Chrystina HÄUBER und Franz X. SCHÜTZ
Abstract
The objective of this paper is to demonstrate the development and application of a new method for the documentation of persistent structures, using GIS technology. These structures can be found by analysing different digital maps. For the level of correlation of spatial structures we have coined the technical term PX = Index of persistence ("Persistenzindex", SCHÜTZ 1999). Our test area is the Mons Oppius in Rome. For places with a complex history, such as Rome, the search for persistent structures in the current layout of the city (`Stadtgrundriss´) is one of the main objectives of topographical studies. Therefore, one aim of our interdisciplinary research while developing the archaeological information system FORTVNA (HÄUBER & SCHÜTZ 1999, 2001a,b) was the development of a corresponding method (SCHÜTZ 1999). The system is named after the Roman goddess Fortuna who had a sanctuary in our test area. C. Häuber is a classical archaeologist, F. X. Schütz is a geographer and a programmer. The PX method is applicable to vector data, it is formulated as an algorithm and can be implemented in every programming language. In this contribution we describe the method as a function of the information system FORTVNA. This function was programmed with C++ using ESRI Shapefiles. In SCHÜTZ (1999) the function was programmed with Visual Basic using data of the GIS SICAD/SD in C60 format.
The PX method allows quantitative analyses of maps and cadastres. It is especially useful for everyone in the scholarly community interested in historical urban studies, and, as far as Rome is concerned, for the many disciplines interested in the (ancient) City. This method allows spatial analyses of structures belonging to previous time periods, be they still extant or only documented on maps. It, therefore, supports the detailed visualization of urban development. The term `archaeological information system´ was first used by other disciplines (BILL 1996, pp. 325-326) and has not yet been defined by archaeologists. We ourselves define it as follows: Archaeological information systems are information systems applied to archaeology. Information systems consist of a database and a database management system (DBMS), hardware, software and data. DBMS provide functions for storage, retrieval and analysis of data.
1 Introduction
So far most geographical publications on persistent structures have been `paper based´ (BOBEK & LICHTENBERGER 1966, AYMANS 1988, SCHWARZ 1989). This is also true for topographical studies undertaken by scholars who are interested in ancient Rome (inter alia classical archaeologists and ancient historians) (STEINBY 1993-2000). Contrary to geographers, German speaking scholars of these fields do not use the technical terms `persistence´ (`Persistenz´) and/ or `topology´, although they may even deal more frequently with the matters in question than geographers. In fact, many ancient literary sources, as well as reports concerning archaeological finds since the Middle Ages until today contain topological descriptions and data related to persistent structures. These topological descriptions may refer to two or more topographical features which were “connected, lay adjacent to, behind/above each other”, etc. Although archaeologists and ancient historians offer correct interpretations of these descriptions, they ignore the fact that geographers and researchers of other disciplines have developed special methods for the analysis of topological matters, such as GIS. In the U.S.A. the application of GIS is more common than elsewhere (GOODCHILD 1996); there the wide range of topics for which GIS technology has already been used contains also the study of persistence. In German speaking countries GIS is rarely applied to the various scholarly activities related to ancient sites, especially so when these are located in modern urban areas (where the use of maps of the scale 1:500 and finer becomes essential). Our research related to the development of the information system FORTVNA has resulted in new findings concerning the persistence of topographical structures in our test area in Rome and were obtained by the application of GIS technology. These structures belong to different time periods and are therefore interesting for scholars specializing in various fields. For this contribution we decided to concentrate on ancient buildings, of which we show reconstructions in plan. The degree of precision obtained in our maps is very high. This was possible, because we found persistent structures in the current cadastre which proved to be remains of ancient buildings, since they are characterized as such on old maps and/ or etchings. Maps and cadastres show persistent structures as lines or areas. Our method is based on the analysis of lines, because areas are also defined by lines. In this contribution we describe the application of our method only briefly, a full account will follow elsewhere. Our partner in the `FORTVNA project´, the classical archaeologist Prof. Eugenio La Rocca of the Sovraintendenza BB.CC. of the Comune di Roma (the archaeological agency of the Municipality of Rome, Italy), kindly provided us with the maps and the cadastre shown here and the aerophotogrammetric survey used for our research, which were generated for his `Centro di Documentazione Forma Romae´ (`documentation centre for the map of Rome´). We also thank our partner in the FORTVNA project, Prof. Michael F. Goodchild (professor of geography, University of California, Santa Barbara, CA, U.S.A.) and the classical archaeologist Prof. Clementina Panella (professor of `metodologia e tecniche della ricerca archeologica´, Università `La Sapienza´, Rome, Italy) for discussing this text with us. We also thank the anonymous reviewers of this contribution for their comments.
2 Definitions and Methods
The main focus of this contribution is the concept of persistence, which `describes the effects caused by historical structures until the present day´ (WIRTH 1979, p. 279). A detailed discussion of the term persistence and of its establishment in geography is provided by WIRTH (1979, pp. 91-100). The index of persistence PX is a measure for the degree of correspondence between spatial patterns, for example line objects, in relation to a given research topic. In our example the PX is the ratio of the length of line objects in two maps, which were drawn at different times and which show the same structure in different stages of preservation. The index of persistence is possibly best grouped under the category of `density´ (`Dichte´), which, according to LICHTENBERGER (1998, p. 96), `has been so far almost neglected by studies in urban geography´. The study of density should be of interest for geography though, since it is closely related to space. The index of persistence consists of two components, one spatial and one timerelated. The first part of the index adds the ratios of lengths of those structures which were found by spatial analysis. It is calculated by using the coordinates of the `line patterns´, for example line objects or edges, and those of the lines found. The second part, the `time component´ incorporates the different time layers which are investigated. Both result in the following equation for the index of persistence (PX):
PX = f(L1 .. Ln); (T1 - T2) (f = function; L = Line; T = Time/date)
In theory the index of persistence is not only useful for the analysis of maps, which have been drawn since only a relatively short period of time, but could just as well be used by geologists, who compare mappable natural formations, who may be millions of years old.
f(L1 .. Ln) can assume values between 0 and 1. This function indicates the degree of `identity´ of two spatial structures in percent (0-100%). The parameters are specified in relation to a given research topic (see infra, 3.2) depending on the source material and data and on the question to be addressed. (T1 - T2) stands for two dates indicating a timespan for which one wishes to test, whether there are persistent structures. T2 is always the `earlier´ date (e. g. 1748). The difference T1 - T2 (1995-1748) indicates, therefore, the length of time (in this case: 247 = years) for which the persistence has been demonstrated. This number is always positive. For dates before BC the numbers should be negative. The choice to incorporate time in the formula for the index of persistence makes sense, because studies in persistence inevitably turn out to compare at least two different moments in time. A PX of 1;247 means that the studied spatial structures are identical (persistent) throughout the timespan which the compared maps represent. In case the compared spatial structures represent parcels, this PX would indicate that the boundaries of all parcels remained unchanged between 1748 and 1995. The PX of 0,5;247 means that only 50 % of the studied structures remained identical throughout the studied timespan. The reason could be that between 1748 and 1995 all parcels were divided in two equal parts.
Spatial structures, for example lines, can be called persistent in case their course in two maps dating in different times is equal or else very similar. Equal/ similar means in this case that two lines are in full length parallel or approximately parallel, or parts of these lines are parallel. Mathematically, the parallel course of lines can be demonstrated by the comparison of their gradients (see figure 1). The core of our method is based on this simple fact. The direction of a line is defined by its gradient. The gradient between point A and B is defined as equation:
mAB = ( YB-YA ) / ( XB-XA )
This gradient is equal to the tangens of the angle : mAB = tan ().
Figure 1: The gradient of a straight line (after: SCHEID 1994, p. 575).
For the problems discussed here it is sufficient to concentrate on the first quadrant in a coordinate system. All coordinates of the maps we used here are in `Gauss-Boaga´ (the national geodetic system, Comune di Roma 1996, p. 46). The PX method is applicable only to orthogonal coordinate systems; for others the coordinates can be transformed. This means that the rules formulated here are generally valid.
Similar line objects visible on different maps and cadastres can represent persistent structures only in case they are located very closely to each other and have at the same time approximately the same gradients. For identical lines visible on different maps the parameters for the distance between these and for the deviations of the gradients should be 0. Because the analysed source material may contain error and uncertainty (GOTTSEGEN et. al. 1999), the parameters for the distances and for the gradients of the lines to be compared have to be adjustable. These parameters must be estimated according to the source material used (see for this problem, infra 3.2). The PX method can be formulated as algorithm as follows:
1. Selection of the source objects and of the objects to be compared with.
2. Selection of the test area (for example selection of objects inside a box drawn on the map).
3. Selection of the parameters distance and deviation of gradient depending of the source material and of the uncertainty of data.
4. Start of search for objects with the defined distance in the set of objects to be compared with the first source objects.
5. Marking of the objects found and comparison of the gradients of these objects with the given deviation of gradient in the set of objects to be compared with (calculated by using the tan (), figure 1 ).
6. If the gradients of the objects found fall within the given limits write the ID of the source-object and the ID of the compared object in a new file.
7. Return to 4 until all source objects are compared.
8. Calculation of the PX as value of the length of lines in the defined test area (for example for building blocks).
For (XB-XA) = 0 a special function for exception handling is needed (SCHÜTZ 1999, p. 39). The major requirements concerning the data and the GIS applied for this method are:
line objects are defined by two points (`Spaghetti-Daten´). All lines have equal length. This is very important for the comparison of the corresponding gradients. What the GIS is concerned, it should have facilities to access the vector data via a script or a programming language. The algorithm can be implemented in this language.
3 The application of the PX method in our test area in the city of Rome
Figure 2: Digital cadastre (1995) of our test area in Rome and map of Rome by G.B. Nolli (1748), details. Publication of Nolli and the current cadastre in figs. 2-4 with kind permission of the Sovraintendenza BB.CC. of the Comune di Roma.
Figure 2 shows our test area in Rome. It is bounded by modern roads, in the south by the Via Labicana and in the east by the Via Merulana. Notice the huge oval shaped parcel in the south-west corner which reflects the groundplan of the extant ancient building located there, the Colosseum. The area between the two streets comprises one of the foot-hills of the Esquiline, which in antiquity stood ca. 30 m (today 10 m) above the bottom of the valley. From the 4th century BC until the 5th century AD some important sacred and public ancient buildings stood on top of this hill (see detail in figure 4), which have been documented since the 17th century until the beginning of the 20th. In the last decade, different interpretations und reconstructions of these architectures have been suggested (BAUER 1997, DE VOS 1993 & 1997, ENSOLI 1997, HÄUBER 1990 & 1998, HÄUBER & SCHÜTZ 1999, 2001a,b). Some of the interpretations of the ancient structures vary according to the authors assumed course of the so-called Servian city wall of the 4th century BC in this area (CASSATELLA 1995, PALOMBI 1999). Throughout antiquity this wall remained important as a boundary between different city wards (Regiones III, V).
The method described in 2.2 was implemented in the archaeological information system FORTVNA as a function called "PERSIST". The function was written in C++, the format of the data provided by our Roman partner Prof. La Rocca was ESRI Shapefile. Because all the vector data had to be split in line objects with equal lengths and two points, we wrote a programme in C++, which allows the splitting of line objects, and in particular of polylines. The reliability of the results obtainable with this method depends on the correct choice of the parameters for distance and for the deviation of gradient. Some statistical tests are necessary in order to find these parameters.
According to GOODCHILD et al. (1995, p. 356) „in some cases, where the line in reality is itself a polyline, and the truth is defined as straight mathematical lines between points, as is often the case for surveyed boundaries, each estimation of the line can be treated as an assemblage of point estimates“. In our test case all line objects in Nolli´s map and in the current cadastre have been surveyed very precisely. See fundamentally on `Nolli and GIS´ BRIENZA (1998a,b). The current cadastre is based on an aerophotogrammetric survey (COMUNE DI ROMA 1996, p. 624). In order to calculate the differences between Nolli´s map and the current cadastre, we compared first Nolli´s map with the aerophotogrammetric survey, since both show ancient structures. We found 14 ancient structures on both and used edges of these structures as passpoints. The difference in the location of the same two points in the two map sources were used in order to calculate the mean and the standard deviation for the 14 pairs of passpoints. Using the real deviations of the passpoints, we calculated as mean parameters: distance = 2.34 m and deviation of gradient = 6°. With these results and applying the method described above (supra, no. 2.2) we calculated the zones shown in figure 3. In figures 2-4 we used the current cadastre instead of the aerophotogrammetric survey because the latter is too detailed for small reproductions.
The PX method is applicable under the circumstances described here: line objects with two points, comparison of lines with equal length. Both of which are necessary. In case these data should not be available it is advisable to apply more sophisticated methods described by GOODCHILD et al. (1995). These methods comprise statistics such as the Perkal epsilon band.
3.3 Results and Interpretation
Figure 3: The black lines in the current cadastre indicate `zones´ containing persistent ancient structures that are already visible on G.B. Nolli’s map of Rome (1748).
Figure 4: Our test area in Rome, reconstruction of ancient buildings in plan. The reconstruction is based on G.B. Nolli’s map of Rome (1748), excavation reports of the 19th and 20th centuries, the current cadastre and persistent ancient structures which were found with the PX method (see for data sources figure 2).
Figure 3 shows the results obtained in a comparison between the vectorized version of G.B. Nolli's map of Rome (1748) and the digital current cadastre, using the function PERSIST. The gray lines represent the current cadastre. The broad black zones emphasize those sections of lines in the current cadastre, which are identical with lines that are already visible on Nolli's map, which means that the areas represented by the black zones contain persistent structures. Our choice of highlighting the persistent structures this way not only facilitated the visual control of the results obtained with PERSIST, but also the following analysis of those structures.
4 Discussion of the results
Using archaeological evidence, C. Häuber came to the conclusion that a number of those persistent structures found by using PERSIST represent remains of ancient buildings. On G.B. Nolli's map of 1748 these lines formed part of extant ancient structures which he characterized as such. Other maps and etchings of this period show some of these ancient ruins in elevation (FRUTAZ 1962). Today most of the ancient persistent structures visible on figure 3 do not extend above street level any more. As the area under discussion was almost completely built over at the turn of the last century, it is currently impossible to verify, whether or not the ancient persistent structures shown in figure 3 represent still existing ancient structures, which might be buried deeply in the ground. But there can be no doubt that these ancient structures determined the positioning of several parcels in the current cadastre, very much like the parcel mentioned above which `registers´ the current groundplan of the Colosseum (figure 2).
The results visible on figure 3 and first attempts to reconstruct the former ancient buildings in this area in plan, incorporating these results (inter alia figure 4), were presented and discussed at various workshops (HÄUBER & SCHÜTZ 2001 a,b).
The index of persistence in the example visible in figure 3 is PX = 0,036; 247. The calculation of the PX takes into consideration the 64344 analysed `linemeters´ in the current cadastre and the found 2300 `linemeters´ in Nolli’s map. Testing only this small area, it is, of course, impossible to know what the index of persistence for the current cadastre and Nolli’s map for the complete historical centre of the city of Rome might be.
The fact that PERSIST helps to find line objects in the current cadastre which Nolli’s map and other sources describe as ancient ruins should not lead to the assumption, that these persistent structures, for which we can also calculate the PX, represent all the ancient ruins which existed at the time the analysed maps were produced. It is not even easy to define the `existence´ of those structures, because also those below street level (`Bodendenkmäler´) may leave their mark in the layout of the city and thus in the digital cadastre. It is, therefore, important to remember that it is so far impossible to establish the `correct´ number of ancient structures (nor their sizes in plan) at any given time for which old maps and cadastres are available.
The reliability of the obtainable results can be improved by using as precise passpoints as possible. It is important to test the precision of the passpoints, since an estimate of their precision determines the estimated parameters for distance and gradient deviation. The obtainable results can be improved by a high number of (precise) passpoints.
What their quality is concerned, maps produced with `paper based´ methods (see introduction) are not inferior to digital maps created by using GIS technology. The decision for one or the other method is mostly dictated by other criteria. At the beginning, developing a method using GIS technology may be even more time consuming than applying `paper based´ methods, especially when it should be necessary to produce ones own digital data or when they have to be found and purchased, which may be not only time consuming but also expensive. It is, therefore, only worth while to develop a method using GIS technology in case one wants to use it more than once, and with different data and parameters. The greatest advantage of a digital method over a `paper based´ one is the fact that it can be repeated at will and at a very low price. This means also that digital maps can be evaluated more easily by others.
In case one decides to use digital data produced by others, which have not been properly documented, it may turn out to be very difficult to define their quality. It is essential to test these data and to document the results. Because the definition of the quality of digital data is so important, the data used here were tested by Prof. La Rocca´s collaborators in the project `Nuova Forma Urbis Romae´ (LA ROCCA 2001). The directors of this project have invited interested scholars to use their data, which means that in the future different scholars studying the history and/ or the topography of (ancient) Rome will use the `same´ digital maps and cadastres. This means nothing less than that the results of those studies will be easily comparable for the first time.
4.3 Concerning the application in archaeology
We assume that the PX method will prove to be especially useful for the study of large areas, mapped in small scale. It is obviously impossible that one single person could visualize all the data relevant for the documentation of the so-called `pressure of change´ (“Veränderungsdruck”) (BRÄUNING 2000, p. 26) for a place like the modern city of Rome (size: 1200 km2). The Sovraintendenza BB.CC. and the Soprintendenza Archeologica (the national archaeological agency) undertake important excavations in the historical centre of Rome, in which large areas are uncovered. Applying the PX method could become one of the methods which archaeologists use for planning such excavations.
The example chosen here represents a third possible application, the detailed, `non-destructive´ archaeological study using large scale maps. For the area shown in figures 3 and 4 (size: ca. 450 x 320 m), the result of the application of the PX method was quite surprising, because more ancient structures were found than with `paper based´ methods, applied earlier (compare HÄUBER 1990). The earlier results were due to false assumptions, as the analysis of the result of the test with PERSIST (figure 3) clearly demonstrated. What is left of the ancient structures in the current cadastre, are not line objects which represent complete walls or even parts of groundplans which are recognizable as ancient, but instead only small sections of these. But of these parts there are many more than expected. The result of the PX method applied to this example (figure 3) allows a much more detailed reconstruction of the former ancient buildings in plan than was possible before, because these persistent lines can be supplemented by data collected and documented during excavations of the late 19th and early 20th centuries (figure 4).
The persistent lines visible in figure 3 represent structures dating in different time periods. It is, therefore, important that the archaeologists who apply the PX method know the area which they are studying very well.
5 References
Aymans, G. (1988): Historische Karten aus der Sicht eines Geographen, in: Auswertung und Erschliessung historischer Karten. Landschaftsverband Rheinland, Archivberatungsstelle Rheinland, Archivheft 18, Köln, pp. 203-221.
Bauer, F.A. (1997): Einige weniger bekannte Platzanlagen im spätantiken Rom, in: Pratum Romanum. Richard Krautheimer zum 100. Geburtstag. Wiesbaden, pp. 37-41.
Bill, R. (1996): Grundlagen der Geo-Informationssysteme. Band 2, Analysen, Anwendungen und neue Entwicklungen. Heidelberg.
Bobek, H., Lichtenberger, E. (1966): Wien. Bauliche Gestalt und Entwicklung seit der Mitte des 19. Jahrhunderts. Graz-Köln.
Brienza, E. (1998a): La planimetria di G.B. Nolli e la topografia di Roma. Unpublished Tesi di laurea, Università di Roma `La Sapienza´.
- (1998b): Cartografia storica e cartografia numerica. La pianta del Nolli e il GIS . In: Bevilacqua, M.: Roma nel secolo dei Lumi. Napoli, pp. 189-202.
Bräuning, A. (2000): Bausteine Archäologischer Stadtkataster. Archäologische Informationen aus Baden-Württemberg 42, bearbeitet von A. Bräuning et al.
Cassatella, A. (1995): Domus Aurea, in: LTUR IV. Roma, fig. 18.
Comune di Roma (1996): Atlante di Roma. La forma del centro storico in scala 1:1000 nel fotopiano e nella carta numerica. Venezia. 4.edizione.
De Vos, M. (1993): Il tempio di Iside in Via Labicana a Roma. In: A. Mastrocinque (ed.), I Grandi Santuari della Grecia e L'Occidente, Trento, 81-91.
- (1997): Dionysos, Hylas e Isis sui monti di Roma. Roma, pp. 99-154.
Frutaz, A.P. (1962): Le piante di Roma. I-III. Roma.
Ensoli, S. (1997): various articles in: Iside. Il mito il mistero la magia. Milano. pp. 306-321.
Goodchild, M.F. (1996): Geographic Information Systems and Spatial Analysis in the Social Sciences. In: Aldenderfer, M., Maschner, H.D.G. (eds.): Anthropology, Space, and Geographic Information Systems. Oxford University Press. pp. 241-250.
Goodchild, M.F., Cova T.J., Ehlschlaeger, C.R. (1995): Mean Objects: Extending the Concept of Central Tendency to Complex Spatial Objects in GIS. In: Proceedings GIS/LIS ´95, ASPRS/ACSM, Nashville. pp. 354-364. URL <http://everest.geo.hunter.cuny.edu/~chuck/MeanLines/> (2/18/2001).
Gottsegen, J., Montello, D., Goodchild, M. (1999): A Comprehensive Model of Uncertainty in Spatial Data. In: K. Lowell and A. Jaton, Spatial accuracy assessment: land information uncertainty in natural resources. Ann Arbor Press. pp. 175-181.
Häuber, R.C. (1990): Zur Topographie der Horti Maecenatis und der Horti Lamiani auf dem Esquilin in Rom. In: Kölner Jahrbuch für Vor- und Frühgeschichte 23. Köln. pp. 11-107.
- (1998): "Art as a weapon". Von Scipio Africanus maior bis Lucullus, domus, horti und Heiligtümer auf dem Esquilin. In: M. Cima - E. La Rocca (eds.), Horti Romani, 6. Suppl. BullCom., Roma. pp. 83-112.
Häuber, C. & Schütz, F.X. (1999): The multi-disciplinary multimedia Geographical Information System applied to archaeology GIS[A] FORTVNA: the basics of development. In: Proceedings of the XVth International Congress of Classical Archaeology, Allard Pierson Series 12. Amsterdam, pp. 194-196, figs. 70-71.
- (2001 a), FORTVNA: A Research Tool: The Archaeological Information System for Ancient Rome: I. The test area Mons Oppius; II. The information system, in: American Philological Association (APA), 132nd Annual Meeting, Abstracts, San Diego, CA, January 3-6, 2001, pp. 197-198; cf. John Bodel, Writing on the Oppian, op. cit., p. 199.
- (2001 b), Joint AIA/APA Workshop, FORTVNA: A Research Tool: The Archaeological Information System for Ancient Rome, AJA 105 (2001), in press.
La Rocca, E. (2001): The project „Nuova Forma Urbis Romae“, in: APA, 132nd Annual Meeting, Abstracts, San Diego, CA, January 3-6, 2001, p. 200.
Lichtenberger, E. (1998): Stadtgeographie. Begriffe, Konzepte, Modelle, Prozesse. Band 1. 3. überarbeitete und erweiterte Auflage. Stuttgart, Leipzig.
Palombi, D. (1999): Regiones Quattuordecim (topografia), in: Lexicon Topographicum Urbis Romae IV. Roma, fig. 84 fuori testo.
Scheid, H. (1994) [ed.]: Duden Rechnen und Mathematik. 5. Auflage. Mannheim.
Schütz, F.X. (1999): GIS-gestützte Analyse von persistenten Strukturen in der Kölner Innenstadt auf der Basis der Kataster von 1836/37, 1938/39, 48 und dem aktuellen Kataster. Unpublished Diplomarbeit, Universität Bonn, Geographische Institute.
Schwarz, G. (1989): Allgemeine Siedlungsgeographie. 4. Auflage, Teil 2. Die Städte. Berlin, New York. (= Lehrbuch der Allgemeinen Geographie Band. 6, Teil 2).
Steinby, E.M. (1993-2000) [ed.]: Lexicon Topographicum Urbis Romae I-VI, Roma.
Wirth, E. (1979): Theoretische Geographie. Stuttgart.