Compilation of novel and renewed, goal oriented digital soil maps using geostatistical and data mining tools 49Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. Compilation of novel and renewed, goal oriented digital soil maps using geostatistical and data mining tools László PÁSZTOR1, Annamária LABORCZI1, Katalin TAKÁCS1, Gábor SZATMÁRI2, Endre DOBOS3, Gábor ILLÉS4, Zsófia BAKACSI1 and József SZABÓ1 DOI: 10.15201/hungeobull.64.1.5 Hungarian Geographical Bulletin 64 2015 (1) 49–64. 1 Institute for Soil Science and Agricultural Chemistry, Centre for Agricultural Research, H-1022 Budapest, Herman Ott ó út 15. E-mails: pasztor.laszlo@agrar.mta.hu, laborczi.annamaria@agrar.mta.hu, takacs.katalin@agrar.mta.hu, bakacsi.zsofi a@agrar.mta.hu, szabo.jozsef@agrar.mta.hu 2 Department of Physical Geography and Geoinformatics, University of Szeged, H-6722 Szeged, Egyetem u. 2–6. E-mail: szatmari.gabor.88@gmail.com 3 Department of Physical Geography and Environmental Sciences, University of Miskolc, H-3515 Miskolc- Egyetemváros, E-mail: ecodobos@uni-miskolc.hu 4 Forest Research Institute, National Agricultural Research and Innovation Centre, H-9600 Sárvár, Várkerület 30/a. E-mail: illesg@erti.hu Abstract Due to former soil surveys and mapping activities signifi cant amount of soil information has accumulated in Hungary. Present soil data requirements are mainly fulfi lled with these available datasets either by their direct usage or aft er certain specifi c and generally fortuitous, thematic and/or spatial inference. Due to the more and more frequently emerging discrepancies between the available and the expected data, there might be notable imperfection as for the accuracy and reliability of the delivered products. With a recently started project we would like to signifi cantly extend the potential, how soil information requirements could be satis- fi ed in Hungary. We started to compile digital soil maps, which fulfi l optimally the national and international demands from points of view of thematic, spatial and temporal accuracy. In addition to the auxiliary, spatial data themes related to soil forming factors and/or to indicative environmental elements we heavily lean on the various national soil databases. The set of the applied digital soil mapping techniques is gradually broadened incorporating and eventually integrating geostatistical, data mining and GIS tools. Regression kriging has been used for the spatial inference of certain quantitative data, like particle size distribution com- ponents, rootable depth and organic matt er content. Classifi cation and regression trees were applied for the understanding of the soil-landscape models involved in existing soil maps, and for the post-formalization of survey/compilation rules. The relationships identifi ed and expressed in decision rules made the compilation of spatially refi ned category-type soil maps (like genetic soil type and soil productivity maps) possible with the aid of high resolution environmental auxiliary variables. In our paper, we give a short introduction to soil mapping and information management concentrating on the driving forces for the renewal of soil spatial data infrastructure provided by the framework of Digital Soil Mapping. The fi rst results of DOSoReMI.hu (Digital, Optimized, Soil Related Maps and Information in Hungary) project are presented in the form of brand new national and regional soil maps. Keywords: classifi cation and regression trees, digital soil mapping, regression kriging, spatial soil information Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.50 Introduction Demands on spatial soil information Demands on soil related information have been signifi cant worldwide and are still in- creasing (Bullock, P. 1999; Mermut, A.R. and Eswaran, H. 2000; Tóth, G. et al. 2008; Sanchez, P.A. et al. 2009; Baumgardner, M.F. 2011). Recent requests oft en do not refer to primary or even secondary soil properties, but to various processes, functions, services and/or systems related to soils (Omuto, C. et al. 2013). Soil maps were typically used for a long time to satisfy these needs. Due to the rela- tively high costs of new data collection and the spreading of Geographic Information technology, Spatial Soil Information Systems (SSISs) and Digital Soil Mapping (DSM), these approaches have taken over the role of traditional soil maps in the fi eld of data service. Nevertheless, legacy soil data are still heavily relied on, as they include an abundance of information exploitable by proper methodology in GIS/SSIS/DSM en- vironment. Not only the degree but also the nature of current needs for soil information has changed. Traditionally focus was on the agricultural functions of soils, which was also refl ected in the methodology of data collec- tion and mapping. Recently the information related to addi- tional soil functions is becoming identically important (Blum, W.E.H. 2005; Panagos, P. et al. 2012). This types of information require- ment generally cannot be fulfi lled with new data collections, at least not on such a level as in the frame of traditional soil surveys (Montanarella, L. 2010). As a consequence of these issues the framework of spatial soil information service has also altered signifi - cantly (Figure 1). Main issues of soil mapping The goal of soil mapping is to reveal and visualize the spatial relationships of the the- matic knowledge related to soil cover. Soil maps are thematic maps, where theme is determined by some specific information related to soils. This can be a primary or secondary (derived) soil property or class as well as any knowledge characterizing func- tions, processes or services of soils (Pásztor, L. et al. 2014). The greatest and inevitable challenge of the compilation of soil maps is the regionaliza- tion of the local knowledge, its spatial infer- ence (Várallyay, Gy. 2012). Reconnaissance of specifi c soil properties is carried out by sam- pling, which provides defi nitely point-like information. To create maps, the data related Fig. 1. Framework of spatial soil information services. Dashed line: information sources; dott ed line: data fl ow/transportation between various elements 51Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. framework for this representation. It should be remarked, there are also compromised ap- proaches between the two concepts (like cer- tain fuzzy methods suitable for soil mapping; McBratney, A.B. and Odeh, I.O.A. 1997). Each soil map can be characterized by three basic aspects which are more or less inter-re- lated. A map displays a theme, a regionalized soil (or more generally soil related) property expressed by either quantitatively or qualita- tively using categories (thematic issues). The map is compiled for a geographic region in a predefi ned scale, with some spatial resolution (geometrical issues). Finally the map has an overall and also spatially variable accuracy, purity, reliability (uncertainty issues). A de- mand on at least a tiny change in any of these issues theoretically induces the compilation of a new map with the required parameters. In traditional soil mapping the creation of a new map was troublesome and laborious. As a consequence robust maps were elaborated and rather the demands were fi tt ed to the available map products. Formation of digital soil mapping A soil map is an object specifi c spatial model of the soil cover, whose compilation is domi- nated by the consideration of soil forming processes (Böhner, J. et al. 2002). There have been signifi cant and essentially concurrent changes concerning three central elements of this defi nition. The growing and spread of digital soil mapping in the last decade can be att ributed to the eff ects of these changes (Dobos, E. et al. 2006; Lagacherie, P. et al. 2007; Lagacherie, P. 2008; Boettinger, J.L. et al. 2010). Spatial and at the same time dig- ital (that is GIS conform) information related to various segments of soil formation proc- esses has become available in more and more quantity, with bett er and bett er spatial reso- lution and on lower and lower costs. Mathematical (geo)statistical and data min- ing methods have been developed, which are effi ciently applicable in the lack of determinis- tic models for the quantifi cation of the some- to locations should be spatially inferred using appropriate methods. From a certain point of view, the development of soil mapping is the conscious expansion of the repository of these methods: from mental space usage, along (base)map delimitation based on soil- landscape models till the various (mechani- cal, geometrical, geostatistical) interpolation methods and further until the introduction of ancillary, environmental, spatial data as auxiliary co-variables related to various com- ponents of soil forming processes. Sampling based mapping is inherently predictive, the value or class of the mapped variable can only be estimated at unvisited locations (Gessler, P.E. et al. 1995; Scull, P. et al. 2003). Spatial prediction can be carried out (i) taking exclusively the mapped vari- able into consideration based on its spatial features; (ii) also based on the mapped vari- able, but the constraints of spatial validity are provided by further spatial, ancillary infor- mation; (iii) in every predicted locations sup- ported by environmental, auxiliary co-vari- ables (McKenzie, N.J. and Ryan, P.J. 1999). Basically there are two, virtually confl ict- ing but rather complementary conceptions for the description of the spatial heterogene- ity of soils (Heuvelink, G.B.M. and Webster, R. 2001). One builds on similarity and is basi- cally object based. It represents the soil cover with soil patches, which are either homoge- neous mapping units or aggregates with es- timated composition. The map realization of this concept is the traditional crisp soil map. According to the inherent model of these maps, at the given spatial resolution the soil properties within the mapping units are ei- ther homogeneous or heterogeneous but in cartographically unmappable way; and there is discontinuity in the mapped soil feature at the borders (Dobos, E. and Hengl, T. 2009; Szabó, J. et al. 2011). The other approach emphasizes the con- tinuous spatial variation of soil properties. The mapped soil property is predicted in cells and the spatial resolution is determined by the cell size (Mark, D.M. and Csillag, F. 1989). Raster data models of GIS provide ideal Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.52 times really complex and indirect relation- ships between soil features and the formerly mentioned, so called environmental auxiliary co-variables. Originally these methods were elaborated for the treatment of substantially diff erent specialties, but they proved to be well adaptable in soil mapping, too. Along the globalization processes the sig- nifi cant inhomogeneity in the knowledge of the world’s soil cover has become evident. In one hand this has induced the compilation of relatively reliable soil maps based on limited soil data on the majority of the world, this way achieving at least a minimal coverage of these regions with spatial soil information. On the other hand it has initiated the elaboration of the principles of unifi cation. The former surveys and mappings were carried out on national level based on independent meth- odologies, which caused disturbing eff ects in the mapping of the geographically continu- ously varying soil cover along administrative borders showing artifi cial disrupt changes. The framework of DSM (McBratney, A.B. et al. 2003; Lagacherie, P. and McBratney, A.B. 2007; Hartemink, A.E. et al. 2008) in- volves spatial inference of the information collected at sampled points based on ancil- lary environmental variables related to soil forming processes (Figure 2). DSM is formal- ized by the so called SCORPAN equation: Sproperty or class = f (S, C, O, R, P, A, N), where on the left side Sproperty or class is the (either numerical or categorical) mapped soil fea- ture, while on the right side the predictive soil forming factors are Climate, Organisms, Relief, Parent material, Age and Geographic position. An original but well-established feature of the SCORPAN approach as op- posed to Jenny’s formula of soil formation (Jenny, H. 1941), that it also takes further Soil related spatial information into considera- tion in the spatial prediction of a given soil variable. The most commonly used spatial auxiliary data layers are terrain att ributes Fig. 2. Concept of digital soil mapping 53Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. derived from digital elevation models, and spectral reflectance bands from satellite imagery. Furthermore f refers to a specifi c function relating to mapped property with the actually used predictors, which may be realized in various forms. Predictive mapping, taking exclusively the mapped variable into consideration, is nu- merically realized by spatial interpolation, which is supported by Tobler's First Law of Geography (Tobler, W. 1970). It states "Everything is related to everything else, but near things are more related than dis- tant things” and it is basically an analogous formulation of the concept of spatial autocor- relation. The main feature of these types of methods, they operate in the geographical space. The branch of various interpolation methods dominated by stochastic modelling of environmental features is geostatistics. If prediction is supported by environmen- tal auxiliary co-variables, the quantifi cation of the relationship between the mapped soil parameter and the ancillary data is the main challenge. Generalized classifi cation, that is data mining methods proved to be suitable for the solution of these types of tasks. These methods investigate essentially the feature space, analysing its structure, thus unfold- ing the hidden and/or complex relationships. Regression and classifi cation trees, random forests, neural networks, Bayesian belief net- work, support vector machines and some more techniques were successfully tested. There are also compromised approaches between the concepts which concentrate purely on geographical or feature space. The two most widely used are co-kriging and re- gression kriging. In co-kriging a more dense- ly sampled ancillary parameter supports the interpolation as opposed to ordinary krig- ing. In regression kriging the variation of the mapped variable is subdivided into two parts: the trend is estimated by MLRA and the residual of the explained part is then kriged (Hengl, T. et al. 2004). Due to the simultaneous richness of spatial inference methods and the potentially avail- able auxiliary environmental information (Grunwald, S. 2009; Hengl, T. 2009; Mulder, V.L. et al. 2011), there is a high versatility of possible approaches for the compilation of a given soil (related) map. The framework of digital soil mapping also provides opportunity for the elaboration of goal specifi c soil maps, since the parameters characterizing the map product (thematic, resolution, accuracy, reliability etc.) may be predefi ned. The activity of DSM goes beyond mapping purely primary and secondary soil properties, the regionalization of further lev- els of soil related features (processes, func- tions and services) is also targeted (Minasny, B. et al. 2012). Spatial soil information in Hungary Hungary has long traditions in soil survey and mapping. Large amount of soil informa- tion is available in various dimensions and generally presented in maps, serving diff er- ent purposes as to spatial and/or thematic aspects (Várallyay, Gy. 2012). Increasing pro- portion of soil related data has been digitally processed and organized into various spatial soil information systems (Pásztor, L. et al. 2013a). The existing maps, data and systems served the society for many years, however the available data are no longer fully satis- factory for the recent needs of policy mak- ing. There were numerous initiatives for the digital processing, completion, improvement and integration of the existing soil datasets. Presently soil data requirements are ful- fi lled with the recently available datasets ei- ther by their direct usage or aft er certain spe- cifi c and generally fortuitous, thematic and/or spatial inference (Szabó, J. et al. 2007; Dobos, E. et al. 2010; Szatmári, G. et al. 2013; Sisák, I. and Benő, A. 2014; Waltner, I. et al. 2014). Due to the frequent discrepancies between the available and the expected data, notable imperfection may occur in the accuracy and reliability of the delivered products. A recently started project (DOSoReMI.hu: Digital, Optimized, Soil Related Maps and Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.54 Fig. 3. Framework of the DOSoReMi.hu project Information in Hungary; Figure 3) aims to signifi cantly extend the potential how soil information requirements could be satisfi ed in Hungary. In the frame of our project hitherto we have carried out spatial and thematic data mining of a signifi cant amount of soil related infor- mation available in the form of legacy soil data as well as digital databases and spatial soil information systems (Pásztor, L. et al. 2013b, 2014). In the course of the analyses auxiliary, spa- tial data themes related to soil forming fac- tors as well as indicative environmental ele- ments are relied on. Our objective is to com- pile digital soil related maps that optimally fulfi l the national and international demands from points of view of thematic, spatial and temporal accuracy. In the following we short- ly present some developments achieved so far in the frame of our activities. Materials and methods Digital mapping of soil properties in Zala County Impact assessment of the forecasted climate change and the analysis of the possibilities of the adaptation in the agriculture and for- estry can be supported by scenario based land management modelling, whose results can be incorporated in spatial planning. This framework requires adequate and spatially detailed knowledge of the soil cover. For the satisfaction of these demands in Zala County (3,784 km²; Hungary), the soil conditions of the agricultural areas were digitally mapped based on the most detailed, available recent and legacy soil data. The agri-environmental conditions were characterized according to the 1:10,000 scale genetic soil mapping meth- odology and the category system applied in 55Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. the Hungarian soil-agricultural chemistry practice. The factors constraining the fertility of soils were featured according to the bio- physical criteria system elaborated for the delimitation of naturally handicapped areas in the EU. Production related soil functions were regionalized incorporating agro-mete- orological modelling. Various soil related information were mapped in three distinct sets: (i) basic soil properties determining agri-environmen- tal conditions (soil type according to the Hungarian genetic classifi cation, rootable depth, sand and clay content for the 1st and 2nd soil layers, pH, OM and carbonate content for the plough layer); (ii) biophysical criteria of natural handicaps defi ned by common European system and (iii) agro-meteoro- logically modelled yield values for diff erent crops, meteorological and management sce- narios. The applied method(s) for the spatial inference of specifi c themes was/were suit- ably selected: regression and classifi cation trees for categorical data, indicator kriging for probabilistic management of criterion in- formation; and typically regression kriging for quantitative data. The appropriate derivatives of a 20 m dig- ital elevation model were used in the analy- sis. Multitemporal MODIS products were selected from the period of 2009–2011 repre- senting diff erent parts of the growing season and years with various climatic conditions. Additionally two climatic data layers (mean annual precipitation and mean temperature of summer months), the 1:100,000 Geological Map of Hungary (Gyalog, L. and Síkhegyi, F. 2005) and the map of groundwater depth pre- pared by Water Research Institute (VITUKI, 2005) were used as auxiliary environmental co-variables. Disaggregating category type soil maps Numerous formerly elaborated thematic soil maps are not available in Hungary in the recently required scale. The original maps were compiled (i) in analogue environment and (ii) applying hardly identifi able soil-land- scape models and unrecorded rules, so their reproducibility is problematic. Their theme, however, represents a widely used, embed- ded information source, which is expected to be (re)produced in larger scales. Various pos- sibilities were studied for the solution of the problem. Decision trees proved to be adequate data mining technique to improve the spatial resolution of category-type soil maps disag- gregating their soil mapping units (SMUs). The agro-ecological units in the AGROTOPO (1994) database, compiled as a result of a substantial scientifi c synthesiz- ing work (Várallyay, Gy. et al. 1985), were elaborated dominantly on the basis of map- ping units originating from Kreybig soil maps (Kreybig, L. 1937), applying appro- priate spatial and thematic generalization. Consequently, the Kreybig patt ern contains signifi cant and potentially utilizable infor- mation on the heterogeneity of these agro- ecological units, as do the elevation models characterizing the relief features. Digital Kreybig Soil Information System is a countywide SSIS, which synthesizes the full soil information collected and processed during the Kreybig survey (Pásztor, L. et al. 2010, 2012). The readiness of AGROTOPO and DKSIS spatial soil information systems together with appropriate Digital Elevation Models and further environmental ancillary data available for the whole country has huge potential, which can be exploited in an integrated manner for the disaggregation of the thematic soil layers stored exclusively by AGROTOPO. The new maps display the same thematic but with increased spatial resolution and accuracy. Compilation of country-wide physical soil property maps The increasing demands on spatial soil in- formation in order to support environmental related and land use management decisions vigorously concern physical soil properties, which also played important role in tradi- Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.56 tional soil mapping. Physical soil properties are directly related to water-holding capac- ity and nutrient supply, they aff ect water in- fi ltration, runoff , and movement within the soil (Várallyay, Gy. 2011; Tóth, B. et al. 2014; Farkas, Cs. et al. 2014). Soils can be characterized by diff erent phys- ical soil parameters; one of the most widely used is particle size distribution (Rajkai, K. and Kabos, S. 1999; Nemes, A. et al. 2011). Particles according to their size are catego- rized as clay, silt or sand. The size intervals are defi ned by national or international tex- tural classifi cation systems. The relative mass percentages of sand, silt, and clay in the soil constitute textural classes, which are also specifi ed miscellaneously in various national and/or specialty systems. The most commonly used is the classifi cation system of the United States Department of Agriculture (USDA 1987, Shirazi, M.A. and Boersma, L. 1984). Soil texture information classifi ed according to USDA system is essential input data in (agri-)meteorological and hydrological mod- elling (Vereecken, H. et al. 1989; Saxton, K.E. and Rawls, W.J. 2006), which are also widely used in Hungary (Kozma, Zs. 2012; Ács, F. et al. 2014; Fodor, N. et al. 2014). Our work for producing the very fi rst texture class map according to USDA classifi cation for Hungary has been recently presented in detail by Laborczi, A. et al. (2015). In addition to tex- ture, particle size fractions (clay, silt and sand content) are also important in themselves. They are mandatory variables of the GlobalSoilMap data structure according to its Specifi cations (2014) as well as main indicators used by bio- physical criteria to defi ne natural constraints for agriculture in Europe (Van Orshoven, J. et al. 2013). Firstly clay, silt and sand content were independently predicted spatially using regression kriging with a predefi ned, 150 me- ter spatial resolution. Reference data have orig- inated from the Hungarian Soil Information and Monitoring System. Auxiliary spatial in- formation was represented by digital elevation model and its derived components, geological, climatic, landuse maps and last but not least the physical property SMU layer of DKSIS. The applied geostatistical and data mining tools In the framework of DSM numerous dig- ital mapping methods have been elaborated that apply on of or integrate geostatistical and data mining tools (Goovaerts, P. 2000; Hengl, T. 2009; Moran, C.J. and Bui, E.N. 2002; Lagacherie, P. et al. 2007; Boettinger, J.L. et al. 2010; Hartemink, A.E. et al. 2008). Here only three of them are shortly dis- cussed, which were used in the works pre- sented in this paper. Regression kriging (RK) is a spatial predic- tion technique, which jointly employs correla- tion with auxiliary maps and spatial correla- tion. It is widely used for the spatial inference of quantitative soil properties (e.g. Hengl, T. et al. 2004; Illés, G. et al. 2011; Szatmári, G. and Barta, K. 2013). Similarly to other DSM methods, RK is based on the application of auxiliary environmental variables (deriva- tives of digital elevation model (DEM), re- motely sensed images, etc.), which can be widely interpreted, that is spatial informa- tion on independent soil features can also be involved. In RK fi rstly correlation of the envi- ronmental factors and the predicted variable is determined by MLRA. Then kriging of the residuals provides the stochastic factor which is added to the regression result thus produc- ing the fi nal map. Essentially, RK respects the fact; neither environmental correlation nor geostatistical interpolation alone is able to account for the whole spatial variation that is to produce map products with satisfactory accuracy. They can be used as complementary spatial inference approaches where one can improve the other’s drawbacks. Indicator kriging (IK) is a nonparametric in- terpolation method without any assumption on concerning the distribution of the mod- elled variables providing estimation of prob- ability. Based on these features IK is a useful tool for the spatial inference of categorical variables. Regionalization of specifi c simple and/or simplifi ed secondary and functional soil (related) data can be supported by IK. IK was heavily based on in the process of the delineation of areas aff ected by natural 57Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. constrains in Hungary defi ned by common European biophysical criteria related to soil. The elaborated European system consists of detailed defi nitions, justifi cation and asso- ciated critical limits or threshold values for each biophysical criterion. The fulfi lment of a specifi c criterion had to be regionalized, that is the fi nal product should have to be a binary map displaying yes/no categories. Decisions carried out on soil profi le resulted in binary (indicator) form, which had to be spatially ex- tended. As a consequence, indicator kriging proved to be proper approach. It provided probability (spatial) distribution maps, indi- cating the probability of fulfi lling the criteria within the block used for the calculation. Classifi cation and regression trees (CART) are also widely applied in Digital Soil Mapping too (Moran, C.J. and Bui, E.N. 2002; Scull, P. et al. 2005; Bou Kheir, R. et al. 2010; Giasson, E. et al. 2011; Greve, M.H. et al. 2012), due to their manifold advantages. CART is one of the most successful, wide- spread and effi cient data mining techniques for supervised classifi cation learning, which builds complex relations by a sequence of sim- ple decisions. The tree is built from a training database by recursive method. The decision rules can be easily interpreted; they start with a single node, and then look for the binary distinction which gives the most information on the classifi cation. At each node the condi- tions (based on homogeneity indices) split the reference data into two child nodes. Each of the resulting new nodes is taken and the proc- ess is repeated continuing the recursion until reaching certain predefi ned stopping criterion. CART is easy to interpret and discuss, when a mix of continuous and categorical type envi- ronmental parameters are used as predictors, furthermore, they have excellent predictive capabilities (Breiman, L. 2001; Lawrence, R. et al. 2004; Henderson, B.L. et al. 2005). CART can be applied for the understand- ing of the soil-landscape models involved in existing soil maps, and for the post-for- malization of survey/compilation rules. The relationships identified and expressed in decision rules make the creation of spatially refi ned maps possible with the aid of high resolution environmental auxiliary variables. Among these co-variables, a special role could be played by larger scale soil informa- tion with diverse att ributes. Results Digital mapping of soil properties in Zala County Some results of our activities for the map- ping of soil properties in Zala County are discussed in recent papers. Szatmári, G. et al. (2013) debate in details the experiences gathered during the application of RK in spatial inference of quantitative soil prop- erties. The two main findings, which we also want to emphasize here, referred to the application of ancillary data. In one hand, usage of various co-variables may result in remarkable diff erences of the fi nal map. On the other hand inclusion of spatial soil data signifi cantly improves the performance of RK. Illés, G. et al. (2014) have presented the elaboration of a unifi ed large scale soil type map according to the Hungarian genetic soil classifi cation system. The reference soil data originated from various legacy datasets with diff ering density provided for areas with diff erent land use. Some further, formerly unpublished soil property maps compiled for the agricultural areas of the country are presented in Figure 4. The results of the agro-meteorological modelling for the regionalization of produc- tion related soil functions are scheduled to be published soon. Disaggregating category type soil maps The disaggregation of categorical soil maps with the aid of auxiliary spatial soil informa- tion was successfully applied in cases with diff erent thematic and spatial extent. Some results have been recently presented in detail by Pásztor, L. et al. (2013b). The most use- ful product has been the spatially refi ned Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.58 nationwide soil productivity map, which was operationally used for the delineation of Areas with Excellent Productivity in the framework of the National Regional Devel- opment Plan. The other challenge has been the charac- terization of the soil cover in terms of genet- ic soil types at a scale of 1:50,000–1:25,000, which is required due to various purposes, like the digital implementation of large-scale mapping methodology designed for irriga- tion planning or spatial planning activities. For the fulfi lling of these demands the ge- netic soil type layer of AGROTOPO was dis- aggregated for pilot areas based on DKSIS, environmental auxiliary variables and using decision trees. The downscaled soil type map according to the Hungarian soil classifi cation system compiled for the Danube–Tisza Interfl uve is presented in Figure 5. The map was com- piled with the aid of decision trees using Fig. 4. Basic soil property maps of Zala County 59Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. non-soil environmental auxiliary variables together with the SMUs of DKSIS. Virtual reference point sets were created by multi- ple point sampling. One point per km2 was randomized in the geographical space, rep- resenting virtual sampling locations. Conditional generalization was ap- plied, prescribing a minimum spacing of 100 meters (equal to the cell size applied in spatial modelling) between the generated points. Randomized points closer than 100 meters to SMU borders were eliminated to avoid transition zones between neighbour- ing soil types. The values of the dependent (predicted) and independent (predictor) vari- ables were identifi ed at the randomized lo- cations and their records were used in data mining classifi cation. The rules established during the building of the classifi cation tree were applied to the spatial layers as opera- tions providing soil type prediction for the whole area of interest. The randomization process was repeated 100 times providing 100 classifi cation results for each cell. The fi nal categorization was done by maximum- likelihood decision; the most frequent class was att ributed to the cell as most likely soil type. The vagueness of the classifi cation is also inferred by the occurrence value of the most frequent class. Figure 5 contains an in- set map, which displays the reliability of the spatial prediction expressed this way. Fig. 5. Predicted soil type map for the Danube–Tisza Interfl uve (on the left ) and estimated reliability of the spatial prediction (on the right) Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.60 Compilation of country-wide physical soil property maps Maps of particle size classes (clay, silt and sand content) have been multi-mapped, that is the same target variable has been pre- dicted in various ways. Either the method or the predictor variables have been selected in diff ering way. The purpose has been to identify the best performing constellation. The performance of the various approaches can be identifi ed by proper validation and in our case it has been measured by three vali- dation parameters: mean error (ME); mean absolute error (MAE) and root mean square error (RMSE). Validation was based on particle size dis- tribution data provided by the Hungarian Detailed Soil Hydrophysical Database (MARTHA, Makó, A. et al. 2010). MARTHA has been developed to collect information on measured soil hydraulic and physical char- acteristics in Hungary. Recently this is the largest and most de- tailed national hydrophysical database. 780 records were used having both georef- erenced location and topsoil particle size distribution data entries. The results for two independent approaches (both using RK, but with diff erent ancillary dataset) are presented in Table 1. According to the fig- ures of the table, the two independent approaches result in slightly differ- ent maps as for their per- formance. The fi nal map products of sand and clay fraction for the upper- most (0–5 cm) soil layer are presented in Figure 6. Table 1. Validation of particle size fraction maps by various parameters for two diff erent approaches Para- meters RK1 RK2 clay silt sand clay silt sand ME MAE RMSE -2.77 7.63 10.42 0.05 11.20 15.28 2.72 13.49 18.38 -2.50 7.72 10.50 -0.13 11.03 15.10 2.62 13.36 18.19 Fig. 6. Countrywide maps of two main particle size fractions (sand, clay) for the uppermost soil layer (0–5 cm) 61Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. Conclusions In the frame of DOSoReMI.hu project a sig- nifi cant amount of experiences have been ac- cumulated so far concerning the compilation of novel and renewed, goal oriented, digital soil maps using geostatistical and data min- ing tools either at national or regional level. By the aid of selected and optimized geostatis- tical, data mining and GIS tools some basic soil properties were multi-mapped and optimized, but it is also planned to conduct the spatial ex- tension of certain more sophisticated pedological variables featuring the state, processes (including degradation), functions and services of soils. It is hoped to achieve further progress in the performance by expanding the pool of environmental co-variables applied and by testing additional methods (random forests, neural networks, support vector machines, etc.). The environmental correlation used in RK expressed by MLRA is also suggest- ed to be substituted by further, knowledge based data mining methods for improving the modelling of the complex relationship between a specifi c soil variable and its aff ect- ing/determining/indicating factors. The fi rst planned step is the substitution of MLRA with Regression Tree Analysis to generalize the linear model between the predicted and the explanatory environmental variables. The country-wide physical soil property maps elaborated according GlobalSoilMap specifi ca- tions will represent the fi rst Hungarian contribu- tion to the GlobalSoilMap.net project (Minasny, B. and McBratney, A.B. 2010). Further GSM.net conform map products are also under develop- ment in the frame of DOSoReMI.hu to contrib- ute to the worldwide activities. Based on more extended data infrastructure, it is suggested that national initiatives could produce more accurate and reliable products. The application of an inset map for the ex- pression of the inherent vagueness of spatial prediction is a rather recently introduced, but we propose its general usage for displaying the results of digital maps elaborated using geo-mathematical methods and/or environ- mental models. Acknowledgements: Our work has been supported by the Hungarian National Scientific Research Foundation (OTKA, Grant No. K105167). Authors thank J. Matus for her indispensable contribution. REFERENCES Ács, F., Gyöngyösi, A.Z., Breuer, H., Horváth, Á., Mona, T. and Rajkai, K. 2014. Sensitivity of WRF- simulated planetary boundary layer height to land cover and soil changes. Meteorologische Zeitschrift , PrePub DOI 10.1127/0941-2948/2014/0544 AGROTOPO 1994. AGROTOPO database of RISSAC. Budapest, RISSAC HAS, http://maps.rissac.hu/ agrotopo_en Baumgardner, M.F. 2011. Soil databases. In Handbook of Soil Sciences: Resource Management and Environmental Impacts. Eds.: Huang, P.M., Li, Y. and Sumner, M.E. Boca Raton, CRC Press, 21–35. Blum, W.E.H. 2005. Functions of soil for society and the environment. Reviews in Environmental Science and Biotechnology 4. 75–79. Boettinger, J.L., Howell, D.W., Moore, A.C., Hartemink, A.E. and Kienast-Brown, S. Eds. 2010. Digital Soil Mapping: Bridging Research, Environmental Application, and Operation. Heidelberg, Springer, 473 p. Böhner, J., Köthe, R., Conrad, O., Gross, J., Ringeler, A. and Selige, T. 2002. Soil regionalisation by means of terrain analysis and process parameterisation. In Soil Classifi cation 2001. Eds.: Michéli, E., Nachter- gaele, F. and Montanarella, L. EUR 20398 EN. The European Soil Bureau, Joint Research Centre, Ispra, 213–222. Bou Kheir, R., Bøcher, P.K., Greve, M.B. and Greve, M.H. 2010. The application of GIS based decision- tree models for generating the spatial distribution of hydromorphic organic landscapes in relation to digital terrain data. Hydrology and Earth System Sciences 14. 847–857. Breiman, L. 2001. Decision-tree forests. Machine Learning 45. (1): 5–32. Bullock, P. 1999. Soil resources of Europe – An over- view. In Soil Resources of Europe. Eds.: Bullock, P., Jones, R.J.A. and Montanarella, L. European Soil Bureau Research Report 6. Luxembourg, Offi ce for Offi cial Publications of the European Communities, 15–25. Dobos, E. and Hengl, T. 2009. Soil mapping appli- cations. In Geomorphometry – Concepts, Soft ware, Applications. Eds.: Hengl, T. and Reuter, H.I. Deve- lopment in Soil Science series 33. 461–479. Dobos, E., Bialkó, T., Micheli, E. and Kobza, J. 2010. Legacy soil data harmonization and database development. In Digital Soil Mapping Bridging Research Environmental Application, and Operation: Eds.: Boettinger, J.L., Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.62 Howell, D.W., Moore, A.C., Hartemink, A.E. and Kienast-Brown, S. Heidelberg, Springer, 309–323. Dobos, E., Carré, F., Hengl, T., Reuter, H.I. and Tóth, G. Eds.. 2006. Digital soil mapping as a support to production of functional maps. EUR 22123 EN. Luxembourg, Offi ce for Offi cial Publications of the European Communities, 68 p. Farkas, Cs., Gelybó, Gy., Bakacsi, Zs., Horel, Á., Hagyó, A., Dobor, L., Kása, I. and Tóth, E. 2014. Impact of expected climate change on soil water regime under different vegetation conditions. Biologia, Manuscript, accepted for publication. Fodor, N., Pásztor, L. and Németh, T. 2014. Coupling the 4M crop model with national geo-databases for assessing the eff ects of climate change on agro- ecological characteristics of Hungary. International Journal of Digital Earth 7. (5): 391–410. Gessler, P.E., Moore, I.D., McKenzie, N.J. and Ryan, P.J. 1995. Soil-landscape modelling and spatial prediction of soil att ributes. International Journal of Geographical Information Systems 9. (4): 421–432. Giasson, E. Sarmento, E.C., Weber, E., Flores, C.A. and Hasenack, H. 2011. Decision trees for digital soil mapping on subtropical basaltic steeplands. Scienta Agricola 68. (2): 167–174. Goovaerts, P. 2000. Geostatistical approaches for incorporating elevation into the spatial interpo- lation of rainfall. Journal of Hydrology 228. (1–2), 113–129. Greve, M.H., Kheir, R.B., Greve, M.B. and Bøcher, P.K. 2012. Quantifying the ability of environmental param- eters to predict soil texture fractions using regression- tree model with GIS and LIDAR data: The case study of Denmark. Ecological Indicators 18. 1–10. Grunwald, S. 2009. Multi-criteria characterization of re- cent digital soil mapping and modelling approaches. Geoderma 152. 195–207. Gyalog, L. and Síkhegyi, F. Eds. 2005. Magyarország geoló- giai térképe, 1:100 000 (Geological Map of Hungary, 1:100 000), Budapest, Geological Institute of Hungary, Digital version: htt p://loczy.mfgi.hu/fdt100/ Hartemink, A.E., Mcbratney, A.B. and Mendonça- Santos, M.De L. Eds. 2008. Digital Soil Mapping with Limited Data. Springer, The Netherlands, 445 p. Henderson, B.L., Bui, E.N., Moran, C.J. and Simon, D.A.P. 2005. Australia-wide predictions of soil properties us- ing decision trees. Geoderma 124. (3–4): 383–398. Hengl, T., Heuvelink, G. and Stein, A. 2004. A generic framework for spatial prediction of soil variables based on regression-kriging. Geoderma 122. (1–2), 75–93. Hengl,T. 2009. A Practical Guide to Geostatistical Mapping. Amsterdam, University of Amsterdam, 291 p. Heuvelink, G.B.M. and Webster, R. 2001. Modelling soil variation: past, present, future. Geoderma 100. 269–301. Illés, G., Kovács, G. and Heil, B. 2011. Comparing and evaluating digital soil mapping methods in a Hungarian forest reserve. Canadian Journal of Soil Science 91. (4): 615–626. Illés, G., Kovács, G., Laborczi A. and Pásztor L. 2014. Zala megye egységes talajtípus adatbáz- isának összeállítása (Developing a unifi ed soil type database for Zala County using classifi cation algo- rithms). Erdészett udományi Közlemények. (in press) Jenny, H. 1941. Factors of Soil Formation. New York, McGraw-Hill, 281 p. Kozma, Zs., Derts, Zs., Kardos, M. and Koncsos, L. 2012. A mezőgazdasági termelés mint ökoszisz- téma-szolgáltatás értéke: hidrológiai modellhez kapcsolt számítási módszertan (The value of agri- cultural crops as an ecosystem service: calculation methodology connected to a hydrological Model) Tájökológiai Lapok 10. (1): 55–69. Kreybig, L. 1937. A Magyar Királyi Földtani Intézet talajfelvételi, vizsgálati és térképezési módszere (The survey, analytical and mapping method of the Hungarian Royal Institute of Geology). Magyar Királyi Földtani Intézet Évkönyve 31. 147–244. Laborczi, A., Szatmári, G., Takács, K. and Pásztor, L. 2015. Topsoil texture class map of Hungary compiled using classifi cation trees. Journal of Maps (submitt ed paper) Lagacherie P. 2008. Digital soil mapping: A state of art. In Digital Soil Mapping with Limited Data. Eds.: Hartemink, A.E., Mcbratney, A.B. and Mendonça- Santos, M.De L. Dordrecht, Springer, 3–14. Lagacherie, P. and Mcbratney, A.B. 2007. Spatial Soil Information Systems and Spatial Soil Inference Systems: perspectives for digital soil mapping. In Digital soil mapping: an introductory perspective. Eds.: Lagacherie, P., Mcbratney, A.B. and Voltz, M., Amsterdam, Elsevier, 3–22. Lagacherie, P., Mcbratney, A.B. and Voltz, M. Eds. 2007. Digital soil mapping: an introductory perspective. Amsterdam, Elsevier, 658 p. Lawrence, R., Bunn, A., Powell, S. and Zambon, M. 2004. Classifi cation of remotely sensed imagery using stochastic gradient boosting as a refi nement of classifi cation tree analysis. Remote Sensing of the Environment 90. 331–336. Makó A., Tóth, B., Hernádi, H., Farkas, Cs. and Marth, P. 2010. Introduction of the Hungarian Detailed Soil Hydrophysical Database (MARTHA) and its use to test external pedotransfer functions. Agrokémia és Talajtan 59. (1): 29–38. Mark, D.M. and Csillag, F. 1989. The nature of boundaries on ’area-class’ maps. Cartographica 26. (1): 65–78. McBratney, A.B. and Odeh, I.O.A. 1997. Application of fuzzy sets in soil science: fuzzy logic, fuzzy measure- ments and fuzzy decisions. Geoderma 77. 85–113. McBratney, A.B., Mendonça-Santos, M.L. and Minasny, B. 2003. On digital soil mapping. Geoderma 117. 3–52. McKenzie, N.J. and Ryan, P.J. 1999. Spatial prediction of soil properties using environmental correlation. Geoderma 89. (1–2): 67–94. 63Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64. Mermut, A.R. and Eswaran, H. 2000. Some major developments in soil science since the mid-1960s. Geoderma 100. 403–426. Minasny, B. and McBratney, A.B. 2010. Methodologies for Global Soil Mapping, Dordrecht, Springer Netherlands, 429–436. Minasny, B., Malone, B.P. and McBratney, A.B. Eds. 2012. Digital Soil Assessments and Beyond. London, Taylor and Francis Group, 466 p. Montanarella, L. 2010. Need for interpreted soil in- formation for policy making. In 19th World Congress of Soil Science, Soil Solutions for a Changing World, 1–6 August 2010, Brisbane, Australia. Published on DVD. Moran, C.J. and Bui, E.N. 2002. Spatial data mining for enhanced soil map modelling. International Journal of Geographic Information Science 16. 533–549. Mulder, V.L., de Bruin, S., Schaepman, M.E. and Mayr, T.R. 2011. The use of remote sensing in soil and ter- rain mapping. – A review. Geoderma 162. 1–19. Nemes, A., Pachepsky, Y.A. and Timlin, D.J. 2011. Toward improving global estimates of fi eld soil water capacity. Soil Science Society of America Journal 75. (3): 807–812. Omuto, C., Nachtergaele, F. and Rojas, R.V. 2013. State of the Art Report on Global and Regional Soil Information: Where are we? Where to go? Global Soil Partnership Technical Report. Rome, FAO, 69 p. Panagos, P., van Liedekerke, M., Jones, A. and Montanarella, L. 2012. European Soil Data Centre: Response to European policy support and public data requirements. Land Use Policy 29. 329–338. Pásztor, L., Bakacsi, Zs., Laborczi, A. and Szabó, J. 2013b. Kategória típusú talajtérképek térbeli felbontásának javítása kiegészítő talajtani ada- tok és adatbányászati módszerek segítségével (Downscaling of categorical soil maps with the aid of auxiliary spatial soil information and data mining methods). Agrokémia és Talajtan 62. (1): 205–218. Pásztor, L., Dobos, E., Szatmári, G., Laborczi, A., Takács, K., Bakacsi, Zs. and Szabó, J. 2014. Application of legacy soil data in digital soil mapping for the elabo- ration of novel, countrywide maps of soil conditions. Agrokémia és Talajtan 63. (1): 79–88. Pásztor, L., Szabó, J. and Bakacsi, Zs. 2010. Digital processing and upgrading of legacy data collected during the 1:25 000 scale Kreybig soil survey. Acta Geodaetica et Geophysica Hungarica 45. 127–136. Pásztor, L., Szabó, J., Bakacsi, Zs. and Laborczi, A. 2013a. Elaboration and applications of spatial soil information systems and digital soil mapping at the Research Institute for Soil Science and Agricultural Chemistry of the Hungarian Academy of Sciences. Geocarto International 28. (1): 13–27. Pásztor, L., Szabó, J., Bakacsi, Zs., Matus, J. and Laborczi, A. 2012. Compilation of 1:50 000 scale digital soil maps for Hungary based on the digital Kreybig soil information system. Journal of Maps 8. (3): 215–219. Rajkai, K. and Kabos, S. 1999. A talaj víztartóképesség- függvény (pF-görbe) talajtulajdonságok alapján történő becslésének továbbfejlesztése (Estimation of Soil Water Retention Characteristics (pF Curves) From Other Soil Properties. Agrokémia és Talajtan 48. (1–2): 15–32. Sanchez, P.A., Ahamed, S., Carré, F., Hartemink, A.E., Hempel, J., Huising, J., Lagacherie, P., McBratney, A.B., McKenzie, N.J., Mendonça-Santos, M.L., Minasny, B., Montanarella, L., Okoth, P., Palm, C.A., Sachs, J.D., Shepherd, K.D., Vågen, T.G., Vanlauwe, B., Walsh, M.G., Winowiecki, L.A. and Zhang, G.L. 2009. Digital soil map of the world. Science 325. 680–681. Saxton, K.E. and Rawls, W.J. 2006. Soil Water Characteristic Estimates by Texture and Organic Matt er for Hydrologic Solutions. Soil Science Society of America Journal 70. (5): 1569–1578. Scull, P., Franklin, J. and Chadwick, O.A. 2005. The application of classifi cation tree analysis to soil type prediction in a desert landscape. Ecological Modelling 181. 1–15. Scull, P., Franklin, J., Chadwick, O.A. and McArthur, D. 2003. Predictive soil mapping: a review. Progress in Physical Geography 27. (2): 171–197. Shirazi, M.A. and Boersma, L. 1984. A Unifying Quantitative Analysis of Soil Texture. Soil Science Society of America Journal 48. (1): 142–147. Sisák, I. and Benő, A. 2014. Probability-based harmo- nization of digital maps to produce conceptual soil maps. Agrokémia és Talajtan 63.(1): 89–98. Specifi cations Version 1 GlobalSoilMap.net products, Release 2.1, (2014) . Szabó, J., Pásztor, L. and Bakacsi, Zs. 2011. Demand, feasibility and construction stages of a national spa- tial soil information system. Agrokémia és Talajtan 60. (Suppl.), 149–160. Szabó, J., Pásztor, L., Bakacsi, Zs., László, P. and Laborczi, A. 2007. A Kreybig Digitális Talajinformációs Rendszer alkalmazása térségi szintű földhasználati kérdések megoldásában (Application of the Kreybig Digital Soil Information System to solve land use problems at regional level). Agrokémia és Talajtan 56. (1): 5–20. Szatmári, G. and Barta, K. 2013. Csernozjom talajok szervesanyag-tartalmának digitális térképezése erózióval veszélyeztetett mezőföldi területen (Digital mapping of the organic matt er content of chernozem soils on an area endangered by erosion in the Mezőföld region). Agrokémia és Talajtan 62. (1): 47–60. Szatmári, G., Laborczi, A., Illés, G. and Pásztor, L. 2013. A talajok szervesanyag-készletének nagyléptékű térképezése regresszió krigeléssel Zala megye példáján (Large-scale mapping of soil organic matt er content by regression kriging in Zala County). Agrokémia és Talajtan 62. (2): 219–234. Pásztor, L. et al. Hungarian Geographical Bulletin 64 (2015) (1) 49–64.64 Tobler,W. 1970. A computer movie simulating urban growth in the Detroit region, Economic Geography 46. (2): 234–240. Tóth, B., Makó, A. and Tóth, G. 2014. Role of soil properties in water retention characteristics of main Hungarian soil types. Journal of Central European Agriculture 15. (2): 137–153. Tóth, G., Montanarella, L., Stolbovoy, V., Máté, F., Bódis, K., Jones, A., Panagos, P. and van Liedekerke, M. 2008. Soils of the European Union. EUR 23439 EN. Luxembourg, Offi ce for Offi cial Publications of the European Communities, 85 p. USDA 1987. Soil Mechanics Level I. Module 3 – USDA Textural Soil Classifi cation Study Guide. National Employee Development Staff , Soil Conservation Service, United States Department of Agriculture, USA, 232 p. Van Orshoven, J., Terres, J-M. and Tóth. T. 2013. Updated common bio-physical criteria to defi ne natural constraints for agriculture in Europe. Defi nition and scientific justification for the common criteria. Technical Factsheets. Luxembourg: Office for Offi cial Publications of the European Communities 2012. Scientific and Technical Research, EUR 25203 EN. Joint Research Centre, Institute for Environment and Sustainability. Várallyay, Gy. 2011. Water storage capacity of Hungarian soils. Agrokémia és Talajtan 60. (Suppl.), 7–26. Várallyay, Gy. 2012. Talajtérképezés, talajtani adatbázisok. Agrokémia és Talajtan 61. (Suppl.), 249–267. Várallyay, Gy., Szűcs, L., Zilahy, P., Rajkai, K. and Murányi, A. 1985. Soil factors determining the agro-ecological potential of Hungary. Agrokémia és Talajtan 34. (Suppl.), 90–94. Vereecken, H., Maes, J., Feyen, J. and Darius, P. 1989. Estimating the soil moisture retention characteristic from texture, bulk density, and carbon content. Soil Science 148. 389–403. Waltner, I., Micheli, E., Fuchs, M., Láng, V., Pásztor, L., Bakacsi, Zs., Laborczi, A. and Szabó, J. 2014. Digital mapping of selected WRB units based on vast and diverse legacy data. In Global Soil Map: Basis of the Global Spatial Soil Information System. Eds.: Arrouays, D., McKenzie, N., Hempel, J., Richer de Forges, A. and McBratney, A.B. London, Taylor and Francis Group, 313–318. << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Dot Gain 20%) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Error /CompatibilityLevel 1.3 /CompressObjects /Tags /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams false /MaxSubsetPct 100 /Optimize false /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness true /PreserveHalftoneInfo false /PreserveOPIComments true /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 300 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.50000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description << /ARA /BGR /CHS /CHT /CZE /DAN /DEU /ESP /ETI /FRA /GRE /HEB /HRV (Za stvaranje Adobe PDF dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. Stvoreni PDF dokumenti mogu se otvoriti Acrobat i Adobe Reader 5.0 i kasnijim verzijama.) /ITA /JPN /KOR /LTH /LVI /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.) /NOR /POL /PTB /RUM /RUS /SKY /SLV /SUO /SVE /TUR /UKR /ENU (Use these settings to create Adobe PDF documents best suited for high-quality prepress printing. Created PDF documents can be opened with Acrobat and Adobe Reader 5.0 and later.) /HUN >> /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ << /AsReaderSpreads false /CropImagesToFrames true /ErrorControl /WarnAndContinue /FlattenerIgnoreSpreadOverrides false /IncludeGuidesGrids false /IncludeNonPrinting false /IncludeSlug false /Namespace [ (Adobe) (InDesign) (4.0) ] /OmitPlacedBitmaps false /OmitPlacedEPS false /OmitPlacedPDF false /SimulateOverprint /Legacy >> << /AddBleedMarks false /AddColorBars false /AddCropMarks false /AddPageInfo false /AddRegMarks false /ConvertColors /ConvertToCMYK /DestinationProfileName () /DestinationProfileSelector /DocumentCMYK /Downsample16BitImages true /FlattenerPreset << /PresetSelector /MediumResolution >> /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ] >> setdistillerparams << /HWResolution [2400 2400] /PageSize [612.000 792.000] >> setpagedevice