Study proposes the introduction of predictive archaeological models in Hungary. Based on studies involving the known Hungarian site register. The chance of protecting sites we do not know the location of is next to none.
Study proposes the introduction of predictive archaeological models in Hungary. Based on studies involving the known Hungarian site register. The chance of protecting sites we do not know the location of is next to none.
Study proposes the introduction of predictive archaeological models in Hungary. Based on studies involving the known Hungarian site register. The chance of protecting sites we do not know the location of is next to none.
ARCHAEOLOGICAL PREDICTIVE MODELS AND THE HUNGARIAN
CULTURAL HERITAGE PROTECTION
GBOR MESTERHZY MT STIBRNYI 1
The first and foremost task of archaeological heritage protection is the spatial identification of archaeological sites. Our study proposes the introduction of predictive archaeological models in Hungary based on studies involving the known Hungarian site register. These could mean extensive tools, already available at the initial planning phase of the construction works, by which the preliminary protection of national sites can be secured.
Keywords: site, GIS, predictive modelling
The approach , where these decisions are taken on the basis of (expert) knowledge of the known archaeological site sample, is in our view an irresponsible approach to archaeological heritage management. VERHAGEN 2010, 438
When asking archaeologists to show us the location of their research, we are bound to be shown a wealth of thrilling sites. Yet they will be in trouble when we ask them to show us each and every one of them. This is due to the fact that no one knows the actual number of archaeological sites in Hungary. This is not a theoretical problem, 2 since the chance of protecting sites we do not know the location of is next to none. 3 Although almost all archaeological research deals with sites, there is disproportionately little discourse on the possibilities of identifying a register of sites. Nevertheless, without doubt such a huge amount is in question that it simply cannot be managed without a geographic information system (GIS). Any archaeological data within the archaeological register of sites which lacks more
1 Hungarian National Museum Centre for National Cultural Heritage, Department of Topography; gabor.mesterhazy@mnm-nok.gov.hu, mate.stibranyi@mnm-nok.gov.hu 2 Act LXIV 11 and 7 (20) of 2001. 3 It is not possible to discuss the definition of archaeological sites here; we interpret the archaeological site as the distinctive spatial location of the special collective of finds, archaeological features and ecological remains (RACZKY 2006), augmented by the fact that its expanse is defined by the surrounding area lacking of finds. However, this does not answer a number of questions: e.g., when finds from several periods are unearthed together should they be regarded as one or more sites; or how many sites comprise, for example, the Roman period city town of Aquincum? We have already discussed the question briefly (REMNYI STIBRNYI 2010). 2 precise geographical data cannot be studied; it simply cannot be identified as a site. In the following essay we will summarize the state of the Hungarian archaeological register of sites, connecting GIS-based analyses, as well as those options which serve as a more protective solution for the sites from such a viewpoint.
The known archaeological site register in Hungary
At first sight, the situation is promising. A unified and centralized data archive exists where information about archaeological sites is collected. This is quite an outstanding feat knowing the divided state of Hungarian archaeology. The official site register of the National Office of Cultural Heritage (KH) can be regarded as a reliable base for the study of the site registers state as it contains the topographical data collected by the Archaeological Topography of Hungary (MRT), 4 the archival databases of the majority of museums, as well as archaeological excavational data conducted since 2001. Currently it contains more than 70 000 records, while the number of sites in Hungary is estimated to be about 100 000 (WOLLK 2009). According to this, we already know 70% of the sites. However, this is not true. First of all: we certainly do not know 70 000 sites. Sites can only be considered known when they possess spatial data, that is, simply their geographical location is known. The database of the KH was not contrived as a geographic information system at the time of its compilation; thus, it contains a lot of site descriptions that do not have, and indeed cannot have any geographical data. Only about two thirds of the given information, 50 000 sites have some kind of geographical data added (whether they have an extension or are only dot-like). Furthermore, only ca. 10% among these can be found within city borders (this means 5.42% of the area of Hungary), which sheds light on its disproportionate state caused by the fact that due to divisions usually the same site receives different registry numbers according to the assigned plots of land. Fairly often a single site is registered under several numbers. In all, we estimate the actually known number of sites to be between 25 00030 000. 5 As we can identify several periods and characteristics at the majority of sites this number cannot be given unequivocally. Nevertheless, for the time being let us be content with the current practice based on spatially defined units: site means a location where archaeology is found surrounded by an area free from artefacts.
4 See below for further details. 5 In 2004 this number was 10 00012 000 (WOLLK 2004, 76); however, this did not yet contain the geographic data resulting from the Transdanubian research of the MRT (cf. WOLLK 2004, 77, fig. 5). 3 Many scholars have already extrapolated the number of sites on regions systematically researched up till now to the whole area of the country (JANKOVICH-BSN AND NAGY 2004; WOLLK 2004), which resulted in an estimate of 100.000150.000 sites within Hungary. This implies ca. 11.5 sites to each square kilometre. This is feasible judging by the study of the sites observable on the surface, which number is more or less supported by current research as well. The staff of the KH and the Directorate of Somogy County Museums conducted topographical research at 17 settlements of Inner Somogy between 20002004, where 517 sites were located in an area of 426 square kilometres (1.21 sites/km) (FEKETE et al. 2005). Our field survey conducted in spring 2011 at Perkta and environs with an area of almost 160 km surveyed resulted in 220 sites (1.375 sites/km). These surveys, just as those conducted by the MRT, certainly have not found all the sites of a given area, but we can state that they have given significant results. The 11.5 sites/km ratio is reasonable in comparison to European site registers as well, although we meet rather different figures according to each region. In Poland 435 000 sites were located on a 270 000 km area (1.6 sites/km) (PRINKE 2009, 72); in Sweden 570 000 sites were identified on a 450 000 km area in 2009 (1.2 sites/km) (NORMAN - SOHLENIUS 2009, 83); the 20 000 km area of Slovenia yielded 26 000 sites (1.3 sites /km) (DJURI et al. 2009, 90). Due to the variation in the data (e.g. site definitions) we must not draw far-reaching conclusions from these figures, but the ratios are observable. Thus the 100 000150 000 sites estimated for the area of Hungary seems reasonable. We should not omit from our calculations the fact that there are also archaeological sites which are unidentifiable by mere field surveys. However, for the moment we do not know of such an extensive survey which could significantly elucidate the question of the ratio of these types of sites within the site register. All in all, we think it is not the numbers which are important, but what results from them; and based on these we know 2030% of the archaeological sites in Hungary, while ca. 25% of the country has been systematically researched. 6 We feel this is a significant problem from several aspects. From an archaeological perspective, apart from academic relevance, this low figure means that practically 70% of the site register is unprotected, as current legislation only protects registered sites. In line with Act LXIV of 2001, protection primarily refers to
6 The MRT examined 11.7% of Hungary, researching ca. 10 000 sites, while there are a further two volumes and almost 2 000 sites still in manuscript form (WOLLK 2004, and for data from manuscripts Istvn Torma, personal communication). 4 preventing the disturbance of the site and not its excavation. Archaeology is thus concerned with constructions avoiding the sites, although due to its underfinanced state in the past few years we seem to forget that our primary ethical and professional duty is not the excavation of sites, but protecting them for the future. From the viewpoint of constructions, this state is also problematic as the high percentage of unknown sites does not allow for planning to avoid them. If such a large percent of the sites is unknown, it is practically futile to ask for archaeological data during the course of planning as there is only a little chance that it will be of any use. An average 70% of the sites ideally only emerge during the compulsory field survey conducted because of the authorization. During the planning phase of the Nabucco gas pipelines running along 384 km through seven counties of Hungary 77 sites were already registered. After a single field survey in 2010, this was modified to 339, but including the percentage of areas where the survey could not be conducted we can estimate at least 400 sites altogether. This of course means an unforeseen risk with regard to schedules and monetary matters for the construction works, which can only be more precisely revealed ideally during authorization, less ideally during construction itself.
Solution possibilities
1. As we have already mentioned in the previous chapter, we regard current legislation, which tries to carry out the protection of sites based solely on the rather small list of sites known, very dangerous. One possibility was outlined ahead of its time by the Archaeological Topography of Hungary (MRT): to conduct intensive field surveys throughout the country. Under the circumstances prevalent in Hungary, due to the extremely high number of cultivated land, field surveys are the cheapest and easiest methods for identifying sites. By this procedure we can gain reliable information in ploughed fields regarding how much they were affected by archaeology. However, its drawbacks are that although the surface-collected finds are deeply connected to the features hidden under the surface, their relationship does not always mean they can be more precisely associated. The fact that a site exists at the spot can usually be determined with certainty, but its precise extent cannot, grounded simply on surface features. Furthermore, in certain areas and with certain types of sites this method cannot be employed. 5 The early undertaking of the MRT started with a positivist approach in the 1960s, growing into a progressive topographical workshop, but was eventually broken off by the end of the 1990s; it had surveyed 11.7% of the area (WOLLK 2004). We may regard this as the basis. What we know from the site register was compiled for the most part by these studies. It is enough to think of the fact that almost 12 000 sites from the total number of 25 00030 000 known sites were identified by the MRT (WOLLK 2004, and for data from manuscripts Istvn Torma, pers. comm.). Even though the numbers suggest so, the task is not impossible, yet it requires significant resources. In Poland the National Archaeological Survey (Archeologiczne Zdjcia Polski, AZP) organized the field survey of almost the whole of the country with a huge amount of human and material resources involved. The end of the first phase (field survey) of the project started in 1978 is to be expected in a few years time. By 2009 with the involvement of 500 archaeologists more than 435 000 sites were identified in Poland (PRINKE 2009). In Sweden archaeologists have conducted field surveys twice so far in areas not covered by forests, resulting in 600 000 sites; currently the archaeological survey of forested areas is underway (NORMAN - SOHLENIUS 2009). For the moment, we do not see reality in any form for systematic topographical surveys, similar to the abovementioned, to start in Hungary in the near future. This is partially due to financial reasons; for archaeology in such an underfinanced state, it can only be conceived with increased state engagement. A greater, more complex problem is the disinterest of the archaeological profession on this matter. The reasons for this would deserve an essay of their own, but we think its roots lead back to the fact that the archaeological revolution spreading throughout Europe following the Malta Convention in 1992 has only partially reached Hungary (BNFFY 2004). The profession regards the necessity of preliminary rescue excavations as self-evident, but not that this would also entail the establishment of a Cultural Resources Management (CRM) independent from academic research. 7 CRM, however, does not only cover archaeological companies and motorway excavations, but also implies a change in approach that results from the complex task of protecting every archaeological feature. Nevertheless, this raises different, not too scholarly questions in the academic sense of the word.
2.
7 For the definition of the phrase cf. KING 2004, 4. 6 We can approach the solution to this problem from another angle as well: let us turn the question around and try to anticipate where archaeological sites and features can be expected. The method is based on model experiments that are employed by several sciences from medicine to zoology, collectively called predictive models (ARC-WOFE 1998).
Fig. 1.: DEM of the Srrt-region at Fejr County. (SRTM 2006)
Archaeological predictive models try to predict the location of archaeological sites or materials in a region, based either on a sample of that region or on fundamental notions concerning human behaviour (VERHAGEN 2007, 13). The majority of these are based on two assumptions: on the one hand, the choice of the location of a human settlement was largely influenced by certain characteristics of the natural environment. On the other hand, these natural factors influencing settlement location appear at least indirectly on modern maps (WARREN ASCH 2000, 6-7). Thus, human settlement is not random in the region and the dispersion of sites found so far can be modelled along natural and cultural resultants. In such a way the locations where sites may occur with more possibility or where this is less likely can be filtered by geographic information and geostatistical analyses. This is quite a pragmatic attitude and obviously comes from a CRM stance: if time and money are not sufficient for a 7
Fig. 2.: Archaeological predictive model of Srrt-region in Fejr County (MESTERHZY 2011 75.) 8 full-scale examination, but we definitely need to give an answer and fast, then we must be aware of the possible solutions. In the past 3040 years that these models have been in use, however, the methodology has outgrown the status of a simple aid. Today we use these data to better understand and study connections between human activity and the natural environment. Likewise, the spatial analyses prepared during the modelling are not only to be interpreted as elements of the model, but as an additional benefit, so to speak, they can be useful data on their own. The effect of natural geography, thus, primarily watercourses, relief features and soil, on human settlement has already been studied both ethnographically and geographically in the first half of the 20th century; Zsigmond Btky and Istvn Gyrffy from an ethnographical perspective (BTKY-GYRFFY-VISKI 1941), and the geographer Tibor Mendl also researched the topic (MENDL 1932). In terms of archaeological settlement structure, Istvn Mri was among the first to emphasize the significance of proximity to water and the features of the terrain (MRI 1952), but this realization also led the directors of the field surveys conducted by the MRT. The relationship of waters and settlements is an obvious fact for archaeologists today. Hungarian archaeological research in the past 20 years has been primarily influenced by environmental archaeology, a science closely connected to and parallelly developing with predictive modelling. We can regard these results as the basis for predictive models. Thus, in the following we will briefly summarize those Hungarian studies which we find progressive from such a perspective and which can be useful for predictive modelling. Archaeological GIS in Hungary
Among GIS systems used for archaeological purposes in Hungary we must first mention those that categorize and systematize data collected during archaeological excavations and display these spatially. The primary aim of these database-like systems based on features and/or stratigraphical units is to store the wealth of data collected from excavations with large surfaces, otherwise rather difficult to manage, according to unified system requirements (WOLF 2002; EKE et al. 2007; TOLNAI 2009; HOLL PUSZTAI 2011). Nonetheless, we will not go into detail regarding these processing methods and applications within the scope of this essay even if they were one of the first steps taken in Hungarian GIS research. Most of all for the reason that from the perspective of predictive models those Hungarian works are more important that not only structurally organize and store data, but have also prepared spatial studies and reconstructions, and among them those that have been 9 spatially displayed, that is, they can be presented on a map. It has to be stressed, that predictive archaeological modelling makes use of archaeological data on the level of and above the site surface, thus we concentrated on these during the study. From the viewpoint of a study relying on the predictive archaeological model that is mostly based on environmental factors, it is of great importance to determine environmental changes. A significant difference between predictive models and environmental reconstructions is that the latter deal with a single level or coverage of the environment. As will be seen later, they do contain predictions regarding favourable and non-favourable zones of prospective archaeological site locations; however, this only refers to the given coverage. On the other hand, the predictive archaeological model is a complex prediction based on several environmental factors implemented by geographical information and geostatistical analyses regarding the probable location of archaeological sites. Consequently, environmental reconstructions can specify the model itself as the input data of the predictive model. It is important to emphasize that this is only possible, if these input data meant for improvement form databases with well-defined precision and spatial characteristics that can be managed by geographical information systems and are not just textual statements. The following research historical sketch serves to point out, amongst others, that: the knowledge necessary for the predictive archaeological models (or geographical information coverages, if you will) is used more and more in Hungarian archaeology; although collective and unified processing through geographical information systems is still rarely to be seen (FEKETE 2008; PADNYI-GULYS 2010; PADNYI-GULYS 2011; MESTERHZY 2011). It is easy to see that during the reconstruction of any archaeological period, all the changes in the region and the environment that happened after the period under examination, or after 1711 8 at the latest, are mostly negligible from an archaeological point of view. Thus, one of the important sources and starting points for environmental reconstructions is the information to be found on modern maps. One of the most significant roles of these contemporary representations is to unravel the land formation effects of the river regulation and swamp reclamation works that had commenced in the 19 th century; but it also offers valuable data on topography and vegetation. The changes that came about in the natural landscape, and thus in the environment of the archaeological sites as well, can be sufficiently demonstrated by the three military surveys (ARCANUM 2004, ARCANUM 2006a, ARCANUM
8 Act LXIV 19 of 2001. 10 2007) and other, county (ARCANUM 2009) or various local (ARCANUM 2006b, ARCANUM 2006c) maps referring to the area of research. Most examples apply to the adjustment of hydrographic data and the refinement of contemporary circumstances. This already goes to show that water, whether still or flowing, is one of the most important factors in human settlement, either from a positive or negative aspect. Since it not only attracts settlements, but it also repels them if we think of swampy regions and floods. The opportunity presents itself to reconstruct the configurations of the terrain prior to river regulations directly next to the site and to trace back the waterlogged areas on them (RACZKY et al. 1997, 170; SZAB et al. 1997a, 81; SZAB et al. 1997b, 87; FISCHL 2006, 12). In the case of larger, microregional studies, however, it is necessary to employ geological data as well (RACZKY et al. 2002; GYUCHA DUFFY 2008; FZESI 2009). Logic in such cases is the opposite contrary to old maps, as the direction of approaching prehistoric circumstances is not from the top but from below, through the changes occurring in the Pleistocene and Holocene epochs determined by geological methods. For the moment only along shorter lengths, but it is proven that the location of archaeological sites is also suitable for the delineation of old river-beds and flood plains (GYUCHA DUFFY 2008, 20; FZESI 2009, 382) at areas where the beds are known to change frequently; this is characteristic of wide glacial valleys such as those at the lower section of the Saj and Hernd rivers. These data also result in the refinement of geological data. Usage of old maps, however, is often not sufficient anymore. Despite the given sources varying according to each area, they provide reliable information only until the 17 th 18 th centuries. Thus, the other significant sources of environmental reconstructions are the results of various scientific analyses that are less defined by periods. Environmental archaeological research based on general samples spreading across large expanses are also known in Hungary (GL et al. 2005; ZATYK et al. 2007); the complex geomorphological, sedimentological, radiocarbon, palaeo- and macrobotanical, and malacological analyses serve the better understanding of the related settlement history overviews. The studies of Pl Smegi cannot be left out from our survey, who is an authoritative figure within the field of Hungarian environmental archaeological research (an incomplete list of examples: KERTSZ SMEGI 1999; RACZKY et al. 2002, 840-842; SMEGI et al. 2003; SMEGI MOLNR 2004; SMEGI 2009). For our survey the spatially displayed, more accurate periodic hydrographic models are the most relevant (SMEGI 2007, 379-382; SMEGI et al. 2007, 250-251). 11 A significant result of soil boring conducted directly in the vicinity of archaeological sites is precisely the more accurate delineation of the contemporary relief of an area. The results of the drills conducted usually along a 10x10 square grid can be utilized when deducted from the height values of the current relief model (VARGA 2000, 75; SALISBURY 2008, 53). However, the problem with this method is that on a vast territory this also culminates in significant expenses in the budget. During the research conducted at Kelemr-Mohosvr, the access of the castle and the direction the finds were washed down could be determined by calculating the gradient categories (PUSZTAI 2007b, 56-57). Likewise, visibility could be studied by the reconstruction of a digital elevation model and the height of the tower (PUSZTAI 2005, 419), which made it possible to define the microenvironment of the castle and the exact locations of the nearby settlements connected to the castle. Two studies have to be mentioned at the end of our overview. It can be considered a great progress that a hydrographic reconstruction supported by remote sensing data was made in the flat terrain of the Great Hungarian Plain (TMR 2004). At areas where the half metre sea level differences on even the best contemporary Unified National Mapping System (EOTR) maps practically appear as flat surfaces on a relief model, much more advantageous possibilities arise with the processing of remote sensing images, which means the redrawing of river-beds that appear on them in the case of a hydrographic reconstruction. Its drawback, however, is that in exchange for the exact spatial data one must conduct chronological examinations at the given area. One of the most important elements of the research conducted at Ecsegfalva besides the hydrographic reconstruction was the mapping of vegetation data resulting from the environmental reconstruction. Thus, visibility studies conducted on plain and vegetation- covered surfaces present varying prospects (GILLINGS 2007, 39-42). Likewise, a visibility study conducted from a single given location, which already considers lengths within distances still seen by the unaided eye (GILLINGS 2007, 44-46), can be very useful in studies regarding relationships between the individual settlements. The creation of predictive archaeological models in Hungary is still in the making. Until now two essays have been published: one of research conducted in Somogy County (FEKETE 2008, 147-156) and one in the area of the Zsmbk Basin (PADNYI-GULYS 2010, 1-6); while a further data collection is known from Borsod-Abaj-Zempln County (FISCHL 2008). So far, two predictive models have been prepared at the professional workshop of the Centre for National Cultural Heritage of the Hungarian National Museum. These are shortly 12 to be published in detail, where one deals with the results of modelling conducted at Srrt (MESTERHZY 2011) and the other at the Srvz Basin (PADNYI-GULYS 2011). We must also mention the predictive model made in connection with Kelemr-Mohosvr, which defined the location of the medieval village of Fancsal based on visibility and exposure; unfortunately, a field survey has not been carried out thus far to verify the hypothesis (PUSZTAI 2007, 63).
Predictive models: history of research
P. Verhagen attributes the research historical basis of modelling to the settlement network analyses of New Archaeology (Verhagen 2007, 14). Its background is built up of the archaeological interpretation of the geographical location theory and the site catchment analysis developed from the latter (CHISHOLM 1962). At the end of the 1960s these theoretical approaches received a quantitative background for employment in the period of processual archaeology; while by the end of the 1970s two volumes discussing archaeological spatial analysis were born (HODDER ORTON 1976; CLARKE 1977). After this, the theoretical background of the predictive model in archaeology was practically ready. The first predictive models were prepared in the USA. According to the National Historic Preservation Act, approved by the Federal Government in 1966, archaeological sites were not renewable resources and were thus values to be protected. Since, however, this referred to the heritage protection of such vast expanses where field surveys could not be conducted, there was a lot of pressure on the departments involved to find a solution. This resulted in the establishment of the first data-driven predictive models (MEHRER WESCOTT 2004, 6-9). These were prepared in large numbers after the second half of the 1970s. The basis of preparing the American-type predictive model was widely published (KOHLER PARKER 1986; JUDGE SEBASTIAN 1988); later on easy-to-use tools were also compiled for the modelling (WESCOTT BRANDON 2000). Predictive models actually received their working base with the creation of GIS, which was necessary for their applicability. It was equally difficult to explain the theory of statistical employment and to introduce the results; the possibility of spatial representation was a huge leap either way. In the 1990s, the rapid development of GIS occurred parallel to the spreading of the postprocessual approach within archaeology. Most scientific criticism aimed at predictive models, according to P. Verhagen, actually stems from the processual 13 postprocessual contrast, where the quantitative approach overly simplifies the requirements of habitation (VERHAGEN 2007, 16). From this time on the opinions on predictive models became somewhat divided. Its professional reception was quite low, with continuous criticism as the basis of the analyses was not sufficiently supported by scientific evidence. At the same time, it was a readily usable, practical tool for CRM in order to draw the attention of decision makers to the actual number of archaeological risks at hand. The archaeological agencies of the USA and Canada use it as an everyday tool in administrative and planning tasks even today. In Europe we can reckon with the large demand of heritage protection tasks previously unaccounted for starting from the Malta Convention in 1992, which resulted in the creation of CRM. The first predictive models were prepared in connection with this, which had the most impact on the archaeology of the Netherlands. The Netherlands uses the only nationwide model that was created in 1997 (Indicatieve Kaart van Archeologische Waarden IKAW) that is continuously developed through a wide scientific cooperation (VERHAGEN 2007): they are now using a third generation model for public administration today.
The structure and methodological background of predictive models
The comprehensive collection and handling of data at motorway excavations involving vast areas could only be managed with the help of geographic information systems, which meant a significant and partially constrained shift of paradigm compared to the previous small surface excavations. Precisely such a leap in scale can be observed in field survey site locations, as the surface data of large areas can only be handled and employed by geographic information databases. Just as in the case of the geographic informational background of excavations, the most exact spatial location of surface finds and sites also obtains a significant role. As we have already mentioned above, we have located about 20 30% of Hungarian archaeological sites in the past 50 years. The accuracy of these data is to be questioned, moreover, simply because the surveys were by necessity usually conducted on a rough estimate without the help of geodesic tools (GPS). With field surveys only even if the systematic studies discontinued since the past 15 years would advance more rapidly a long time would elapse until the majority of 14 archaeological sites could be identified and protected. Predictive modelling in contrast offers an alternative that can draw an exact, reliable and approximate picture of the archaeologically concerned areas of Hungary. During predictive modelling, the spatial location probability of new, thus far unknown sites can be determined in a given region by considering the environmental characteristics of the archaeological sites (human settlement) already known. Today this process is usually defined by statistical methods. However, we need to briefly review the techniques employed during modelling before outlining the environmental factors, as various new methods have appeared following almost 30 years of development. A significant difference in the case of predictive models is whether the known archaeological data are to be used or not. This helps in separating the inductive models employing archaeological data and the deductive models that do not. Due to lack of space we cannot compare the positive and negative aspects of the two model types, but we can state that here in Hungary the inductive models have more possibilities, thus we will be dealing with these in the following. The most important characteristic of the initial model types with dichotomous variables (Boolean overlay, weighted binary addition) was that the modelling specialist could determine which environmental conditions, in which zone would predict the location of unknown archaeological sites (EJSTRUD 2003, 128-129). Nowadays, the base of modelling is placed on modern methods of mathematics, such as the Bayesian probability theory, the Dempster-Shafer theory based on the latter, the Weights of Evidence, and fuzzy logics (for an overview of their archaeological use cf. EJSTRUD 2003; NYKNEN SALMIRINNE 2007). During their use, when choosing the environmental factors a mathematical (function) relation is determined between the location of the known archaeological sites and the environmental factors; thus the indicators not showing significant relations can be filtered out before the modelling itself. On the other hand, accurate probability rates and reliability values can be determined by these methods, which are important for measuring the stages of models. The most variation between models appears in the choice of environmental factors. There is no prescription for these choices; the variations in natural geography regarding each studied area call for different processing methods and data. As it was seen earlier, these environmental factors are not unknown in Hungarian research either. Therefore, in the following we will summarize the characteristic environmental factors employed and the geographical information to be derived from these. 15 One of the basic requirements for human settlement is the proximity of water, whether still or flowing. As we have already stated above, an archaeologically accurate hydrographic map can be obtained if we do not consider hydrographic changes following the archaeological periods; and furthermore, incorporate dry watercourses into our database. There are three methods to outline it. Firstly, each change can be defined with the help of old maps; however, this is lengthy and costly work with large areas involved. Secondly, the opportunity arises to generate outflow models based on the (improved) relief model (TELBISZ 2007). Thirdly, experience (MESTERHZY 2011; PADNYI-GULYS 2011) shows that the layers of geographic maps related to flowing and still waters can also give a good point of departure. The distance from water or flooded areas can be employed as coverage for geographic information, which can be defined generating zones; in our experience the depth of ground water levels can also be an important factor. Geographical and pedological data can be employed as well; and not just in predicting the flooded areas already mentioned. It is absolutely necessary to simplify these data, and to merge their categories, for example according to characteristic rock-forming minerals, to enable them to be authoritative information resources that can be managed by statistical analyses. The special role of geographical data is the identification of temporary, former watercourses, as we have already discussed in detail. Digital elevation models (DEM) are available either free (SRTM, ASTER GDEM), or can be bought (the products of The Institute of Geodesy, Cartography and Remote Sensing and the MoD Mapping Non-Profit Ltd in Hungary). They fundamentally influence the achievement of predictive models. This is caused by the fact that on the one hand, the freely available elevation models (DEM) are actually digital surface models (DSM). That is, the DSMs produced by remote sensing methods contain not only the relief features of the terrain, but also incorporate all the features, buildings, vegetation and their height values are also added to the heights of the relief map. Nevertheless, whichever elevation model will be used its precision measurements and resolution must be well considered, as it gives the lower end of the scale of the modelling as well. Usually three coverages from the relief data can be built into the predictive archaeological model: exposure, gradient categories and individual surface forms; but several other modes of application are also possible. The choice of environmental factors and their categorization requires varying values whether in flat, hilly or mountainous areas. Thus, such near homogenous units need to be found in an area where the coverages do not have extreme values. The Cadastre of the Microregions of Hungary (MKK 2010) offers precisely such delimitations for the whole of 16 Hungary, and provides the scale of modelling too for microregions usually 100500 km 2 that have a sufficient number of sites for the building of a substantial model (STIBRNYI 2010, 352). The end-product of modelling is a raster image file where fixed probability and reliability values are determined for its basic construction units: the pixels. These values can be categorized and result in a clear, colour-coded map, where three or four zones can be identified to indicate the very low, low, middle, and high archaeologically significant areas. Yet, it is important to emphasize that the actual site expansion cannot be read from such maps, only the probability values showing the probability of the location of archaeological sites.
Concerns about predictive models
Archaeological predictive modelling has divided the international archaeological community since its beginnings, stirring up heated debate within scientific circles. The main objections have been summarized by P. Verhagen in the following four points (VERHAGEN 2007, 17): 1) use of incomplete archaeological datasets; 2) the biased selection of environmental parameters, often governed by the availability of cheap datasets such as DEM; 3) as a consequence, a neglect for the influence of cultural factors, both in the choice of environmental parameters, as well as in the archaeological dataset; 4) and lastly, a neglect of the changing nature of the landscape
The archaeological database currently available and the archaeological site register in it, as we have already mentioned, is far from complete. Nonetheless, this idea would hold true for the comprehensive processing of any archaeological object or feature type; while the goal of modelling is non other than to serve data for these unplumbed areas of archaeology. The accuracy of inductive models is defined by the accuracy of the worst resolution coverage. One of the primary viewpoints of selecting environmental factors besides the quality of data is that the piece of information should be available in digital format or able to be digitalized. The incentives behind it are naturally to cut expenditures, on the one hand; and 17 as we have mentioned with regard to environmental archaeology above, on the other hand, the textually formatted environmental reconstructions are difficult to form into geographic informational data and thus incorporate a significant amount of uncertainty. One method of defining cultural effects, and culturally specific characteristics of a given period or culture is precisely predictive modelling, inasmuch as we can consider the differences between the models building on periods when creating a complex model. There are a growing number of methods to define natural changes in the landscape. The elimination of the effects of erosion (CENTERI 2001, 12-14), the definitions of the change in vegetation (BR 2006; BR et al. 2010) and vertical surface movements (PAPP et al. 2005) are all steps towards achieving this. Yet we must realize that however much these researches help in making the image of the natural environment more accurate in a given period, they will not be complete or we cannot be sure about their completeness, just as we cannot know all the archaeological sites in Hungary. Of course, this does not mean we should not strive to achieve it.
Applying predictive archaeological models in Hungarian CRM
Archaeological studies also regularly make use of models specific to periods or predicting features (e.g. GRAVES 2011). Academic research, however, is rarely interested in every feature, much rather the individual feature types or cultures are focused on. Nevertheless, the models are primarily connected to CRM, already in connection with their creation, usage and development. They are suitable for producing reliable predictions on relatively large, archaeologically less known areas. This should not be confused with site predictions because, as we have mentioned, the models do not tell us where the sites are, but where they are highly likely to be located. During the planning phase of investments it is equally important for both the archaeologist and the investor to avoid archaeological sites. Naturally, from the investors point of view this is only a single phase of the whole planning but archaeology can only be of use in this respect if the construction is planned on a known archaeological site. Due to the low level of known sites, the risks of heritage protection are like time-bombs to all construction works, which after detonation causes damage to the investor, the site, and the social status of archaeology as well. A good predictive model can minimize this risk by 18 offering an overlay that indicates this to the investor or planner. This can be regarded as just another overlay among the overlays of risks to be considered, but as the most important characteristic: it can be anticipated and can give quantifiable predictions. The employment of models can thus be pivotal, on one hand during the planning phase of investments as, if other terms permit it, the constructions can simply be re-planned to an area of lower risk in the knowledge of the predictable risks. This has a further advantage in real estate developments: predictions can be made without the investors disclosing information about the location of the planning, which is often handled as an anxiously protected trade secret in the initial phase. Currently, if an investor was to call on an archaeologist to ask for advice on such matters, he or she would not be able to help without the knowledge of the exact location. In the case, however, when a predictive model would be available for the area chosen (microregion), then the basic heritage risks can be clarified without knowing the actual (lot number of the) location. The model may help prior to construction as well, calling attention to the areas where for various reasons no sites have been identified but these can be expected (e.g. at motorways). Calculating with such areas, the amount of time and expenses needed for the archaeological excavation can be estimated more precisely. The model can also be of immense use for the archaeological tasks during construction, pointing out those regions where archaeological monitoring should be more active and earthworks more careful as we can expect to find unforeseeable archaeological features. The model offers further assistance during everyday heritage protection. Whilst the field surveys compulsory for the heritage protection of large-scale investments signify a considerable amount of security in the case of unknown sites, this is not mandatory with other investments; thus, unknown sites are practically not protected at all under these circumstances. When the archaeological inspector does not have sufficient data to specify any type of preliminary measures, the regional museum authorized to conduct excavations can only acquire information about unexpectedly appearing sites by chance. However, should there be a predictive model available of the area in question, then the archaeological inspector can use this data for constructions planned in zones of high predictability to at least prescribe monitoring, which can already assure archaeological presence and protection. Incidentally, this is also granted by law to enable the protection of areas archaeologically affected. Models are such an important part of administrative procedures in heritage protection in the Netherlands that excavations have to be conducted at locations of high probability prior to constructions even when no actual site is known there. 19 Interestingly enough, Hungarian legislative regulation is familiar with the category of highly probable areas defined in predictive models, which was added to Act LXIV in 2001 as an influence of the Dutch example: this is the category of area of archaeological concern (Katalin Wollk, pers. comm.). According to this all areas at which archaeological sites can be predicted are to be regarded as areas of archaeological concern. 9 The employment of the model could fill this phrase with meaning, defining exactly which areas count as archaeologically concerned and on what grounds. Although the models can and should be continuously developed it would be nave to expect the model to be infallible. Nevertheless, even the worst model offers more protection for unknown sites than the present practise. Moreover, its values can even be set to be less precise but more accurate (WHITLEY 2004, 238). That means that the high probability zone will appear on a larger scale at the surveyed area, but the majority of the sites will be included within it. Hidden sites pose as major problems for heritage protection. These are features that cannot be observed through surface surveys either because they cannot be found by field surveying methods (e.g. row cemeteries), or because they are covered (by forests, pastures, architecture, etc.). Prior to earthworks, the identification of site remains on such areas is usually very expensive, but the model draws attention to the risks affecting these locations as well.
Possible considerations of modelling
The Hungarian microregional system could be a good base for the implementation of models. As mentioned above, these microregions form an excellent possibility for delimitation in connection with creating models for the description of homogeneity. We estimate that the 230 microregions would fit into more or less 20-30 separate models. Each model is exact inasmuch as the data entered are, but our experience shows that archaeological data is the least accurate of these (MESTERHZY 2011; PADNYI-GULYS 2011). Thus, each modelling period should be preceded by a field survey during which 20-30% of the area is thoroughly surveyed, where each feature observable on the surface is identified by GPS. This base, including the archaeological excavations conducted earlier on the site, can be extrapolated later on. As referred to above, we have prepared the experimental models of two microregions so far. Based on our experience, one microregional model can be prepared in
9 Act LXIV 7 of 2001 (17) about the protection of cultural heritage. 20 about a month involving 8 people (including field surveys and monitoring). Models can (and should) be continuously improved with the development of both the input data and the know- how, and will be increasingly more accurate by channelling the results of development in remote sensing (LiDAR, hyperspectral images). This, however, requires a base that can be developed. The tools of archaeology are developing at an increasing speed providing more and more possibilities; while putting more responsibility on archaeologists shoulders to protect archaeological heritage with all methods available. Predictive models form a cost-effective and highly accurate, continuously developing part of these tools. The application of predictive archaeological models and the field surveys are not two opposite methods, both can be organically integrated into each other. In any systematic field survey we are to conduct, we must always consider those areas where perception is not possible. Models can predict the heritage risks concerning these areas, moreover, any further archaeological examination will strengthen the precision of the model. We are convinced that in the current situation of Hungarian archaeology predictive models offer such a possibility for heritage protection, which enables responsible safeguarding that can be continuously developed. Irodalom
ARCANUM 2004
Az els katonai felmrs Magyar Kirlysg 1782-1785. [First Ordnance Survey Kingdom of Hungary] Arcanum DVD-ROM, Budapest. ARCANUM 2006a
Msodik Katonai Felmrs: Magyar Kirlysg (s a Temesi Bnsg) - georeferlt. [Second Ordnance Survey Kingdom of Hungary (and the Bnsg of Temes) - georeferenced] 1806-1869. Arcanum DVD-ROM, Budapest. ARCANUM 2006b
MOL Trkptra I. Kamarai trkpek [MOL Map Collection I. Maps of the Chamber] 17471882. Arcanum DVD-ROM, Budapest. ARCANUM 2006c
A MOL trkptra II. Helytarttancsi trkpek [MOL Map Collection II. Maps of the Procuratorships Council] 17351875. Arcanum DVD- ROM, Budapest. ARCANUM 2007
Harmadik katonai felmrs 1869-1887, a Magyar Szent Korona Orszgai, 1:25.000 georeferlt [ Third Ordnance Survey 1869-1887 21 georeferenced]. Arcanum DVD-ROM, Budapest. ARCANUM 2009
Magyarorszg megyetrkpei a Hadtrtneti Trkptrban [Countymaps of Hungary from Military Historical Map Collection] 17311948. Arcanum DVD-ROM, Budapest. ARC-WOFE 1998 Arc-WofE User Guide. http://www.ige.unicamp.br/wofe/documentation/wofeintr.htm BNFFY 2009 Bnffy E., Fldinduls? A rgszet vlsga s vltozsa napjainkban. [Earthquake? The crisis and change of archaeology today] In: Archeometriai Mhely 2009/1. 1-3. BTKY-GYRFFY- VISKI 1941 Btky Zs. Gyrffy I. Viski K.: A magyarsg nprajza [Ethnography of Hungary] I-IV. Budapest. BR 2006 Br M.: Trtneti vegetcirekonstrukcik trkpek botanikai tartalmnak foltonknti gazdagtsval. [Historical vegetation reconstructions with the method of teaching the maps] Tjkolgiai Lapok 4. 357-384. BR et al. 2010 Br M. Horvth F. Blni L. Molnr Z.: Vegetcis adatbzisok s a CORINE felsznbortsi trkp szintzisnek mdszertani krdsei az Ipoly-vzgyjt nvnyzeti trkpe kapcsn. [Synthesis of different habitat databases and CORINE landcover map methodological approaches and a regional scale case study]. In: Tjkolgiai Lapok 8. 607-622. CHISHOLM 1962 Chisholm, M.: Rural settlement and land use: an essay in location. Hutchinson University Library, London. CLARKE 1977 Clarke, D.L.: Spatial archaeology. Academic Press, London. DJURI et al. 2009 Djuri, B. Mason, P. Mlakar, B. Kovaec Nagli, K. Petek, B.: Listing archaeological sites heritage: the case of Slovenia. In: Schut, A. C. P. (ed): Listing Archaeological Sites, Protecting the Historical Landscape. EAC Occasional paper no. 3. 87-94. EKE et al. 2007 Eke I. Frankovics T. Kvassay J.: Els tapasztalatok a nagyfellet rgszeti feltrsok trinformatikai feldolgozsa sorn Zala megyben. [Primary experiences in processing geographic information of large- scale excavations of Zala County] Zalai Mzeum 16. 259-269. FEKETE et al. 2005 Fekete Cs. Honti Sz. Horvth F. Jankovich-Bsn. D. Korom A. 22 Klt L.: Terepbejrsok Somogy megyben 19992004. [Field walkings in Somogy County] In: Kisfaludi J. (szerk.): Rgszeti kutatsok Magyarorszgon 2004-ben. Budapest 91120. FEKETE 2008 Fekete Cs.: Predictive archaeological modelling in Somogy county. Somogy Megyei Mzeumok Kzlemnyei 18. 147-156. FISCHL 2006 P. Fischl K.: rokt-Donghalom bronzkori tell telep. [Bronze age tell settlement in rokt-Donghalom] Borsod-Abaj-Zempln megye rgszeti emlkei 4. Miskolc. FISCHL 2008 P. Fischl K.: Adatbzis a Borsodi Mezsg s a Borsodi rtr prediktv rgszeti lelhely meghatrozshoz. [Database for the predictive determination of archeological sites in the Mezsg of Borsod and the flood basin of Borsod] DVD-ROM. Herman Ott Mzeum, Miskolc. FZESI 2009 Fzesi A.: A neolitikus teleplsszerkezet mikroregionlis vizsglata a Tisza mentn Polgr s Tiszacsege kztt. [Microregional study of the neolithic settlement pattern between Polgr and Tiszacsege] Tisicum 19. 377-398. GL et al. 2005 Gl E. Juhsz I. Smegi P. (eds.): Environmental archaeology in North-East Hungary. Varia Archaeologica Hungarica XIX. Budapest. GRAVES 2011 Graves, D.: The use of predictive modelling to target Neolithic settlement and occupation activity in mainland Scotland. Journal of Archaeological Science 38. 633-656. GILLINGS 2007 Gillings, M.: The Ecsegfalva Landscape: Affordance and Inhabitation. In: Whittle, A. (ed.): The Early Neolithic in the Great Hungarian Plain: investigations of the Krs culture site of Ecsegfalva 23, Co. Bks. 31- 46. GYUCHA DUFFY 2008 Gyucha A. Duffy, P. R.: A Krs-vidk holocn kori vzrajza. [Holocene hydrology of the Krs area] In: Gyucha A. (szerk.): Krs- menti vezredek Rgszeti kolgiai s teleplstrtneti kutatsok a Krs-vidken. Gyula. 11-40. HOLL PUSZTAI 2011 Holl B. Pusztai T.: Trinformatika alkalmazsa a rgszeti feltrsokon [GIS applications on archaeological excavations] In: Grf P. Horvth F. Kulcsr V. F. Romhnyi B. Tari E. T. Bir K. (szerk.): Rgszeti Kziknyv CD ROM, Budapest, 321-376. 23 HODDER ORTON 1976 Hodder, I. Orton C.: Spatial analysis in archaeology. Cambridge. JANKOVICH-BSN NAGY 2004 Jankovich-Bsn D. Nagy M.: Felmrs a rgszet helyzetrl 1989 1999. [Survey on the state of archaeology 1989-1999] Budapest. JUDGE SEBASTIAN 1988 Judge, J.W. Sebastian, L. (eds.): Quantifying the Present and Predicting the Past: Theory, Method and Application of Archaeological Predictive Modelling. Denver. KERTSZ SMEGI 1999 Kertsz R. Smegi P.: Terik, kritika s egy modell: Mirt llt meg a Krs-Starevo kultra terjedse a Krpt-medence centrumban? [Theories, critiques and a model: why did the expansion of the Krs- Starevo-culture stop in the centre of the Carpathian Basin?] Tisicum 11. 9-23. KERTSZ SMEGI 2003
Kertsz R. Smegi P.: skrnyezeti tnyezk s a Krpt-medence neolitizcija: egy j geoarcheolgiai modell nhny aspektusa. [Palaeoenvironmental factors and the neolithization process of the Carpathian Basin: some aspects of a new geoarchaeological model] MFM Studia Archaeologica 9. 25-37. KING 2004 King, T. F.: Thinking about cultural resource management. Altamira Press, Walnut Creek. KOHLER PARKER 1986 Kohler, T.A. Parker, S.C.: Predictive models for archaeological resource location, In: Schiffer, M.B. (ed.): Advances in Archaeological Method and Theory, Vol. 9. New York, 397-452. MEHRER WESCOTT 2004 Mehrer, M. Wescott, K.: GIS and archaeological site location modelling. London. MENDL 1932 Mendl T.: Tj s ember. Az emberfldrajz ttekintse. [Landscape and man. Overwiev of the human geography.] Budapest MRI 1952 Mri Istvn: Beszmol a Tiszalk-Rzom s a Trkeve-mrici satsok eredmnyeirl. [Report on the results of the excavations in Tiszalk-Rzom and Trkeve-Mrichalom] Archrt 79. 204-226. MESTERHZY 2011 Mesterhzy G.: Prediktv rgszeti modellek magyarorszgi alkalmazhatsgnak lehetsgei. Szakdolgozat [Possibilities in the application of archaeological predictive models in Hungary, BSc Thesis] NYME-GEO. 24 MKK 2010 Dvnyi Zoltn (szerk.): Magyarorszg kistjainak katasztere. [Gazetteer of the subregions of Hungary] Budapest. NORMAN SOHLENIUS 2009 Norman, P. Sohlenius R.: ASIS more than a register of ancient monuments. In: Schut, A. C. P. (ed.): Listing Archaeological Sites, Protecting the Historical Landscape. EAC Occasional paper no. 3. 83- 86. PADNYI-GULYS 2010 Padnyi-Gulys G.: Prediktv modellezs a Zsmbki-medencben. [Predictive modelling int he Zsmbk Basin] http://www.agt.bme.hu/szakm/software/pgg_terinfo_szoftverism_01.pdf PADNYI-GULYS 2011 Padnyi-Gulys G.: Rgszeti cl prediktv modellezs a Srvz vlgyben. Szakdolgozat [Archeological predictive modelling in the Srvz valley, BSc Thesis] BME. PAPP et al. 2005 Papp B. dr. Jo I. Balzsik V.: A fggleges felsznmozgsok vizsglata s modellezse a Duna-Tisza kzn Kecskemt- Kiskundorozsma. [Investigation and modelling of the recent vertical movements int he space between Danube and Tisza river] Geodzia s Kartogrfia 2005/6 11-18. PRINKE 2009 Prinke, A.: Seven years after Seville: Recent progress in managing the archaeological heritage in Poland. In: Schut, A. C. P. (ed.): Listing Archaeological Sites, Protecting the Historical Landscape. EAC Occasional paper no. 3. 71-76 PUSZTAI 2005
Pusztai T.: The archaeological investigation of Kelemr-Mohosvr and the medieval settlement history of the Kelemr area. In: Environmental Archaeology in North-Eastern Hungary. Szerk.: Gl E. Juhsz I. Smegi P. Budapest. 411-424. PUSZTAI 2007b
Pusztai T.: A kelemri Mohosvr Egy 13-14. szzadban hasznlt vr kutatsnak lehetsgei. [Mohosvr at Kelemr Possibilities of the investigation of a 13-14th century castle] Castrum 5. 39-64. RACZKY et al. 2002
Raczky P. Meier-Arendt, W. Anders A. Hajd Zs. Nagy E. Kurucz K. Domborczki L. Sebk K. Smegi P. Magyari E. Sznt Zs. Gulys S. Dob K. Bcskay E. T. Br K. C. Schwartz: Polgr-Csszhalom (1989-2000): Summary of the Hungarian-German Excavations on a Neolithic Settlement in Eastern 25 Hungary. In: R. Aslan S. Blum G. Kastl F. Schweizer D. Thumm. (eds.): Mauerschau Festschrift fr Manfred Korfmann. Remschalden-Grunbach. 833-860. RACZKY 2006 Raczky P.: A rgszeti lelhely fogalmnak tudomnyfilozfiai alapon trtn axiomatikus meghatrozsa. [The axiomatic definition of the concept of archaeological site based on scientific philosophy] Archrt 131. 246248. REMNYI STIBRNYI 2010 Remnyi L. Stibrnyi M.: Rgszeti topogrfia ugyanaz msknt. [Archaeological topography the same otherwise] In: Fl vszzad terepen Tanulmnyktet Torma Istvn tiszteletre 70. szletsnapja alkalmbl (megjelens alatt) SALISBURY 2008
Salisbury, R. B.: Az Alfld ks neolitikus s kora rzkori teleplsszerkezetnek vizsglatai frsadatok elemzsvel: A Neolithic Archaeological Settlements of the Beretty-Krs Project (NASBeK) elzetes eredmnyei. [Investigating Settlement Structure through Soil Chemistry on Late Neolithic and Early Copper Age sites in the Krs- Beretty Region. The NASBeK Project.] In: Gyucha A. (szerk.): Krs- menti vezredek Rgszeti kolgiai s teleplstrtneti kutatsok a Krs-vidken. Gyula. 41-64. SRTM 2006 Jarvis, A. - Reuter, H. I. Nelson A. Guevara E.: Hole-filled seamless SRTM data V3, International Centre for Tropical Agriculture (CIAT), http://srtm.csi.cgiar.org. STIBRNYI 2010 Stibrnyi M.: Gondolatok a rgszeti topogrfia lehetsgeirl. [Thoughts on the possibilities of archaeological topography] In: Kvassay J. (szerk.): vknyv s jelents a Kulturlis rksgvdelmi Szakszolglat 2008. vi feltrsairl. Budapest. 343359. SMEGI 2007
Smegi P.: The vegetation history of the Srkeszi area. In: Zatyk Cs. Juhsz I. Smegi P. (eds.): Environmental Archaeology in Transdanubia. Varia Archaeologica Hungarica XX. Budapest. 377-384. SMEGI 2009
Smegi P.: Ember s krnyezet kapcsolata a kzps bronzkorban: az skori gazdasgi tr fejldse egy bronzkori tell geoarchaeolgiai s krnyezettrtneti feldolgozsa nyomn. Az Alfld ks neolitikus s kora rzkori teleplsszerkezetnek vizsglatai frsadatok 26 elemzsvel: A Neolithic Archaeological Settlements of the Beretty- Krs Project (NASBeK) elzetes eredmnyei. Tisicum 19. 457-480. SMEGI et al. 2003
Smegi P. - Kertsz R. - Tmr G. - Hebrich K.: Palaeoenvironmental factors and Neolithization process of the Carpathian Basin: some aspects of a new geoarchaeological model. In: Cattelain, P. (ed.): Acts of the XIVth Union Int. of Prehistoric and Protohistoric Sciences, Lige, Belgium, 2-8 September 2001, Section 1: Theory and methods. BAR Series 1145. Oxford. 135-141. SMEGI et al. 2007
Smegi P. Bodor E. Juhsz I. Hunyadfalvi Z. Herbich K. - Molnr S. Tmr G.: A Balaton dli partjn feltrt rgszeti lelhelyek krnyezettrtneti feldolgozsa. [Environmental historical processing of the excavated sites in the southern bank of Balaton] In: Kiss V. Belnyesy K. Honti Sz. (szerk.): Grdl id - Rgszeti feltrsok az M7-es autplya Somogy megyei szakaszn Zamrdi s Ordacsehi kztt. Budapest. 241-253. SMEGI MOLNR 2004
Smegi P. - Molnr S.: The Kiri-t meander: sediments and the question of flooding. In: Whittle, A. (ed.): The Ecsegfalva project. Varia Archaeologica Hungarica XXI. 67-82. SZAB et al. 1997a
Szab M. Guillaumet, J. P. Kriveczky B.: SajpertiHossz-dl. In: Raczky P. Kovcs T. Anders A. (szerk.): Utak a mltba Az M3-as autplya rgszeti leletmentsei Paths into the Past. Rescue excavations on the M3 motorway. Budapest. 81-86. SZAB et al. 1997b
Szab M. J.-P. Guillaumet Kriveczky B.: Polgr-Kirly-rpart. In: Raczky P. Kovcs T. Anders A. (szerk.): Utak a mltba Az M3-as autplya rgszeti leletmentsei Paths into the Past. Rescue excavations on the M3 motorway. Budapest. 87-89. TELBISZ 2007 Telbisz T.: Digitlis domborzatmodellekre pl csapadk-lefolys modellezs. [Precipitation runoff modelling based on digital elevation models] http://www.fw.hu/geografus2008/ix_felev/hm/hm.doc TMR 2004
Tmr G.: Space and GIS technology in Paleoenvironmental Analysis (Old maps, satellite images and digital elevation models in archaeology). Anteus 27. 135-144. TOLNAI 2009 Tolnai K.: Rgszeti feltrs trinformatikai tmogatsa. [GIS 27 supported archeological analysis] Geodzia s Kartogrfia 61/2 (2009/2). 23-27. VARGA 2000
Varga A.: Coring results at Szzhalombatta-Fldvr. In: Poroszlai I. Vicze M. (eds.): SAX Szzhalombatta Archaeological Expedition Annual Report 1. Field Season 1998. Szzhalombatta. 75-81. VERHAGEN 2007 P. Verhagen (ed.): Case studies in archaeological predictive modelling. Leiden. VERHAGEN et al. 2010 Verhagen P. Kamermans, H. Leusen, M. v. Ducke B.: New developments in archaeological predictive modelling. In: Bloemers, T Kars, H. Valk, A. v. d. Wijnen, M. (eds): The Cultural Landscape & Heritage Paradox. Protection and Development of the Dutch Archaeological-Historical Landscape and its European Dimension. Amsterdam, 431-444. WARREN ASCH 2000 Warren, R. E. Asch, D. L.: A predictive model of archaeological site location in the eastern Prairie Peninsula. In: Wescott, K. L. Brandon, R. J. (eds.): Practical applications of GIS for archaeologists: a predictive modelling kit. London. 5-32. WESCOTT BRANDON 2000 Wescott, K. L. Brandon, R. J. (eds): Practical Applications of GIS for Archaeologists. A Predictive Modeling Toolkit. London, 2000. WHITLEY 2004 WhitLey, T. G.: Causality and Cross-purposes in Archaeological Predictive Modeling. In: Fischer Ausserer, A. Brner, W. Goriany, M. Karlhuber-Vckl, L. (eds.): Enter the past: the E-way into the Four Dimensions of Cultural Heritage. Computer Applications and Quantitative Methods in Archaeology 2003. BAR International Series 1227. Oxford. 236239. WOLF 2004
Wolf M.: A felszsolcai fldvr s krnyezetnek trinformatikai rekonstrukcija. [GIS reconstruction of the hillfort of Felszsolca and its environs] Archrt 127. 89-101. WOLLK 2004 Wollk K.: The Protection of Cultural Heritage by Legistlative Methods in Hungary. In: Jerem, E. Mester, Zs. Benczes, R. (eds.): Archaeological and Cultural Heritage Preservation within the light of new technologies. Selected papers from the joint Archaeolingua - EPOCH Workshop. 27. Sept 2. Oct 2004. Szzhalombatta, Hungary. 28 73-82. WOLLK 2009 Wollk K.: Listing precondition of protection? In: Schut, A. C. P. (ed.): Listing Archaeological Sites, Protecting the Historical Landscape. EAC Occasional paper no. 3. 53-62. ZATYK et al. 2007
Zatyk Cs. Juhsz I. Smegi P. (eds.): Environmental Archaeology in Transdanubia. Varia Archaeologica Hungarica XX.