Seeking tombs from space - Automatic detection and extraction of Omani burial monuments in satellite images.

Cover image: view of Wādī al-Jīzī in a WorldView-2 satellite image (Ṣuḥār, Oman) (false colour composite NIR1-Red-Green).

1

SEEKING

TOMBS

FROM

SPACE

Automatic detection and extraction of Omani

burial monuments in satellite images

Author: Giacomo Fontana Student number: s1747223

Course code: ARCH 1046WTY (RMA Thesis) Supervisor: Dr. B. Düring

Specialisation: Research MA Archaeology: Town and Country in the Mediterranean Region and the Near East

University of Leiden, Faculty of Archaeology Leiden, 12 June 2018, final version

3

TABLE OF CONTENTS

ACKNOWLEDGEMENTS 7

1 INTRODUCTION 9

1.1 THE ROLE OF THE BURIAL MONUMENTS IN THE ARCHAEOLOGY OF OMAN 10

1.2 PROBLEMS WITH THE CURRENT METHODOLOGIES 11

1.3 RESEARCH QUESTIONS 13

1.4 METHODOLOGY DESIGN 14

1.5 EXPECTED RESULTS 14

1.6 OUTLINE OF THE THESIS 15

2 THE LANDSCAPE OF ṢUḤĀR 17

2.1 LANDSCAPE ARCHAEOLOGY 17

2.1.1 FUNERARY LANDSCAPES 20

2.1.2 A PALIMPSEST OF DATA 20

2.2 LANDSCAPE TAPHONOMY OF ṢUḤĀR 21

2.2.1 A REGION OF PRESERVATION 23

2.2.2 THE RECENT DESTRUCTION 27

2.3 CONCLUSION 28

3 THE FUNERARY LANDSCAPES OF OMAN 31

3.1 THE HAFĪT PERIOD (3300B.C.E.–2500B.C.E.) 32

3.1.1 THE ḤAFĪT TOMBS 35

3.1.2 THE ḤAFĪT FUNERARY LANDSCAPE 38

3.2 THE UMM AN-NAR PERIOD (2500B.C.E.–2000B.C.E.) 43

3.2.1 THE UMM AN-NAR TOMBS 44

3.2.2 THE UMM AN-NAR FUNERARY LANDSCAPE 47

3.3 THE WĀDI SŪQ PERIOD (2000B.C.E.–1600B.C.E.) 49

3.3.1 THE WĀDI SŪQ TOMBS 50

4

3.5 THE EARLY IRON AGE (1300B.C.E.–600B.C.E.) 53

3.5.1 THE EARLY IRON AGE TOMBS 56

3.6 THE LATE IRON AGE (600B.C.E.–300C.E.) 57 3.7 THE SASANIAN PERIOD (300C.E.–630C.E.) 58

3.8 ISSUES OF THE CURRENT RESEARCH 61

3.9 THE ROLE OF REMOTE SENSING TO ADDRESS THE CURRENT ISSUES 63

3.10 CONCLUSION 65

4 REMOTE SENSING IN THE MIDDLE EAST 67

4.1 SATELLITE REMOTE SENSING 67

4.1.1 DIGITAL IMAGES 71

4.1.2 IMAGE INTERPRETATION IN REMOTE SENSING 71

4.2 THE EVOLUTION OF REMOTE SENSING FOR THE MIDDLE EAST 76

4.2.1 AERIAL IMAGES 76

4.2.2 THE INTRODUCTION OF SPACEBORNE PLATFORMS: THE CORONA PROJECT 77

4.2.3 A PROLIFERATION OF SATELLITES 81

4.2.4 REMOTE SENSING FOR THE MASSES 82

4.3 THE EMERGENCE OF AUTOMATION 84

4.3.1 AUTOMATION IN THE MIDDLE EAST 85

4.3.2 AUTOMATIONBEYONDTHEMIDDLEEAST 89

4.4 CONCLUSION 91

5 METHODOLOGY 93

5.1 SOFTWARE AND DATA 94

5.1.1 DATA DESCRIPTION 95

5.2 DATA PRE-PROCESSING 96

5.2.1 SPATIAL PRE-PROCESSING 98

5.2.2 RADIOMETRIC PRE-PROCESSING 100

5.3 DESCRIPTION OF THE AUTOMATION 100

5.3.1 FIRST STEP – OBJECTS DETECTION 103

5.3.2 SECOND STEP – OBJECT CREATION 110

5

5.3.2.2 Cleaning 112

5.3.2.3 Limit definition 114

5.3.3 THIRD STEP – OBJECT CLASSIFICATION 114

5.3.3.1 Threshold classification 121

5.3.3.2 Random tree classification 123

5.4 CONCLUSION 125

6 RESULTS AND DISCUSSION 127

6.1 APPLICATION OF THE METHOD 127

6.2 ASSESSMENT OF THE DETECTION PROCESS 128

6.2.1 RESULTS OF THE DETECTION PROCESS 129

6.2.1.1 Comparative assessment of the detection process 133

6.2.2 REPRESENTATIVENESS OF THE DETECTIONS 135

6.2.2.1 Qualitative representativeness 135

6.2.2.2 Quantitative representativeness 138

6.3 ASSESSMENT OF THE DRAWING PROCESS 138

6.4 POTENTIAL OF THE AUTOMATION 146

6.5 CONCLUSION 149

7 CONCLUSION 151

7.1 A NEW TECHNIQUE FOR STUDYING AND MONITORING TOMBS 151

7.2 A REVIEW OF THE METHODOLOGY 153

7.3 AIMS FOR THE FUTURE 154

ABSTRACT 155

BIBLIOGRAPHY 157

LIST OF FIGURES 167

LIST OF TABLES 177

6

7

A

CKNOWLEDGEMENTS

The research presented in this thesis is the result of several years of work in the Sultanate of Oman. During this period, I discovered the extraordinary potential of the study of Arabian tombs for developing an understanding of several millennia of human occupation. Simultaneously, I recognised the practical limits of the available methods for the study of the extensive funerary landscapes of the Arabian Peninsula. These considerations motivated my decision to develop a new approach to map and analyse the burials.

This would not have been possible without the support of two key organisations. The DigitalGlobe Foundation granted me exclusive access to WorldView-2 satellite images of the area of Ṣuḥār, while Trimble provided me with their cutting-edge software eCognition for remote sensing. For this reason, I thank them.

I also thank Dr. Bleda Düring for his supervision and, above all, for letting me discover, experiment and research so many different things during my Research Master whilst always supporting me.

At the same time, I thank my friends for the countless hours and moments of laughter spent under the sun of Ṣuḥār. Finally, I thank my family for the invaluable moral support that they provide every day.

9

1 I

NTRODUCTION

In the Sultanate of Oman thousands of burial monuments lie across the landscape. Despite some gaps in the cairn sequence, these tombs are among the most consistent evidence relating to the lengthy period between the third millennium B.C.E. and the seventh century C.E. The continuity of this type of burial practice makes these funerary monuments of exceptional importance for researching the ancient communities that inhabited these territories. However, the lack of an efficient method to detect and record the tombs has made it extremely difficult to develop systematic and comparative studies of these funerary landscapes. As a consequence, a robust understanding of the societies that created the burial monuments is still unavailable. Furthermore, ongoing developments, such as urban and agricultural expansion, as well as a drastic increase in looting activities, have led to the destruction of thousands of tombs. The current situation makes the study of the burial monuments an urgent issue for the archaeology and cultural heritage of the Sultanate of Oman.

This project aims to develop an efficient method of detecting and recording the burial monuments of Oman from remote. The current research uses two methods to detect the tombs. The first consists of archaeological surveys on the ground and, the second, of visual inspection of satellite imagery. These methods are not optimal for application across large territories. Fieldwork in Oman is problematic due to the challenging topography of the landscape and efforts to visually interpret satellite imagery are not efficient because they do not fully exploit the quality and quantity of available data.

This shortcoming in data interpretation is related to the ‘big data problem’. Every day, the entire surface of the Earth is monitored with hundreds of satellites imagery. The data sourced from these satellites contain the necessary information to detect a significant number of archaeological objects by inspecting the imagery. Moreover, thanks to the use of legacy data, it is possible to collect information regarding archaeological remains that have already disappeared. Although manageable for small regions, visually interpreting satellite imagery is not optimal for study of large areas of the Arabian Peninsula. Interpretation of satellite imagery is a very time-consuming task, but the most critical caveat for this method relates to the output of visual interpretation. The results of the interpretation are directly related to the images chosen for the analysis and it is difficult to repeat the process with different data. For the same reason, this method is not

10

efficient for heritage management purposes. The periodic monitoring of the remains would require the repetition of the entire process of interpretation for each application.

This thesis uses recent, cutting-edge approaches of remote sensing to develop an automation for the detection and recording of burial monuments from high-resolution multispectral satellite images. Only with this type of automation it is possible to systematically analyse large territorial areas and thus provide the datasets that are necessary both for archaeological research and for heritage management purposes. This thesis study is only the first step in a broader research ambition aimed at understanding and preserving the funerary landscapes of the Arabian Peninsula.

1.1 T

O

The burial monuments are among the most consistent evidence of human occupation from the Ḥafīt to the Sasanian period in Oman (3300 B.C.E.–630 C.E.). These tombs are small-scale stone structures that can occur both in isolation or in clusters (fig. 1.1). Other types of ancient architecture, such as remains of villages, have also been discovered but only for some phases of this very long timeframe. For this reason, the burials and the associated artefacts are the most ubiquitous evidence available to develop diachronic and comparative analysis of the dynamics of human occupation.

Recent projects have used the burials as proxies to analyse the human relationship with the landscape. A working assumption is that the spatial distribution of the burials reflects the human occupation of the territory. The patterns of burials have, therefore, been correlated to the presence of natural resources and anthropogenic factors in order to identify different models of human land use. These have been used to discriminate between sedentary or nomadic communities, to study the emergence of

11

agriculture and to trace continuity in occupation through different chronological periods (Al-Jahwari 2013, Deadman 2012, Giraud 2010).

However, these studies have often produced contradictory results (e.g., Al-Jahwari 2008; Giraud 2010), and have failed to identify coherent associations between burial patterns and the types of societies they relate to. The major problem consists of the absence of comprehensive studies of the burial evidence. Indeed, several scholars have focused on particular typologies of tombs while neglecting the study of others. Furthermore, newly discovered burial typologies do not clearly fit within existing classes, challenging the previously proposed classifications (Düring and Olijdam 2015). Moreover, the existing studies generally focus on small areas and, therefore, preclude broader comparative analysis. As a result, archaeologists still lack a robust understanding of the tombs and the people who created them.

1.2 P

The lack of comprehensive studies is directly related to the absence of an efficient method of mapping the burial monuments. Archaeological surveys in Oman can be very time consuming and expensive because the areas under investigation are extensive and difficult to access. The burials are often located on mountainous ridges and slopes, which follow old riverbeds for many kilometres (fig. 1.2). As a result of their locations, a survey of the burials precludes the use of cars and necessitates walking through a challenging landscape in high temperatures. These conditions are physically arduous for the surveyors and very slow. As a consequence, survey operations require substantial funding and are usually limited to small areas.

The introduction of aerial and satellite images has improved the situation, allowing the remote study of these burials and other archaeological objects (Leisz 2013). The majority of the typologies of tombs are clearly visible on images taken from above (fig. 1.3). Thanks to this characteristic, several projects have chosen to identify these structures by visually reviewing images taken from satellites. This operation is carried out by a surveyor who reviews small portions of the research area at a time and records the presence of archaeological sites in a geographical information system (GIS). This operation is usually followed by fieldwork aimed at investigating a representative selection of the objects detected in order to verify their interpretation as burials, examine their characteristics and date these structures through associated artefacts.

12

Although manageable for small regions, visually interpreting satellite imagery is not efficient across large areas. Firstly, the results of this approach are heavily based on the expertise of the interpreter. Moreover, they are hardly replicable by other scholars because of the difficulty in describing the applied routines. Secondly, images taken in different periods of the year show a different landscape where the visibility of the archaeological remains changes. For this reason, it would be necessary to analyse different images of a given area in order to increase the reliability of the detection. This repetition of the entire process of visual interpretation for each image results in a dramatic increase in the time required for the analysis. For this reason, the majority of previous research projects have limited their work to a single satellite scene for a given area. Thirdly, approaches based on visual interpretation cannot fully exploit the quantity of data which are available. Every day the entire surface of the Earth is recorded in hundreds of satellite images. These images cannot be processed manually due to the Figure 1.2 View of a series of burial monuments located along an old riverbed (Ṣuḥār, Oman).

13

enormous effort required. As a consequence, the potential use of these data, both for archaeological site detection and monitoring, are under exploited.

Over the last decade there has been growing interest in the scientific community in methods aimed at automatically detecting archaeological objects from remotely-sensed data (e.g., Scheutter et al. 2013; Schneider et al. 2015; Zingman et al. 2016). These methods attempt to replicate the capabilities of human vision in recognising and interpreting objects in images. Being computer-based, they allow the analysis of large territorial areas without the active effort of a human interpreter. A human operator’s role is, instead, limited to a review of the results. Furthermore, the quality of the analysis outcome is constant because it is not based on the expertise of the users. The automated methods are the best available approaches to the analysis of the significant quantity of data acquired every day by the satellites that cover not only Oman but the entire surface of the Earth (Bennet et al. 2014).

At present, efficient methods are not available for the automatic detection of the burial monuments of Oman. A first attempt to address this need was made by Schuetter

et al. (2013). Unfortunately, the technique that these authors designed was inefficient

due to the large number of misdetections produced. Therefore, this thesis aims to develop the first effective automation for detecting and extracting information on the burial monuments of Oman from satellite imagery.

1.3 R

Remote sensing does not currently allow full detection of the target objects using an automatic method. Therefore, to be effective, the method developed in this thesis had to be designed to detect a representative number of burial monuments that could be used to deduce reliable archaeological information. The outcome of the analysis need to be representative both of the spatial distribution of the burials and of their different morphological typologies.

The main research question of this thesis is as follows: Is it possible to develop an

effective method to automatically detect and draw the burial monuments of Oman from satellite imagery?

Furthermore, the following sub-questions are addressed:

1. Can a representative number of burial monuments be detected with automated detection?

14

of burials are not detected?

b. Are the detected objects representative of the spatial distribution of the burial monuments?

2. Is it possible to automatically record the morphological characteristics of the burial monuments?

a. Can the perimeter of the burial monuments be automatically identified and drawn? What degree of accuracy can be reached?

b. Is it possible to use these characteristics to date the burials?

1.4 M

To address the issues outlined above, this thesis developed an automation able to detect burial monuments in high-resolution multispectral satellite imagery, specifically WorldView-2 data (WV-2). The method consisted of a series of basic rulesets written in ‘eCognition’. Each of the rulesets aims to exploit a precise piece of information present in the imagery that was useful for the detection of the tombs. These pieces of information were, generally, basic observations on the morphology of the burials that were visible in satellite imagery. The method was composed of three parts. The first part aimed to identify a series of candidate objects as parts of possible burial monuments. The second part automatically identified and draws the burial perimeter. Finally, the last part classified the objects in burial monuments or other objects.

A pilot study for the method was carried out on an area of 3km2 _{and it was tested}

using an area of 100km2 _{in the region of Ṣuḥār, in the Al Batinah. This region is the focus}

of the Wādī al-Jīzī Archaeological Project (WAJAP), under the direction of Dr. Bleda S. Düring (Düring and Olijdam 2015). One of the aims of the WAJAP project is to develop understanding of the funerary landscape of this region. Some of the burial monuments had, therefore, already been recorded during earlier survey activities. The outcomes of these surveys were used to identify the burials upon which the rulesets were developed and to verify the results in the test area. The accuracy of the drawing method was evaluated on the basis of the analysis of the characteristics of the burials in the satellite imagery.

1.5 E

The method developed in this thesis is likely to have a high impact on the study of the funerary landscapes of Oman. For the first time, comprehensive datasets about burial

15

monuments across different areas of this territory may be created. The operation outlined in this thesis may allow the development of diachronic and comparative studies across large areas. This is likely to be of great value in the study of the patterns of the burials and to lead to a better understanding of the societies that the monuments relate to.

This thesis project also contributes to the development of methods of automation in remote sensing for archaeology in general. Indeed, this is one of the few applications of automation in satellite imagery interpretation for the Middle East. Furthermore, this research contributes to exploration of the 'big data' issue with regard to the use of the extraordinary quantity of data that is acquired every day by satellites around the world.

Finally, the method developed during this thesis produces new possibilities for monitoring and preserving the burials of the Arabian Peninsula. It is possible to apply the method of semi-automatic detection to satellite images taken at different times. This operation allows for the monitoring of the ongoing destruction of burials but also helps to detect and prevent new forms of destruction if applied regularly. The prevention of further destruction is now a critical issue for the burial monuments of the Arabian Peninsula (fig. 1.4).

1.6 O

The automatic method developed in this thesis is presented in the following chapters. After this introductory chapter, the field of landscape archaeology studies and the regional setting of Ṣuḥār are introduced in chapter 2. In chapter 3, the current state of research on the funerary landscapes of Oman is reviewed, together with its major issues. Figure 1.4 The picture shows one of the many sites destroyed in the region of Suhār (Oman). On the left the situation in 2009 (DigitalGlobe data - Google Earth), while on the right in 2018. The large cluster of burials in the centre of the picture has been entirely destroyed (WV-2 data).

16

In chapter 4, the field of remote sensing in archaeology is presented together with the history of the research. Chapter 5 describes the automation in detail and explains the choices made in the development of the method. Chapter 6, then, reviews and evaluates the results obtained. Finally, the last chapter answers the research questions posed above, and reviews the approach chosen for the creation of the automatic method.

17

2 THE

LANDSCAPE

OF

ṢUḤĀR

This chapter discusses the particular suitability of the region of Ṣuḥār for landscape archaeology studies. The archaeological remains of a large part of the region are visible and very well preserved. This situation facilitates their detection and, thus, the reconstruction of past landscapes. The region of Ṣuḥār is, therefore, an optimal study area for the application of remote sensing approaches in archaeology.

The first part of this chapter reviews a number of fundamental concepts in landscape archaeology. It discusses the concept of cultural landscape as the result of the diachronic interplay between environment and culture. Then, it explains how the effectiveness of the landscape study is contingent on the level of visibility and preservation of the archaeological remains. The second part of the chapter describes the physical environment of the area of Ṣuḥār. Most of this area is a preservation landscape, in which the remains of past activities continue to be visible on the surface. This type of region is thus particularly suitable for landscape studies. At the end of the chapter, recent landscape developments in the region are presented. Despite the naturally favourable conditions, the rich archaeological record of Ṣuḥār is currently under threat. The ongoing developments of urbanisation and the looting activities have destroyed numerous archaeological features and continue to constitute a significant danger to the region’s archaeological remains.

2.1 L

The notion of landscape is an exceptionally complex concept. The word itself has multiple meanings. The landscape can be the portion of land overseen from a particular point. It can also be the terrain within which people dwell or the word ‘landscape’ can indicate the topography of a territory. In this thesis, ‘landscape’ is defined as a synergic complex of physical and cultural characteristics that form a lived and experienced environment (Ingold 2000). A landscape is the result of the reworking of a ‘natural landscape’ according to the religious, economic, social and political strategies adopted by the different communities that have inhabited it. This reorganisation consists of a wide range of man-made modifications of the natural landscape, such as the construction of settlements, field systems, roads and sanctuaries (fig. 2.1). Therefore, the concept of landscape does not indicate the natural environmental alone, but also encompasses its artificial features

18 (Ashmore and Knapp 1999).

The human-made modifications of the landscape form the cultural systems that organise the interactions of people with the natural environment. The term cultural landscape thus refers to the historical presence of the human in the territory as an agent of transformation of the natural space, but also as an observer and interpreter of the iteration between nature and culture (Anschuetz et al. 2001, 160–1). Inhabitants perceive the landscape and manipulate it according to imbued meanings. The landscape is lived and experienced through a set of symbolic associations that overlay the physical space. These symbolic meanings are, therefore, not a quality of the landscape itself, but are conveyed by the presence of one or more individuals. The transformation of ‘space’ into ‘place’ is a human intervention (Ingold 1992). The human perception of the landscape can be seen as an overlay—a meaning given to the landscape—or as an inherent component of the landscape. In the first case, it is understood as a metaphysical space of the mind that builds a perspective in which culture is an ‘arbitrary symbolic framework built on the surface of reality’ (Ingold 1995, 66). In contrast, the alternative notion considers the perception of reality to be itself a component of that reality. The second approach rejects the idea that ‘our understanding of the world is somehow a failed attempt to come to terms with things as they really are’ (Thomas 2012, 175). Instead, our thinking is regarded as an activity that takes place in the reality of the world and thus that there are no boundaries between mind and world (Ingold 2007, 28).

One of the most important symbols embedded in the landscape is that of kinship. The landscape itself is a reminder of the relationship between the living and the past generations (Thomas 2012, 176–8). This connection with the past is manifested both in Figure 2.1 View of the Islamic site of Sahlat (Ṣuḥār, Oman). The figure shows the reorganization of the natural landscape with the construction of a settlement and fields systems (WV-2 data).

19

the physical space, with the remains of past human activities, and in the culture, with, for example, the names given to places. It is important to highlight how the meaning of these features depend on the individuals who are observing and interpreting them. The same location can appear as different places between various individuals. Therefore, the landscape can be understood in different ways according to different cultural inheritances (Bender 1998, 87). Despite the complexity of the meaning of landscape, the structure of each place reflects the cultural identity of its inhabitants because the traces of human activities materialise the membership of the inhabitants in a specific culture and territory (Olwing 1993; Terrenato 2000).

The different forms of the physical and cultural landscape are generated and sustained by the ongoing interplay of mutual relationships and transformations between nature and culture. The landscape is, therefore, a dynamic product of natural and human life. As a consequence of the characteristics outlined in the preceding paragraphs, the landscape itself can be the subject of a biographical approach. This approach implies the use of ‘biography’ to describe ‘lives’ of objects. The concept of biography of an object was originally proposed by Appadurai (1986) and Kopytoff (1986). As an object, the landscape is subject to interpretation and, as such, its meaning can change between individuals or communities. This point is important because it allows for the study of the landscape in its diachronic transformation as the result of the interplay between nature and culture, rather than relying on the history of its occupation.

The different manifestation of the landscape in particular times can be seen as layers that constitute a palimpsest of data. The landscape is, thus, a continuous diachronic sequence of information, both temporally and spatially. This characteristic is the major asset of the landscape compared to other types of archaeological sources. Due to this aspect, landscape archaeology allows the study of the territorial complex as a unique diachronic entity. Indeed, the landscape is intended not as a collection of discrete archaeological sites separated by empty spaces and isolated in time, but instead as a continuum product of the iteration between nature and culture. This idea enables the identification of elements of continuity and discontinuity through periods of time. It allows for the identification of constants that remain determinant of the topography and history of the place and, at the same time, variables correlated to specific phenomena in time and space. Overall, landscape archaeology is a form of comparative study that compares models, dynamics, trends and developments on a regional level and according to a diachronic approach (Breuilly 1992).

20

2.1.1 F

The landscape is characterised by a remarkable multiplicity. It is the set of the different natural components and artificial uses in a specific territory. These elements are present in different forms, which are contextual for each territory. As a consequence, the study of the landscape has to be both quantitative and qualitative (Muir 1999, 74). These different constitutive elements can be grouped under specific research themes. This thesis concerns funerary landscapes and, thus, those parts of the landscape that are related to death.

Death is not a separate entity in the landscape because it is perceived as a part of the landscape inhabited by the livings. Furthermore, it can be instrumentally used by the living to fulfil a series of necessities present in both ordinary life and in the ‘extraordinary’ sphere. Death constitutes a major component in the conceptualisation and manifestation of the supernatural in the landscape. As such, burials can mark sacred locations and, thus, identify areas of the landscape imbued with symbolic associations and meanings. On the other hand, they can also assume secular functions. The most common secular use of burials is as territorial markers to identify the ownership of parts of the land. Specifically, formal burials are often located in the spatial proximity of crucial but restricted resources. Lineal descent from the dead, thus, legitimated the right of the people to use those resources in a ‘systematic relationship between the economic base and the ideological superstructure’ (Pearson 2003, 137). This explanation is an example of a functionalist interpretation of death as presented in the landscape. This account of the burial complexes cannot represent the full layers of meanings related to this interpretation but, at the same time, it is a reality of its secular use in the landscape. Death has, therefore, multiple roles in the landscape and it can be associated with economic, social and religious purposes.

2.1.2 A

Landscape archaeology aims to trace the factors at the base of landscape formation, and to analyse the landscape’s diachronic evolution. To achieve these outcomes, it is necessary to isolate the specific landscapes that existed during the different phases of human occupation. This is a challenging process because current landscapes are palimpsests of all the different phases of human occupation that have occurred through time. Being the result of this progressive superimposition of earlier and later occupation, landscapes present elements of all of them, but, at the same time, they are not entirely

21

representative of each of those periods. Indeed, each later landscape modifies the features of earlier phases, changing, in this way, the preservation and visibility of the earlier landscapes (fig. 2.2).

The concept of visibility is crucial to detect the specific landscapes that have occurred through time. Visibility indicates the degree to which the features of a past landscape are visible and preserved in the current situation (fig. 2.3). Landscape toponomy studies both the natural and cultural processes that selectively remove or cover part of the landscape, and thus change the visibility of the features (Wilkinson 2003, 41). The primary natural factors are the processes of erosion and sedimentation. These processes obscure or destroy part of the landscape. Therefore, it is necessary to detect them to properly evaluate the absence of data in specific areas of the landscape as a real absence of past communities. The cultural processes in landscape evolution are related to several factors. The first of these factors is concerned with the type of past landscape that is investigated and the types of societies that created it. Nomadic communities, for example, leave only ephemeral traces of occupation that are extremely difficult to detect in the current landscape. In contrast, sedentary communities generally leave durable remains that require an active investment of resources to eradicate them from the territory. For this reason, they are often better preserved and visible in later landscapes. The second factor concerns the possible reuse of features of earlier landscapes. This phenomenon can either preserve or destroy the features. For example, the reuse of roads or field systems preserved them because they become an integral part of later landscapes. On the contrary, the reuse of construction materials of earlier features destroys their traces in the current landscape.

It is only possible to interpret a landscape after an analysis of landscape taphonomy. Natural and anthropogenic phenomena have changed each landscape. Each generation has used and changed the past landscape according to its own necessities and, therefore, left a different landscape to the next generation (Wilkinson 2003, 7). It is essential to detect the effect of these phenomena in the creation of the current landscape. The following paragraphs evaluate the impact of the natural and anthropogenic processes into the formation of the landscape of Ṣuḥār.

2.2 L

The natural landscape of Oman is characterised by three main types of environments: the desert, the mountains and the alluvial plains. The desert of the Rub al-Khali, ‘Empty

22

Quarter’, occupies the interior of the peninsula. To the east, it is divided from the alluvial plain that precedes the Arabian Gulf by the mountain range of the al-Hajar. These mountains form the topographic backbone of the northern Oman Peninsula. They run from the Musandam peninsula to the north to the western extremity of the Arabian territory to the south, toward the area of Ra

’

A narrow alluvial plain named Al Batinah, meaning the belly, is boundaried to the northeast by the mountain range of the al-Hajar (fig. 2.4). The plain has the shape of a crescent that is approximately 5km wide at the extremities and 20-25km wider in the centre. Seasonal water courses, or Wādi in Arabic, have cut deep valleys through the natural barrier of the al-Hajar mountains. These valleys constitute natural corridors used since antiquity as communication routes between the interior and the coastal areas of Figure 2.2 View of a palimpsest of settlements (Ṣuḥār, Oman). From left to right, remains of structures dated to the late Islamic period, to the Bronze Aga period, and a modern house (WV-2 data).

Figure 2.3 The figure shows different levels of visibility between a cemetery, on the left, and a settlement, on the right (Ṣuḥār, Oman) (WV-2 data).

23

Oman. The three main corridors are Wādī Summa

’

and Wādī Hatta to the North. The city of Ṣuḥār is situated in a strategic position behind the corridor of Wādī al-Jīzī. This locational choice allows the control of a crucial passage point that connects the coastal area to the interior with the oasis of Buraymī, which has been populated since the early third millennium B.C.E.

The region of Ṣuḥār can be roughly divided into two areas. The first consists of the mountainous area and the second of the coastal plain. At the foothill of the main al-Hajar complex, the minor mountain complex of the Jabal Sheikh delimits a corridor parallel to the coast and the mountains. It is accessible from Wādī al-Jīzī to the south, and from the valleys of Wādī Bani Umar al-Gharbi and Wādī Fizh to the north (fig. 2.5). The area of

‘

is to the south of this corridor and in proximity to the passage of Wādī al-Jīzī.

‘

numerous economically workable copper sources that have been exploited since antiquity.

The coastal plain of the region of Ṣuḥār is the result of Pleistocene and Holocene sediments washed from the al-Hajar mountains (Costa and Wilkinson 1987, 24). Today, the al-Hajar complex projects numerous ridges and mesa-like hills toward the coast. These are the remains of river terraces, which have been eroded and dissected by later Wādi flow. The presence of numerous terraces at different altitudes is a common feature in this landscape. These terraces have been exploited for settlement strategies since the Bronze Age. Together with the different ridges, they have been chosen as locations for the numerous burial monuments present in this region (fig. 2.6).

After the mountain ridges, the plain is a gentle sloping terrain 2–6km wide and mainly composed of sand and silt deposits. The catchment of Wādī al-Jīzī is very large, and it guarantees an abundance of water in the region. This area is today the primary agricultural zone of the Al Batinah. Here, a variety of fruits, cereals and dates are grown.

Toward the sea, after this area, the coastal sand belt provided sites for several settlements in the past. Although still inhabited today, the urban development of the region has moved a large part of the city toward the hinterland. Between the plain and the coastal sand belt, there are several salt flats (sabkha). These are flat areas constituted by silt and silty clay deposits characterised by salt encrusted surfaces that are the result of the evaporation of water drawn up from the sea (Costa and Wilkinson 1987, 26).

2.2.1 A

24

Figure 2.4 Terrain map of the Al Batinah region (Oman) with administrative divisions (Esri basemap).

Figure 2.5 Satellite view of the hinterland of Ṣuḥār (Oman) with the indication of the geographical elements discussed in the text (Esri basemap).

25

archaeological features. In particular, the hinterland is characterised by low levels of sedimentation, which preserves the location of the archaeological remains on the surface. Elements such as low walls made of only a few layers of stones or mud are still visible on the ground. Furthermore, they are often visible also from aerial and satellite images (fig. 2.7). In contrast, the coastal area has been the subject of more extensive processes of sedimentation as well as urban development. For these reasons, a large proportion of the coastal archaeological sites have been lost.

Despite the optimal conditions of preservation, the hinterland of Ṣuḥār is a patchwork of zones of preservation and zones of attrition. This is caused by the tendency of human activities to persist in the same areas of the territory. An example of this heterogeneous situation are the valleys and the surrounding mountains. The past settlements and field systems were usually built in the valley terraces. As such, it is common to find later settlements that have been built partially, or totally, over earlier villages. The reuse of the same areas is also evident with the field systems. They usually occur in terraces close to Wādi or natural springs. These fertile areas were reused in later periods because of their limited presence in the landscape (fig. 2.8). Furthermore, the introduction of the falaj irrigation system stimulated the reuse of existing fields due to the effort required to build these infrastructures. By contrast, large parts of the hills and mountains that surround the valleys have been less exploited. They have been used mainly for the construction of funerary monuments that spread across large areas of the territory. These funerary areas are very well preserved, and it is possible to detect a ‘fossil’ landscape (fig. 2.9).

The landscape of Ṣuḥār is, thus, characterised by high levels of visibility of the Figure 2.6 View of a series of burial monuments located along a mountain ridge in Wādī al-Jīzī (Ṣuḥār, Oman).

26

archaeological remains that are situated on the numerous old river terraces of the hinterland; locally variable visibility of the remains in the valleys; and poor preservation of the coastal sites. In general, the overall situation favours the study of the hinterland from satellite and aerial imagery. A large part of the archaeological features is visible on these images. This situation is particularly optimal for the study of the burial monuments. The majority of these archaeological objects are built on the hills and Wādi terraces in the hinterland of Ṣuḥār. Therefore, they are generally well preserved and visible, and thus easy to detect in the landscape.

Figure 2.7 View of a series of field systems around the site of Sahlat (Ṣuḥār, Oman). It is possible to detect the limits of the fields although a large part of them are low walls made of mud (WV-2 data).

Figure 2.8 View of a series of field systems along Wādī Fizh (Ṣuḥār, Oman). It is likely that the modern field systems on the right are built over the remains of the fields associate to the Iron Age settlement on the left (WV-2 data).

27

2.2.2 T

The main threat to the preservation of the archaeological landscape of Ṣuḥār derives from the recent urban and agricultural developments as well as looting activities. During the last 10 years, there has been growing urbanisation of several areas that were not extensively used in the past. Specifically, the creation of large industrial areas at the foothill of the mountains has led to the construction of a series of infrastructures, which have damaged the archaeological record. The works carried out for the construction of the Batinah Expressway have worsened the situation (fig. 2.10). Despite the efforts of the Omani government to conduct a series of rescue excavations along the planned route of the expressway, several companies involved in the construction of the road excavated large portions of the hills to gather construction materials. These activities did not take into consideration the preservation of the archaeological remains and, on the contrary, are the direct cause of the destruction of hundreds of them.

Mining activities in the area have further contributed to the destruction of parts of the landscape. In particular, the area of

‘

area was known for the presence of numerous, economically workable copper sources. Weisgerber has described this landscape as characterised by numerous archaeological remains correlated to metallurgic activities (Weisgerber 1987). Nowadays, the landscape that he described is no longer recognisable. Modern mining has completely changed the area and destroyed a large part of the remains of the past activities.

On a local scale, the construction or expansion of small farms or villages have also contributed to damage to the archaeological record. Local inhabitants have often reused Figure 2.9 A mountainous ridge of Ṣuḥār (Oman) (right) shows a ‘fossil landscape’ very different in comparison to the recent development of the neighbouring plain (left) (WV-2 data).

28

parts of archaeological sites without knowledge of the archaeological remains that were present in the surroundings. This process is clear from the dramatic increase of farming activities that has happened over the last 50 years in the environs of Ṣuḥār city, which have completely covered the archaeological palimpsest. At the same time, there is evidence that the archaeological remains of the hinterland have also been damaged. In this way, important sites such as Zahrah 1-R (Costa and Wilkinson 1987, 95) have been partially destroyed in the construction of contemporary gardens (fig. 2.11).

Another critical issue for the preservation of the archaeological record is looting. These activities have been concentrated on the burial monuments. At the present time, a large number of the tombs in the hinterland of Ṣuḥār have been looted. It is possible to detect these activities due to a characteristic hole at the top of the structures made by the robbers (fig. 2.12). It is difficult to date these lootings, but the presence of recent food packages in a significant number of the looted tombs indicate a peak of looting activities during recent times. Unfortunately, it is difficult or impossible for the local authorities to control the vast territory that these tombs cover. An effective method to prevent or monitor the ongoing destruction has to be developed.

2.3 C

Landscape archaeology can provide unique insights into the past dynamics of human occupation by analysing and comparing the diachronic modification of the landscape. The hinterland of Ṣuḥār represents a favourable situation for the study of landscape archaeology. A large proportion of the features of the past landscape is still visible. This visibility facilitates the detection and reconstruction of past landscapes and, thus, the study of the dynamics of past human occupation.

The burial monuments are among the best preserved archaeological remains in this area. They are among the few consistent types of structures used over an extensive period, between 3300 B.C.E. and 630 C.E. A comparison of the successive funerary landscapes can, therefore, provide excellent data for the understanding of the past communities. The next chapter introduces the funerary landscapes of Oman. It presents a review of the recent research literature and discusses the potential of remote sensing to address the current issues.

29

Figure 2.10 The figure shows the transformation of the landscape of Wādī al-Jīzī due to the construction of the Batinah Expressway (Ṣuḥār, Oman) (WV-2 data).

Figure 2.11 The site of Zahrah 1-R (Ṣuḥār, Oman) has been largely destroyed by the developments of a modern farm. On the left the situation at present (WV-2 data); on the right a map of the site with the indication of the different phases of developments identified in Google Earth (after Costa and Wilkinson 1987, 96).

30

31

3 THE

FUNERARY

LANDSCAPES

OF

OMAN

Burial monuments are of exceptional importance for the archaeology of Oman. Although there are a few gaps in the cairns sequence, they are among the most consistent evidence of four millennia of human occupation, from the Ḥafīt to the Sasanian period (3300 B.C.E.– 630 C.E.). Other types of monumental architecture are absent or sparse for some periods of this very long timeframe, and thus not available or difficult to use to develop diachronic and comparative studies. In contrast, the continuity of use of the burial monuments in funerary practices makes them effective proxies for the study of the long-term social dynamics of the communities that inhabited this territory.

During the period of the use of the cairns, a large proportion of the communities of Oman transitioned from an earlier pastoral-nomadic lifestyle to a social organisation based on farming villages. Due to a lack of evidence, it is not clear when and why this transition took place. The spatial distribution of tombs has been used as a proxy of the human occupation in the landscape to investigate this transition and its later developments. It has been related to natural resources and anthropogenic factors in an attempt to understand the different subsistence strategies adopted by these communities.

Unfortunately, the research has focused on certain periods and neglected the study of others. Several interpretations of specific funerary landscapes have been proposed, but none of them are satisfactory. This is due to the lack of comparative and diachronic studies on the funerary landscapes that developed across extensive areas of Oman. This gap in information is directly related to the inefficiency of fieldwork strategies in covering vast territories. Archaeological surveys in the Arabian Peninsula can be extremely time consuming and expensive due to the large and often inaccessible areas investigated. The tombs are, typically, clearly visible on satellite imagery, and for this reason, several previous research projects have chosen to identify these structures by visually reviewing images taken from satellites.

This chapter reviews the archaeological data available and how they have been used and interpreted by scholars. It describes the socio-economic situations present in the different periods with the aim of highlighting current problems that might be solved in the study of the funerary landscapes of Oman. Recent research literature is presented according to the different periods of construction and use of the burial monuments of Oman. These periods are Ḥafīt (3300 B.C.E.–2500 B.C.E.), Umm an-Nar (2500 B.C.E.–2000

32

B.C.E.), Wādi Sūq (2000 B.C.E.–1600 B.C.E.), Late Bronze Age (1600 B.C.E.–1300 B.C.E.), Early Iron Age (1300 B.C.E.–600 B.C.E.), Late Iron Age (600 B.C.E.–300 B.C.E.), Early Islamic (300 B.C.E.–300 C.E.), and Sasanian (300 C.E.–630 C.E.).

Toward the end of this chapter, some of the studies carried to date are critically examined. Criticisms of previous studies are mainly related to the, often, non-systematic classifications of the burials adopted by scholars, which are frequently based more on precedent bias than on the real data collected in the field. Subsequently, an explanation is given regarding how comparative and diachronic studies on the funerary landscapes can solve issues in the current archaeological findings. The role of remote sensing in creating the datasets necessary to develop these studies is discussed. The tomb features that are detectable from satellite are reviewed in order to demonstrate the potential to classify the burials through remote sensing, using a combination of morphological characteristics and spatial location.

3.1 T

H

(3300

B.C.E.–2500

B.C.E.)

Nomadic pastoral groups occupied the territory of Oman from the start of the Neolithic period, in the eighth millennium B.C.E. During the fourth millennium B.C.E., the end of the Holocene Moist Phase caused the environment to change. The climate deteriorated and, as a consequence, competition for access to resources increased (Magee 2014).

During the early third millennium B.C.E., the evidence provides indications of new aspects of the communities of Oman. The most important of these indications are the traces of possible sedentary communities and the creation of thousands of burial monuments across the landscape. This period is called Ḥafīt, and it corresponds to the Chalcolithic and early Bronze Age (3300 B.C.E.–2500 B.C.E.). The name derives from the mountain of Jebel Ḥafīt (UAE) where the typology of burials characteristic of this period was studied for the first time.

The identification of sedentarism is a critical issue for the understanding of the Ḥafīt period. Some scholars interpret the Ḥafīt communities as sedentary and, furthermore, as already characterised by the oasis economy of the later Umm an-Nar period (2500 B.C.E–2000 B.C.E.) (Cleuziou and Tosi 2007). An oasis economy is a type of agricultural system typical in the Arabian Peninsula. It is based on a series of irrigated fields of palm trees. Once at the mature stage, the shadow created by the palm trees create a micro-environment that allow types of cultivation, which are impossible in arid environments, such as several types of cereals. The oasis system is the only possible

33

method for allowing agricultural activities across most of the Arabian Peninsula (fig. 3.1). Although widely discussed by scholars such as Cleuziou and Tosi (2007), there is little evidence to support the theory of a general transition toward a sedentary agricultural regime during the Ḥafīt period. Only five villages have been discovered so far, and only one presents traces of agriculture dated to this period. These are Bat, Hili-8, HD-6, Al-Khashbah and ALA-2 (Cleuziou 2009; Schmidt and Döpper 2017; Thornton et al. 2016). The principal evidence comes from the site of Hili-8 in the al-Ain oasis. This site resembles the tower-like structure of the late Umm an-Nar period, with a series of mudbrick buildings clustered around a central well (fig. 3.2). The remains of different types of cereals and fruits were discovered at this site, as well as remains of herded animals such as sheep, goats and cattle (Magee 2014, 94).

Based on radiocarbon dating, Hili-8 has been dated to 3300 B.C.E. (Cleuziou and Tosi 2007, 148). It constitutes the first attestation of a settlement with substantial agricultural activities. Several ditches were discovered around the main building. The depth of some of these ditches led Cleuziou and Tosi (2007) to interpret them as an underground water transportation system and a precursor of the falaj system of the later periods. They have argued that the irrigation channels used for the distribution of water in nearby fields always ran higher than the surface to be cultivated. On the contrary, these trenches are deeper than the surface level and, therefore, were probably used to transport water. Other scholars, such as Magee (2014, 95), have been more cautious in their interpretation. Magee has argued that these trenches were only distribution channels used to directly irrigate neighbouring agricultural fields from the central well at Hili-8.

Although not related to agriculture, there is other possible evidence of villages during the Ḥafīt period. In the Ja'alan region, the most southwestern part of Oman, at the intersection between the Gulf of Oman and the Arabian Sea, the site of HD-6 presents a series of mudbrick buildings surrounded by a stone wall with associated fishing materials and copper objects (fig. 3.3). Rather than agriculture, the economy of this village was based on the exploitation of the sea.

Further evidence is from Bat, in the region of Al-Dhahira in northwestern Oman. Several Umm an-Nar towers have been excavated at this site, and some of them led to the discovery of a similar mudbrick construction, usually associated with a well, dated to the Ḥafīt Period.

34

At the same time, other structures have been recently discovered at Al-Khashbah, in the Shamal Al-Sharqiya governorate. The excavation of the site led to the discovery of a series of mudbrick buildings that are encompassed by large ditch systems (Schmidt and Döpper 2017).

Figure 3.1 View of an oasis garden in Wādī Fizh (Ṣuḥār, Oman). The shadow created by the palm trees allows other types of cultivations.

Figure 3.2 Plan of Hili-8 (al-Ain, UAE) with the ditches interpreted by Cleuziou and Tosi as an underground water transportation system (Cleuziou 1997, 5).

35

The few known Ḥafīt villages provide little indications of a shift toward sedentarism. If compared with the remarkably large quantity of burial monuments spread across the landscape, we should expect a substantially larger quantity of village sites discovered. It, therefore, seems likely that the villages discovered only represent local developments. Furthermore, there is even less evidence to argue for a general trend toward the oasis economy at the beginning of the third millennium B.C.E., as proposed by Cleuziou (2009).

Overall, the evidence for the type of social organisation and economy of the Ḥafīt communities remains inconclusive. During the last decade, several scholars have tried to solve this issue using the Ḥafīt tombs as proxies to identify the location preferences of settlements and their strategies for the control and exploitation of natural resources. In the following sections, the characteristics of the Ḥafīt burials are described and these recent studies are reviewed.

3.1.1 T

The Ḥafīt-type tombs are completely without precedent in the archaeological sequence. They are circular stone structures built above ground that range in diameter between 5m to 11m. They have a conical shape that encloses a single burial chamber that usually hosts multiple inhumations. They are often located on rocky ridges and they are extremely visible in the surrounding landscape (fig. 3.4).

36

The Ḥafīt-type burials have been dated, thanks to the recovery of sherds of imported Jemdet Nasr jars from Mesopotamia. Due to the lack of excavations, a large number of the burials of Oman have been dated to the Ḥafīt period, based only on their generic resemblance to the few Ḥafīt-type cairns that were originally excavated. This identification process has created a bias in the research, with a tendency among some scholars to consider the majority of the burials built above ground on rocky ridges as Ḥafīt, even when there is evidence to suggest different dates. As a result, it is probable that many burials have been mistaken as Ḥafīt and, therefore, erroneously included on the study of the Ḥafīt funerary landscape.

The Ḥafīt tombs that were originally identified mainly occur in two types: cairns and beehive tombs. The first type has a small, round, corbelled chamber surrounded and covered by a cairn of stones (fig. 3.5), while the second has a characteristic beehive-type structure that can be made of dressed stones (fig. 3.6). The beehive type, also called Bat-type or transitional tombs, is considered a later evolution of the Ḥafīt tombs and it is interpreted by some scholars as a transitional type of burial that occurred between the Ḥafīt and Umm an-Nar period (Frifelt 1975; Boehme 2011; Williams and Gregoricka 2013). In both types, the tombs range between 5m to 11m in diameter.

The Ḥafīt tombs have a single burial chamber built on ground level, which usually varies in diameter between 1.2 to 2 metres and can be paved with flat slabs (fig. 3.7). The chamber is roofed with corbelling stones in the beehive type, and the top of the monument is flat. They commonly have a narrow trapezoidal door, which usually opens to the east and was blocked by unshaped stones. Both the typologies are made from stones that are, typically, gathered locally. The appearances of the Ḥafīt tombs vary significantly, according to the available materials of the different locations.

The Ḥafīt tombs usually contain multiple inhumations. Older skeletons were pushed to the sides to make room for new corpses. Cleuziou and Tosi (2007, 114) have argued that they were used by family groups, each using one or several graves. This interpretation has been questioned by recent evidence from sites such as Dhank, in central Oman, where the excavation of Ḥafīt tombs has led to the discovery of single depositions (Williams and Gregoricka 2013).

The discovery in Ḥafīt tombs of imported Mesopotamian jars and reproductions made locally provides the first evidence of the use of ceramics in the Arabian Peninsula. At the same time, these burials also contain small quantities of copper objects. These have been interpreted as imports from Iran rather than local products due to the association

37

of other remains of this provenience, such as ivory beads. The absence of any clear differentiation in the dimensions or the architecture of the Ḥafīt tombs or the objects found in them, together with the collective character of the burials, have been the basis for their interpretation as the product of an egalitarian society (Cleuziou and Tosi 2007, 122). Recent evidence from the Al Batinah region may question this interpretation. In the Figure 3.4 Ḥafīt burials in the al-Ain oasis (UAE) (UNESCO archives).

Figure 3.5 Ḥafīt burial of the ‘cairn of stones’ type (Ṣuḥār, Oman).

38

inland of Rustaq, a pattern constituted by a central burial of extremely large proportion, surrounded by smaller burials, has been detected across three different areas. This pattern has been interpreted as possible evidence of hierarchy among the burials (Deadman et al. 2015). Ultimately, more research is needed to prove this interpretation.

3.1.2 T

A large proportion of the burials of Oman is believed to be Ḥafīt. These tombs occur, isolated or grouped, in highly visible positions on rocky ridges overlooking roads, oases and Wādis (fig. 3.8). There is, so far, no satisfactory explanation for the extremely widespread diffusion of tombs during the Ḥafīt period. Several scholars have interpreted the emergence of Ḥafīt tombs as an attempt to define land ownership of new exploitable territories. Environmental studies have indicated that a climate change around 3200 B.C.E. produced an increasingly moist environment, which allowed a new exploitation of the inland area of Oman after the climate deterioration of the fourth millennium B.C.E. (Parker et al. 2007). The Ḥafīt tombs have, then, been interpreted as markers functioning to affirm the position and ownership of newly accessible natural resources for pastoral and agricultural activities.

This observation is at the basis of the recent uses of the burials as proxies to trace human occupation and its social organisation during the Ḥafīt period. The positions of the burials in the landscape has been related to the presence of natural resources and anthropogenic features. This approach has led scholars to suggest two opposing models of social organisation. On the one hand, scholars such as Cleuziou and Giraud have argued that the Ḥafīt communities were based on a sedentary agricultural economy (Giraud and Cleuziou 2009; Giraud 2010). On the other hand, other researchers such as Al-Jahwari and Deadman have argued for a pastoral-nomadic model of social organisation (Al-Jahwari 2008; Deadman 2012; Al-Jahwari 2013; Deadman and Al-Jahwari 2016).

Giraud has investigated the eastern Ja'alan region, applying a series of spatial analyses to a set of 3,000 burials dated according to their morphological resemblance to the Ḥafīt-type. She has not specified if there was any other evidence for their association to the Ḥafīt period (Giraud 2010, 72). According to the results of a buffer zone analysis, she has argued that the Ḥafīt burials of this region were clustered in 54 necropolises (fig. 3.9). The buffer analysis uses a series of buffers centred on each burial to identify common regions. The range of the radius is defined by the user and therefore subjective. She has identified common areas, using a radius of 500m and 1000m, and interpreted them as the

39

regions of different necropolises. She has, then, calculated the centre of each necropolis and noted that in 37% of the cases they relate to other small structures, as well as areas where, today, it is possible to develop an oasis system. On the basis of this slim evidence, Giraud and Cleuziou have theorised that the necropolises reflected the ownership of the land by different groups. Moreover, they have asserted that the relation between arable farmland and the presence of other structures at the centre of the necropolises is evidence for the interpretation of these areas as early centres of occupation based on the oasis system (Giraud and Cleuziou 2009, 172). Their interpretation can be critiqued. The Figure 3.7 Plan of an Ḥafīt burial (Shir, Oman) (after Yule and Weisgerber 1998, 193).

40

correlation of these features does not imply the presence of a village. The small structures identified have not been dated and possibly date to later periods. The possibility that an oasis system could be developed does not necessarily imply its creation, especially during the Ḥafīt period, in which the evidence of agriculture is limited.

Further evidence of sedentarism has been proposed by Giraud using the concept of ‘landscape of identity’ (Giraud 2010, 83). Performing viewshed analysis on the site of Figure 3.9 Map of the Ḥafīt necropolis identified by Giraud (Ja'alan, Oman) (Giraud and Cleuziou 2009, 171).

41

HD-6, she has traced the presence of a possible dwelling area at the visual centre of a local group of burials. She has noted that the known village is situated at the visual centre rather than at the centre of gravity. She has argued that this pattern indicates a phenomenon of spatial appropriation that is typically developed by sedentary communities to identify themselves with the surrounding landscape. This single instance cannot constitute evidence of sedentarism. Furthermore, it contrasts with the general pattern that she has identified in other areas, because the site of HD-6 is not at the centre of gravity of the associated necropolis. Overall, Giraud’s arguments are not convincing.

Al-Jahwari has carried out further investigation in the Ḥafīt funerary landscape in the Wādī Adam region (Al-Jahwari 2008) and the western Ja'alan region (Al-Jahwari 2013). In the Wādī Adam region, he has mapped 2,000 tombs from ground-based surveys, while in the second region, he has mapped 5,000 burials. He has dated them to the Ḥafīt period, based on their generic resemblance to the Ḥafīt burial type, although the vast majority of the artefacts that he has found are dated to the Iron Age period and, as he has reported, a great typological variability was detected in the field.

Both surveys conducted by Al-Jahwari have failed to find any settlements that can be dated to the Ḥafīt period. He has proposed that the Ḥafīt society consisted of nomadic pastoralists living in temporary and ephemeral camps (Al-Jahwari 2013, 163). These communities were using several areas as grazing land for their animals according to seasonal availability. Al-Jahwari has suggested that the limited resources available had to be secured and protected from potential rival pastoral groups. Therefore, he has proposed that the tombs had the function of marking the land ownership in the periods where the communities were exploiting different areas of the territory (Al-Jahwari 2013, 167).

Like Al-Jahwari, Deadman has also interpreted the Ḥafīt communities as nomadic pastoralist. He has studied the region of Wādī Andam remotely, using satellite imagery. He has detected the presence of around 3,000 cairns that he has dated to the Ḥafīt period on the basis of their generic resemblance to the Ḥafīt-type (fig. 3.10). He has, then, performed a series of spatial analyses to highlight the relationship of the tombs with natural resources. Comparing the distance of the burials from the arable land and the Wādi, he has noted that the Ḥafīt tombs tend to be located along Wādi rather than in proximity to arable land. Furthermore, he has indicated how the density of the burials seems proportional to the sizes of the watercourses (Deadman 2012).

42

Deadman has repeated the analysis using the data from the western Ja’alan collected by Al-Jahwari (Deadman and Al-Jahwari 2016). The results have confirmed the correlation between Ḥafīt cairns and watercourses, with a proportional increase of burials according to the size of the watercourses. Furthermore, he has noted how there are no indications of locational preference for burials in proximity to arable farmland. He has argued that this constitutes further evidence that the Ḥafīt community were nomadic pastoralist rather than settled agro-pastoralists.

The relationship of burials and water resources has been questioned in a recent study of the Al Batinah region (Kennet et al. 2016). In the inland of Rustaq, the Ḥafīt cairns are located perpendicular to the Wādi channels instead of parallel to them. This anomaly has been tentatively explained as the results of the presence of a vertical aquiclude parallel to the tombs. If confirmed, the distribution of the tombs could follow these water alignments instead of the Wādi.

Further evidence arguing for the absence of a relationship between water and the Ḥafīt cairns has come from an earlier survey of the Dhahirah region. Gentelle and Frifelt (1989) have indicated that the distribution of Ḥafīt graves was not tied to the availability of water and cultivable soil. Instead, they have noted how the location of the burials mark mining areas and trade routes.

Overall, the interpretation of the Ḥafīt funerary landscape is contested. The different interpretations of the scholars have not been supported by the data recorded in different regions. This may indicate regional differences in organisation between the Ḥafīt Figure 3.10 Graphical view of the criteria used by Deadman to classify the burials (Deadman 2012, 28).

43

communities but, more likely, it is the result of the problematic dating method applied and the lack of systematic and comparative studies across different areas of Oman.

3.2 T

U

-N

(2500

B.C.E.–2000

B.C.E.)

The term Umm an-Nar refers to a timespan of 500 years between 2500 B.C.E. and 2000 B.C.E., the period of the Bronze Age. It is named after the island of Umm an-Nar, near Abu Dhabi, where remains of this period were discovered for the first time.

During this period, there was a remarkable intensification of settlements. The causes of this phenomenon are not clear. Cleuziou and Tosi (2007, 90–91) have argued that the diffusion of settlements was the consequence of the consolidation of the oasis system, which was initially developed during the Ḥafīt period. In contrast, other scholars have placed great emphasis on the role of neighbouring Mesopotamia. During this period, there is evidence of an increased extraction and export of copper from Oman to Mesopotamia, and probably also to India (Weisgerber 2007). This trade network has been interpreted as the driver of the intensification of settlements.

The settlements found are characterised by a round tower, usually made of mudbricks, which can vary between 20m to 40m in diameter. In some cases, the tower was at the centre of a larger settlement populated with structures made by organic materials, of probably residential use. At Bat, evidence attests that the area between these structures was irrigated from the nearby Wādi (Thornton et al. 2016). In addition to these large sites, other smaller settlements have also been recently recorded. These are normally constituted of a series of stone structures, and are probably to be related to small groups of inhabitants.

Agricultural activities are attested during this period and the cultivation of cereals, such as barley and wheat, seems to have had an important role in the diet of the Umm an-Nar communities. Nevertheless, the emergence of cereal agriculture does not have to be a catalyst for sedentarism. From region to region, a wide range of terrestrial and marine resources was exploited in parallel to agricultural products. Therefore, the subsistence economy was based on local availability of resources rather than cereal agriculture (Magee 2014, 102).

Trade with other regions during the Umm an-Nar period are well attested. Several sites have provided evidence of imported ceramics from Iran and India. At the same time, a flourishing trade of copper and copper made artefacts is reported in Mesopotamian textual sources, and indicated in the evidence of mining activities in the centre of Oman