Ivica MILEVSKI (Macedonia), Slavoljub DRAGICEVIC (Serbia), Aleksandra GEORGIEVSKA (Macedonia) Introduction
In this work, one approach of GIS and RS assessment of potential natural hazard areas (excess erosion, landslides, flash floods and fires) is presented.
Pehchevo Municipality in the easternmost part of the Republic of Macedonia is selected as a case study area.
Reasons: high local impact of natural hazards on the environment, social-demographic situation and local economy. WHY?
Natural hazards represent significant economic,




environmental and social problem; Losses from natural disasters in Macedonia are estimated at several million euro per year; Excess erosion result in vast loss of fertile land, reduced production etc; Landslides lead to traffic disruption, damage to structures, and even casualties; With climate change, fires are more and more frequent, larger and cause massive damage; Spring and autumn floods reach high-damage effects, and summer drought can lead to destruction of agricultural production, water management problems and so on. Examples in Macedonia Examples in Pehchevo municipality Examples in Pehchevo municipality WHY PEHCHEVO MUNICIPALITY? PEHCHEVO MUNICIPALITY AS A CASE STUDY AREA IS AFFECTED BY SEVERAL ACTUAL AND POTENTIAL NATURAL RISKS AND DISASTERS:
EXCESS EROSION
LANDSLIDING
FLASH FLOODS
HIGH SEISMICITY
FOREST FIRES Methodology Most relevant static factors for each type of natural hazard are selected (topography, land cover, anthropogenic objects and infrastructure); 2. With GIS and satellite imagery, multi-layer calculation is performed based on available traditional equations, clustering or discreditation procedures. In such way suitable relatively “static” natural hazard maps (models) are produced; 3. Then, dynamic (mostly climate related) factors are included in previous models resulting in appropriate scenarios correlated with different amounts of precipitation, temperature, wind direction etc. 1. Methodology
Finally, GIS-based scenarios are evaluated and tested with field check or very fine (0.5-1.0 m) resolution Google Earth imagery showing good accuracy.
Further development of such GIS models in connection with automatic remote meteorological stations and dynamic satellite imagery (like MODIS) will provide on-time warning for coming natural hazard avoiding potential damages or even casualities. METHODOLOGY FLOWCHART Used data
25k digital topographic map of the State Agency of







Cadastre (6 sections) including 5 m Digital Elevation Model (DEM-SAC), 100k General Geologic Map (raster), section Delchevo with explanatory note (1974), 50k General Geologic Map (raster), section Delchevo with explanatory note (2009), 100k Topographic Map (raster), section Berovo, 100m Corine Land Cover (CLC2006) raster and vector model, Landsat MSS, TM, ETM+ satellite imagery full spectral range (1977-2012), Soil map from the Monograph Malesh and Pijanets, MASA, 1984, Google Earth satellite maps and data extracted from, Other maps and data from the Monograph Malesh and Pijanets, MASA, 1984 etc. Digital elevation models and satellite data Vector data with database Rivers Roads Geology Raster data-preprocessed: slopes, aspects, CLC2006 Convergence index (profile curvature), Terrain relief, Lithology Soils, Temperatures, Precipitations Vector and raster data overlay and processing Rocks waterproof Slopes Settlements Large construction objects GIS ANALYSIS Main road buffer River buffer DATA PROCESSING AND MODELLING FINAL MODEL OF POTENTIAL LANDSLIDE AREAS Example: Erosion risk model
EPM Gavrilovic (1972) model with Z erosion risk coeff.
Z = Y ∙ Xa ∙ (φ + sqrtJ0.5), where:
Y – is soil and rock erodibility coefficient, ranging from 0.1 (very resistant) to 2.0 (unresistant soils and rocks);
X*a – is land cover index, ranging from 0 (water bodies and 0.2 for forest areas) to 1.0 (bare uncovered soils);
φ – is coefficient of visible erosion processes ranging from 0 (no visible erosion) to 1 (severely eroded landscape);
J – is mean catchment slope in % as a decimal value.
GIS-calibrated coefficient Z is calculated according to the equation:
Z=sqrt(sqrt(Y))*φ*((X*a+φ)*log(a+1)+sqrt(a/57.3)) Erosion risk model according to the Z value range in Gavrilovic model Erosion risk scenarios Rains: 30 mm 60 mm 90 mm with H value in Gavrilovic model: H=(Hy*(Hd/(Hy/6))) Annual erosion intensity based on GIS modified Gavrilovic approach (EPM) GIS addopted EPM Scenarios with rains of 10 мм 20 мм 30 мм 40 мм 50 мм 60 мм 70 мм 80 мм 90 мм 100 мм Potential risk of erosion with daily raining of 40 mm Landslide modellling
Heuristic models
Clustering classification models
Statistical models
Statistical Index Method (SIM; Model)
Landslide susceptibility analysis (LSA)
Other statistical models In this research, 6 causative factors were considered: • Slope (DEM), • Lithology (Digital Map), • Land use (CLC2006), • Distance from streams (100 m buffer), • Distance from roads (50 m buffer) and • Curvature (profile curvature) DEM. LSI=A+B+C+D+E+F Another 4 factors are usually part of the lanslide zonation models: • Aspect (DEM), • Relative relief (DEM), • Distance from faults (buffer), • Elevation (DEM) and • Seismic zone (buffer), However, most analyses shows its insignificant or inconsistent influence on landslide processes. FACTOR Lithology Clastic sediments Schists Gneiss Granitoides Quartzlatites Slopes 0-5° 5-10° 10-30° 30-50° >50° Convergence (curvature) Highly concave Concave Flat Convex Highly convex Land Cover Dense forests Transitional forests Pastures Cultivated lands Urban areas Bare rocks Streams 0-100 m >100 m Roads 0-50 m >50 m VALUE 5 4 3 2 1 2 6 10 8 4 0.5 2 1.5 1 0.5 1 2 4 3 3 4 1.5 0 1.5 0 DIGITAL THEMATIC MAPS Combination of landslide factors and its weightening values according to the significance Landslide susceptibility model of Pehchevo Municipality Potential landslide area scenarios Rains: 30 mm 60 mm 90 mm Rains 30 mm 60 mm 90 mm Total Area km2 19.8 50.1 88.6 208.5 % 9.5 24.0 42.5 100.0 Landslide risk scenarios in town of Pehchevo on different rain occurrences. Landslide risk near Crnik village with different daily rain occurrences(10-100 мм). Potential flood areas modelling using rellated:
Index of vegetation (Vi) according to CLC 2006 model with




values from 0.1 for the forest area with high density to 1.0 for uncovered area. Land cover coefficient (Li), Landsat ETM+ band 3 (red) is used shown in value from 0 (dense vegetation) to 1 (bare soils). Catchment area (Ca), extracted in hydrology module (Flow Tracing) with square meters units. Topographic Wetness Index (TWI) which show topographic tendency of water retention. Slope height (Sh) which indicated relative altitude above valley bottoms in meters and the potential width of floodplain. Fa =Vi*Li*log (Ca)*log(TWI)*2/(Sh). Potentially floodable areas in Pehchevo municipality using: 15m DEM CLC2006 Landsat ETM+ Flood areas scenarios in regard to rains 30 mm 60 mm 90 mm Rains 30 mm 60 mm 90 mm Total Area km2 2.7 5.1 12.1 208.5 % 1.3 2.4 5.8 100.0 Forest fire risk modelling
For the production of the forest fire risk map, five fire rating




classes are used according to Ertena et al., 2004. These classes are formed according to: Slope (DEM), Aspects (DEM), Vegetation type (CLC2006)-VT, Distance from roads (DR) and settlements (DS)(buffers).
RC = 7*VT + 5*(S+A) + 3*(DR+DS)
Finally, based on these analysis carried out, a fire risk zone map was produced. Forest fire risk map (model) of Pehchevo municipality according to the Ertena et al. 2004, approach Forest fire risk scenarios 1. Hot weather 2. Hot and dry weather 3. Very hot and very dry weather Scenario 1 2 3 Total Area km2 % 21.0 10.1 44.2 21.2 74.7 35.8 208.5 100.0 Evaluation of models
From field research and GPS records
From detailed orthophoto image (0.5 m) analyses
Informations of local peoples about the past hazards and its extent
Old maps, documents and bibliography
Models showed significant accuracy, especially for excess erosion, landslides and floods (85-97%)
Further improvement will make them even more accurate. Combined map of potential natural hazard areas Hazard Excess eros. Landslides Floods Fires Total Munic. area Area km2 34.0 19.8 12.1 21.0 71.9 208.5 % 16.3 9.5 5.8 10.1 34.4 100.0 Conclusion
GIS and RS-tools with good approach and algorithm provide very good results in potential natural hazard modelling;
Because of lack of measured meteorological, hydrological and other relevant data in Pehchevo area, there are some shifts, uncertainties and probably errors in the potential natural hazard models;
For much better accuracy, at least one automatic meteorological gauge (remotely connected with GIS desktops) must be installed in Pehchevo.
GIS models with geo-database must be permanently maintained and updated by responsible persons continuously offering relevant and fresh data. Conclusion
Because natural hazards do not known borders, broader connection in wider regional cross-border area is necessary bringing the higher level of quality with greater usability to the municipalities.
The ultimate goal is to keep attention on areas vulnerable to natural risks with as high as possible protection of the environment.
In turn, there will be very positive economic and social demographic feedback for the local communities. THANK YOU FOR ATTENTION
Acknowledgment: This paper is result of the project Joint Applicable Research for Natural Recourses Preservation and Environmental Protection in the Cross border Region within the Municipalities of Pehchevo and Simitli, co-financed by European Union through IPA Cross-border Programme CCI 2007CB16IPO007