statistical tools in landscape archaeology

7/29/2019 Statistical Tools in Landscape Archaeology

1/18

339

Archeologia e Calcolatori21, 2010, 339-356

STATISTICAL TOOLS IN LANDSCAPE ARCHAEOLOGY

1. I

Recent evolution of the Romanian national policy regarding the pro-tection of cultural heritage has increased the general interest in organizinginformation about archaeological sites and monuments in georeferenceddatabases. The main result of this systematization was aimed at monitoringthe interaction between archaeological heritage and urban development.However, a better use of these data management systems through Archaeo-logical Predictive Modeling could contribute both to the improvement ofthe heritage management activity and to an advance in scientic research.Our aim here is to discuss the intermediate process, which should ensure thetransition from tables and lists of archaeological discoveries to meaningfulconnections and interactions, using Landscape Archaeology methods andstatistical algorithms.

As GIS applications in archaeology were classied by A(1992) and K (1999), Archaeological Predictive Models (APMs) rep-resent an important evolution of spatial integrated databases of archaeologi-cal records. Most of the development of these techniques has taken place inNorth American archaeology, where the spatial extent of some archaeologicallandscapes was difcult to survey in an inclusive and efcient manner. Re-cently, APMs were also applied in the national management of archaeologicalresources of some European countries, in particular the Netherlands (L, K 2005; V 2007; K, V L,V 2009), and Great Britain (R, B 2004).

APM correlates the distribution and density of certain categories ofarchaeological sites discovered in a particular geographic region with theenvironmental conditions of their location and surroundings. Therefore,it predicts with a variable probability the location of analogous sites byidentifying similar environmental conditions with those quantied for thediscoveries already known. This kind of model is called inductive or cor-relative. There is another type of approach, the deductive model, whichveries a behavioral hypothesis on the existing data sets focusing on thesame assumption that the sites emplacement is connected with environ-mental features.

Our aim is to assess the evolution of archaeological data sets into APMsand to reconsider the real value of such attempts for Romanian heritage pro-tection and for scientic purposes. We will consider as well certain aspectsregarding the deductive/inductive nature of the APMs.


2/18

D. tefan, V. Srbu

340

2. E

We will use as a case study the database of sites of a certain administra-tive region of Romania called Buzu, for which we actually contributed in thebuilding and exploitation of the GIS. Even if the modern area of the BuzuCounty does not represent a prehistoric or ancient reality, being historicallyand culturally related to the neighboring counties of Braov, Prahova, Covasna,Brila and Vrancea, some arguments may support the scientic relevance ofsuch an attempt, besides, of course, the administrative advantages in manag-ing the regional heritage.

The region is characterized by the hydrographic basin of the river Buzu,which crosses over the entire county from its mountainous springs to its mid-

dle course in the plains (Fig. 1). Therefore, Buzu County is quite interestingfor a Landscape Analysis since it contains all the known relief shapes, fromthe 1700 m high peaks of the Curved Carpathian Mountains, hilly terrains,to ooded marshes and large sandy river valleys, distributed in a balancedcoverage of the area. At the same time, the registered archaeological remainsbelong to all prehistoric and historical epochs and represent both excavatedsites and surface discoveries.

Buzu area has been used since prehistory as the main passage routebetween three important geographic and cultural areas: the Danube Valleyto the S, the North Pontus steppes to the E, and Transylvanian Plateau to theN (Fig. 1). This important functionality ensured the cultural homogeneity ofthe archaeological discoveries in the region which showed a strong mixtureof various neighboring inuences.

Surface deposits of salt (Loptari, Bisoca, Rmnicu Srat, Bdila,Mnzleti, Brtileti, Goideti), mineral waters (Srata Monteoru, Bozioru,Fisici, Balta Alb, Siriu, Nehoiu, Loptari) and amber (Mljet, Sibiciu de Sus,Coli, Bozioru, Plotina, Terca) represented important local natural resourcesthat inuenced the land-use patterns and distribution of sites.

3. P

During the Neolithic era, the sub-Carpathian area received inuencesfrom many directions, mainly from the southern and north-eastern regions,establishing complex interactions with the intra-Carpathian area as well (P- 1999; F 2007). This makes the identication of land-usepatterns rather difcult, as they cannot easily be considered relevant in everysituation. The majority of the discovered archaeological sites belonging tothe Neolithic period were settlements located in the plain, along river valleys:Smeeni, Costeti-Pietrosu, Sudii. Several sites were located in higher areas suchas hilly plateaus: Aldeni, Blneti, Fulga, Vadu Soreti. During the Eneolithic


3/18

Statistical tools in Landscape Archaeology

341

era, the number of known sites increases. During the Eneolithic period, hu-man communities inhabited the regions located S to Buzu River, sometimesin tell-settlements. Typical sites were identied at: Aldeni, Gherseni, Sudii,Nicov, Gura Vitioarei, Srata-Monteoru, Spoca, Coatcu (A et al.2008), Pietroasa Mic (S, M, D 2005), Cleti (C,C 2008), Cotorca, Lipia, Pota Clnu (F 2008, 37,tav. 2).Their material culture was recognized as a different version of Gumel-nitsa, called Stoicani-Aldeni, strongly inuenced by the northern neighboringculture of Cucuteni.

During the Bronze Age the number of sites grew signicantly, mainlyassociated with the Monteoru culture (M-C 2003). Themajority of settlements and necropolises were located on higher relief, onhilly terraces and plateaus: Neni (M-C, S 2003),Srata-Monteoru (Z 2000), Pietroasa Mic (S, M, D2005; M-C 2000), Crlomneti, Aldeni and even inmountain areas. Many settlements were fortied. The density of habitationon lower terraces suggests an extensive land use. During this period, theintensity of cultural and material changes seemed to have intensied. Therichness of these nds was connected with the idea of exploitation of localsalt resources. The eponym site of the culture, Srata Monteoru (Z2000), was located in the vicinity of a mineral spring. The site was fortiedand multi-leveled.

Even if, during Neolithic era and the Bronze Age, the Buzu area wasdensely inhabited, beginning with the First Iron Age, the demography of thisarea registered a strong set-back. From the end of the Bronze Age (easternchronology) until the 8th century BC there were no funerary discoveries. Ba-sarabi type discoveries (8th-7th century BC) were made at: Crlomanesti, Berca,Pietrosu and Izvorul Dulce (C 2008b; M 2009b). For the6th-5th centuries BC more discoveries were noted, mostly graves, belonging tothe Ferigile-Brseti Group: Neni-Colarea, Neni-Znoaga, Valea Viei andGherseni. First Iron Age discoveries were made mainly along river valleysin small mountain hollows (M 2009b).

Settlements dated to the two centuries before the Roman conquestof Dacia (which happened in 106 AD) were discovered densely amassedin the high-hilly sector of the Buzu river valley, right before its descentand enlargement in the plains. In addition, several fortresses located instrategic positions guarded the communication route through the moun-tains towards Transylvania (M 2009a). Crlomneti (B et al.2010) and Trcov (T, A 1992; M 2008) are the mostrelevant sites for this period. Pietroasa Mic-Gruiu Drii was a sacredDacian place enclosed with a limestone wall (D, S 2001; S,M, D 2005).


4/18

D. tefan, V. Srbu

342

Fig. 1 The relief and river network of the Buzu County. General emplacement of the discussedregion in the Balkan Peninsula area.


5/18


343

Fig. 2 Archaeological site Crlomneti-Cetuia, Buzu County.


6/18

D. tefan, V. Srbu

344

Fig. 3 Archaeological site Pietroasa Mic-Gruiu Drii, Buzu County.


7/18


345

Only a small number of sites were discovered in the plain which belongedto the period during which Dacia was a Roman province (2nd -3rd centuriesAD). Even if the ofcial Imperial border is believed to have passed a fewdozen kilometers further to the E, on the axis of Boroneu Mare (S1975)-Drajna de Sus ( 1948), important Roman discoveries weremade in the area of Pietroasele (C 2008a), suggesting theexistence of a fort, and a second inner line of defense, whose chronology isstill under debate.

During the Post-Roman Age, habitation intensied as revealed by thelarge number of discovered sites, both settlements and necropolises. Thesewere located along river valleys, either in the plains or in hilly regions. Thehigh density of settlements and inhumation necropolises belonging to theSntana de Mure culture points to the presence of a Gothic authority centerlocated in the area of Pietroasele. Pietroasele was the place of discovery of thefamous hoard Cloca cu puii de aur, weighing over 18 kg of gold, dated tothe beginning of the 5th century AD (O 1976).

During Early Medieval times settlements were located in higher andmore remote places on small tributaries of the Buzu River.

4. A

The data which we considered were partly processed in one of the

few implementations of the Governmental Project EGISPAT (ElectronicGIS Heritage), promoted by the Romanian Ministry of Culture, Cults andNational Heritage (http://www.inmi.ro/egispat.html). An interactive map ofarchaeological sites, based on GIS principles, was meant to be useful to theRegional Heritage Authorities for managing and protecting the monuments,but also to the general public through a web interface for knowing thelocal potential. The database of archaeological records included the ofcialregional list of protected sites and monuments (LMI 2004). Apart from sys-tematic excavations, the majority of information originated from occasionaland old surveys, not properly documented, preserved in the memory of differ-ent researchers (M. Constantinescu, G. Trohani, S. Matei). Little informationwas actually veried and updated in the eld.

However, APM was not an aim of this project, as none of the GIS typelandscape analyses (site catchment, viewsheds, least cost analysis, etc.). Animportant aspect of the EGISPAT project was to deliver the Regional HeritageAuthorities maps with sites, also containing the protected areas calculatedat various distances (buffers) around the known surface of the sites. In theseprotected areas, even if the archaeological nds were not documented, anyintervention which may affect the terrain would require expert archaeologi-cal control.


8/18

D. tefan, V. Srbu

346

At the same time, part of the information was collected in the contextof a scientic research project, ISTVB, funded by the Romanian Ministry ofEducation and Research (2007-2010) and supervised by the National Centrefor Program Management (CNMP), dealing with the study of the Buzu RiverValley as a communication passage from the southern plain towards the innerCarpathian areas, over the ages (http://istvb.net4u.ro/en_index.html). Withinthis project, mapping of sites and understanding their relation with the naturalenvironment represented the major aspects of the research. Limited systematicsurveys were made, in conjunction with systematic excavations of a few key-sites. Predictive models were desirable in the context of this scientic researchproject, in order to allow the formulation of statements regarding the ancientroutes used by people in the past in this region.

5. P . T P-R A

A sequence of predictive models was created as a rst step in the processof integrating eld-research data sets with theoretical assumptions regard-ing the correlation between past human communities and their surroundingenvironment. In the initial stage of developing this predictive model we tookinto consideration only three simple environmental variables mean eleva-tion of sites, spatial continuity within the site surface (roughness), distance

to the nearest permanent water course and two archaeological variables chronology of site, type of space use (habitation or funerary).

In the development of our model we followed three steps: 1) data collect-ing and hypotheses development; 2) development of initial models; 3) testing.As the project is still in progress, for the moment we are just presenting thepreliminary results obtained for a single chronological framework: the caseof Post-Roman Age sites (3rd-4th centuries AD).

The majority of sites belonging to this period were discovered in closeproximity to permanent rivers, along their valleys (Fig. 4, A). When mapped,only a small number of cases from the large group of Post-Roman Age sitesappeared to be located quite distant from permanent water courses. A calcula-tion of the channel network in the plain revealed however that these apparentlyunusual locations were determined by the existence of smaller, temporarywater valleys (Fig. 4, B). The normal distribution curves of mean elevationcalculated for settlements from different periods indicate differences in theheight of the occupied hill (Fig. 6). A general normal distribution curve ofthe elevations calculated for the entire Digital Elevation Model of the regionis represented for a better understanding of the human habitation choices(Fig. 6). Differences may be observed as well, in the comparative analysis ofnormal distribution curves calculated for the spatial continuity within sites


9/18


347

Fig. 4 Detail of the Buzu County map showing the distribution of thePost-Roman Age sites. Hydrological Analysis.


10/18

D. tefan, V. Srbu

348

Fig. 5 Archaeological Predictive Model of the Buzu County for the Post-Roman Age sites.

surface. A roughness index of the site surfaces was assessed statistically, us-ing a range of variograms and curvature analyses. The noticed differencesbetween these analyses are rather small, if judged as individual elements, buta combination of several variables could determine a clearer pattern of land-use (the multiplication result of several probabilities will always be smallerthan any of the individual probabilities).

The Digital Elevation Model was reclassied in three orders accord-ing to the initial model proposed after the observation of the data sets. Asimilar reclassication and correlated representation of roughness index wasperformed. The reclassied grids were afterwards reunited through Booleanprocedures in a single grid with four classes. Afterwards, we kept in the re-sulted modeled representation the intersection between this new grid and theriver valleys buffer calculated at 1000 m (Fig. 5).

The probability for the existence in the region of sites dated to the 3rd to5th centuries AD is greater in the given region as the map color is darker. Theproposed model corresponds to 95% of the registered sites in our database(eighty sites t the pattern and four of them do not, each one explainable ina certain context).


11/18


349

Fig. 6 The normal distribution curves of mean elevation calculated for sites in the BuzuCounty dated in different periods.


12/18

D. tefan, V. Srbu

350

6. R

An important element which enhances the relevance of predictivemodels is the reexive renement of data, hypotheses, interrogations, appliedalgorithms and interpretation.

Predictive models may be improved by addition of supplementary en-vironmental variables, which could constrain the variation of general values.For the case of Post-Roman Age settlement, previously described, the waterresource analysis should be completed with more data types. Water tableconditions are of great importance for choosing settlement emplacement, inthe terms of water supplies. Other relevant data should refer to the paleo-channel network. Flooded areas may benet from the use of APMs, if we

try to obtain information regarding the scale of ooding through systematiccore sampling or integration of old maps depicting lands before modern riverregularization. The changing nature of the landscape should not be seen asa menace to the APM consistency, but rather as an interesting eld to whichAPM may bring important aid. In our case, the integration into the APMsof the results of viewsheds, least cost analyses, hypotheses about circulationroutes helped us in giving additional socio-cultural meaning to the environ-mental parameters.

Another option of improving the APMs results is to better analyze theselected variables. In many cases, a simple normal distribution curve doesnot depict by itself the entire analyzed behavior, as this often bears multipleinuences. For example, a simple kernel density estimation algorithm appliedto mean elevation values of the sites emphasizes the multimodal nature ofdata sets (Fig. 7). At rst view the differences are slightly noticeable, but everyattempt for automatic classications should be checked through condencetests.

7. U

The comprehension of the real nature of the archaeological data sets

represents in our opinion the most delicate operation performed during thedevelopment of predictive models. In order to illustrate this issue, three casestudies placed within the Buzu County will be taken further into considera-tion. They represent the result of the ISTVB project.

The archaeological potential of the mountainous lands of Buzu Countyis almost unknown and hard to cross-over due to the historical developmentof the region. Until the 18th century one of the main circulation routes fromTransylvania towards the S passed through the region on ridge routes, butthe shift of the circulation towards the river valleys, which were arrangedand built as proper roads, changed signicantly our perspective about the


13/18

351

Fig. 7 Histograms showingthe frequency of mean elevationvalues for different categoriesof archaeological sites. Normaldistribution curves. Kernel DensityEstimation curves.


14/18

D. tefan, V. Srbu

352

landscape. Little archaeological information has been available for the highelevated areas, as this region has been difcult to survey.

Old maps of the area (mainly Austrian and Hungarian), used as back-ground layers in APM, offered signicant information: depiction of lost trailsand strategic passing points. Moreover, these documents represent the immensescale of landscape change, which happened in the last half a century, causedby river regularization, drainage of marshes and building of concrete roads.Therefore, the APM was improved by adding supplementary environmentalvariables deduced from the analysis of old maps, bringing at the same time aclearer social meaning to landscape use patterns. APM helped us in reevaluat-ing the importance of mountainous landscapes for the past people, not as abarrier but as a connecting bridge between spaces and cultures.

The Second Iron Age site of Crlomanesti is located on a hilly plateauelevated 20 m above the surrounding lowlands, component of the large Buzuriver (Fig. 2). Systematic excavations, undertaken over 20 years (B 1975,1977; Bet al. 2010), revealed the existence of a multi-leveled site, datedto the Bronze and Iron Age, on top of a hill, on the at plateau. The particularnature of the discoveries from the Second Iron Age led to a scientic con-troversy regarding the character of the site. Recent interdisciplinary researchchanged the scale of the analysis from the site scaled information to integratedregional information. They showed that the archaeological discoveries foundon top of the plateau represented only a part (the acropolis) of a larger site.Human presence was attested on the steep slopes of the massif as well, in theform of different ancient land modeling activities and fortication structures.Moreover, contemporaneous discoveries were made in the lowlands locatedat the foothill of the site, buried under a substantial layer of alluvial deposit,brought by historical ooding of the river Buzu.

Simulations of major ooding of the area suggested the scale of thelandscape change in the region of the site (Fig. 2, C.). In this case, we pointedout the following aspects: the irrelevancy of slope and elevation variablesand the difculty in deciding where to draw the limits of this site in order toquantify it in a GIS application in relation with the surrounding landscape.Practically, this site overlaps a contact area between different types of reliefshapes with opposing environmental features.

Another case, which may raise problems in a regular predictive model,is called Pietroasa Mic-Gruiu Drii, a multi-leveled site, with remains fromthe Eneolithic, Bronze and Iron Ages (Fig. 3) (D, S 2001; S,M, D 2005). The major feature of this context the continuity ofusing the same space over the ages emphasized the need to identify patternsof human land-use. Nevertheless, the manner in which space was occupied,used, transformed and seen was distinct for each particular epoch. In shapingthese models, the real dimensions of the sites (the same site in different epochs)


15/18


353

and the depths between which each archaeological deposit had formed wereimportant variables. The archaeological and stratigraphic information weregathered in systematic excavations of the upper-plateau and systematic andlong lasting surveys of the surrounding region combined with small controltrenches of the stratigraphy and geophysical prospecting. Additional landscapeanalyses completed the models with the cultural-social component (,D 2005).

Nowadays, the site appears to be located on a high oval shaped plateau,on top of a limestone hill, elevated 500 m above the plains placed immediatelybeneath it. A river valley (Dara) surrounds the massif on two sides. However,as excavations indicated, during the Eneolithic period, the outline of the land-scape was quite different in comparison with the modern conguration of theterrain. People lived only in the northern part of the rocky hill, at the base ofwhat was then a slope, close to the river valley. They settled on both banksof the river, overlapping only partially the nowadays recognized site surface.During the Bronze Age, the level differences between the margins of this slopewere lled up with archaeological layers. The plateau and surrounding ter-races were created through human effort for defensive and military purposes.During the Late Iron Age the visibility of the site (Fig. 3, B) recommendedit for the imposing location of a sacred place. The leveling of older layerswas performed in order to obtain the aspect of the at surface of the terrain.Last, but not least, a modern limestone quarry destroyed almost a quarter ofthe plateau, dramatically changing the terms of the real site dimensions andforms of land use.

In this case, the modern landscape will never represent, for example, thereal relation with the environment of the Eneolithic community. Moreover,each community seemed to have had different reasons for which it selected thesame particular location: the proximity to water resources for the Eneolithiccommunities, defense and strategic reasons during Bronze Age and imposingvisibility as a sacred place for the Late Iron Age people. The spatial complex-ity of this site is given by its evolution in time. Multi-leveled sites are thebest examples to sustain the variability of human choices in the matter ofenvironment and landscape and, in this particular case, we point out as wellthe issue of prehistoric large scale built-landscape.

8. C

Our attempt to develop predictive models is only in the preliminarystage, as we are still in the process of testing both the input and output data.It is probably more the case of indicative maps than real predictive maps.Nevertheless, we believe that several features regarding possible relevance andlimitations of APMs for Romanian archaeology may already be emphasized.


16/18

D. tefan, V. Srbu

354

It should be noted that these are the results of scientic research activitiesand are not directly connected with the heritage management. In our case,predictive models together with eld surveys, geophysical prospection andother landscape analyses (slope, site catchment, least cost, viewsheds) helpedus in assessing the past communities relation with the natural environment,giving us the occasion to test hypotheses about land use and to understandwhat are the most suited environmental variables to test hypotheses. In thisway, we gained important support in developing recording data strategies.

A young rescue archaeology movement, as is the case in Romania, needsto establish its own assessment strategies and APMs should not be ignored.Even if, as it was showed before, APMs suffer from major limitations, we wishto highlight the important ability of spatial statistics to impose directions foreld surveys strategies. In addition, predictive models may change the scale ofarchaeological analyses leading to a micro-regional integration of data, thuschanging the perception about the nature of sites. However, APMs should notreplace eld survey. At the same time, in order to obtain consistent modelsone should take into consideration the need to develop particular models forparticular regions. This means that experts may select environmental variablessuited for the particular geographical and archaeological context.

In the end, let us summarize what in our perspective may represent waysof improving APMs: the use of more variables, the understanding of the ana-lytical nature of data sets and of the real nature of archaeological data sets.

D Bucharest UniversityDigital Domain Ltd.

V SMuseum of Brila

Institute of Archaeology Vasile Prvan

REFERENCES

A M.S. 1992, The role of Geographic Information Systems in quantitative an-thropology, in H.D.G. M, M.S. A (eds.), The Anthropology ofHuman Behavior through Geographic Information and Analysis, New York, OxfordUniversity Press.

A R. et al. 2008, Raport preliminar privind cercetrile arheologice efectuate n tell-ulneo-eneolitic de La Coatcu, com. Podgoria, Campania 2008, Mousaios, 13, 41-50.

B M. 1975, Problmes de la chronologie de la culture gto-dace la lumire des fouillesde Crlomneti, Dacia, 19, 125-139.

B M. 1977, Statuetele geto-dace de la Crlomneti (Jud. Buzu), SCIVA, 28, 3, 319-352.B M. et al. 2010, Crlomneti, com. Verneti, jud. Buzu, Punct Cetuia, Cronica

Cercetrilor Arheologice din Romnia, Campania 2009.C M., C C.P. 2008, Rezultatele cercetrilor de suprafa de la Cleti,

com. Pietroasele, jud. Buzu, Mousaios, 13, 59-74.


17/18


355

C E.M. 2008a, antierul arheologic Pietroasele. Campania 2008, Mousaios,13, 137-162.

C E.M. 2008b, Tombe dincinration dcouverte Izvoru-Dulce-Merei, Dp.Buzu, Mousaios, 13, 103-106.D V.I., S V. 2001, Incinta dacic forticat de la Pietroasele - Gruiu Drii (I),

Buzu, Alpha.F A. 2007, Repere cronologice privind cultura cu ceramic liniar a sud de Carpai,

Mousaios, 12, 27-56.

F A. 2008, Epoca eneolitic n nordul Munteniei, Mousaios, 13, 7-39.K H., V L M., V P. (eds.) 2009, Archaeological Prediction and Risk

Management. Alternatives to Current Practice, Leiden, Leiden University Press.

K K.L. 1999, Recent directions and developments in Geographical Information Systems,Journal of Archaeological Research, 7, 2, 153-201.

M S. 2008, Trcov-Piatra cu lilieci, jud. Buzu, Cronica Cercetrilor Arheologice dinRomnia, Campania 2007, 308-309.

M S. 2009a, Dacii din Curbura extern a Carpailor, in V. C, V. S (eds.), Dacii dinCurbura Carpailor. Catalog de expoziie 2009, Sfntu Gheorghe, Angustia, 18-21.

M S. 2009b,Habitation structures and funerary discoveries in the area of Buzu County, in the8th-3rdcenturies BC, in V. S (ed.), The Necropolises and the Environment. Proceedings ofthe 11th International Colloquium of Funerary Archaeology (Buzu 2009), Buzu-Brila.

M-C I. 2000, Observations concerning the Bronze Age cemetery of PietroasaMica, Romanian Journal of Archaeology, 1 (http://apar.archaeology.ro/rja.htm).

M-C I. 2003, Cteva observaii asupra culturii Monteoru [Some regardsconcerning the Monteoru culture], Ea-Online (http://www.archaeology.ro/imc_mont.htm; http://www.archaeology.ro/imc_mont_eng.htm).

M-C I., S D. 2003, Neni-Znoaga, Cronica Cercetrilor Arheologice

din Romnia, Campania 2004.O A. 1976, Tezaurul de la Pietroasele, in M. B (ed.), Opere IV,Bucureti, Edi-

tura Academiei.

P S. 1999, Observaii referitoare la evoluia culturii Boian n nord-estul CmpieiRomne, Istros, 9, 13-46.

R C., B P. 2004, Archaeology: Theories, Methods and Practice, London (4th ed.),Thames & Hudson.

S V., M S., D V.I. 2005, Incinta dacic forticat Pietroasa Mic-Gruiu Drii(II), Buzu, Alpha.

S Z. 1975, Noi descoperiri din epoca roman din sud-estul Transilvaniei, SCIVA,26,3, 343-351.

D., D M. 2005, Abordri teoretice i cercetri interdisciplinare, in S,M, D 2005, 119-138.

G. 1948, Le camp romain de Drajna-De-Sus, Departement de Prahova, Dacia,11-12, 114-144.

T G., A R. 1992, Cercetrile arheologice din punctul Piatra cu Lilieci, satulTrcov, com. Prscov, jud. Buzu, Cercetri Arheologice, 9, 46-51.

L M., K H. (eds.) 2005, Predictive Modelling for Archaeological He-ritage Management: A Research Agenda, Nederlandse Archeologische Rapporten 29,Amersfoort, ROB.

V P. 2007, Case Studies in Archaeological Predictive Modelling, Leiden, LeidenUniversity Press.

Z E. 2000, Monteoru, in C. P (ed.), Enciclopedia Arheologiei i Istoriei Vechi aRomniei, Bucureti, Editura Enciclopedic, 102-104.


18/18

D. tefan, V. Srbu

356

ABSTRACT

Archaeological Predictive Models (APMs) represent an important evolution of spatialintegrated databases of archaeological records. Before the development and the analysis of apredictive model, numerous other steps are required in order to integrate the raw data sets intofunctional archaeological systems. Our aim is to assess the evolution of archaeological datasets into APMs and to reconsider the real value of such attempts for the Romanian HeritageProtection or for scientic purposes. We will consider, as well, certain aspects regarding thedeductive/inductive nature of the APMs. In our perspective, there are a few ways APMs couldbe improved: the use of more variables, as well as the understanding both of the analyticalnature of data sets and of the real nature of archaeological data sets.

statistical tools in landscape archaeology

Documents