surface runoff formation and transport of huge quantities of sediment.The usage of nonrepresentative input data produces inadequate results of computations and poor subsequent dimensioning of protective structures.As a result, the insufficient capacity for maximal discharge and sediment leads to obturation, damage, and destruction of these structures.The higher accuracy of DEM enables a more precise identification of the "source" zones of erosive material production and generation of surface runoff.That was confirmed by this investigation, where the usage of DEM 20 resolution produced a more "realistic" picture of the experimental watershed than the usage of DEM 90 .The results of computations of area sediment yields and transport and maximal discharge on the basis of DEM 20 m are significantly higher, which affects the dimensions of ETCS in the torrent bed and on watershed slopes, the costs of their construction and the achieved level of security.In addition to other measures, the reduction of flood risk is based on the construction of effective and welldimensioned structures, with a capacity that is sufficient for maximal discharge and sediment.An adequate GIS approach can help in the precise evaluation of the factors affecting the generation of destructive erosion processes and torrential floods in order to provide effective erosion control and torrential flood protection in endangered watersheds.

fIGURe 1 .
fIGURe 1. Location of the experimental watershed of the Čađavica River.

fIGURe 4 .
fIGURe 4. Hydrographs of maximal discharge for AMC III (Antecedent Moisture Conditions III -high content of water in the soil and significantly reduced infiltration capacity).

taBle 1 .
(DEM 20 ; DEM 90 ), while the structure of erosion categories is presented in Table4.Main hydrographic characteristics of the Čađavica River watershed.

taBle 2 .
Land use in the Čađavica River watershed.

taBle 3 .
Characteristic outputs of computations of sediment yields and transport.

taBle 4 .
Structure of erosion categories.

taBle 5 .
Structure of erosion categories.