In the Contexts section of the research programme original datasets are integrated and analyzed to create new indicators of long term trends in landscape and territorial development. Landscape and socio-economic territories are the result of many interactions and processes. To understand them requires, on one hand, systematic observations of land use and practices over time and, on the other hand, an anthropological approach focused on interactions between territories and identifying landscape development strategies. The socio-economic indicators produced, of course, vary according to cultural and geographical context, but overall can contribute to a better overview of the co-evolution of landscapes and territories.
Long term trends in land use and occupation patterns
One can study both historical and contemporary land use. Contemporary land use characterization is usually based on processing remotely sensed images- classifying, with more or less detail and accuracy, the entire area under study. Land use distinguishes urban areas from forest or cultivated areas, for example. Similarly, archaeological remains such as manuring traces, field system features, settlement, and burial activities can be used in the production of period by period land use maps (Bertoncello and Nuninger 2006).
Focusing on differences between periods, ModeLTER will provide quantitative indicators of change, described in terms of intensity and rhythms. Each part of space studied can be characterized by its steady or rapid development and, similarly, by the pace of its decline (Tourneux 2000). In addition, some qualitative indicators will be produced to characterise the types of change in order to evaluate their impact on landscape and on territorial structures.
A second approach will focus on the environment of settlements, using descriptors of composition and configuration in the style of landscape ecology. Rather than land cover, physical topography described in terms of homogeneity, variety, diversity, fragmentation, etc. will be the main dataset in these analyses. These variables are well described in the literature, but the question of the size of the area which comprises the environment of a location, which is directly linked to scale-effects, remains. Continuing on the basis of research developed in within our own team (e.g. for bird inhabitation or epidemiological modelling), we will identify local minimums in variability through radial analysis to identify variables and scales where inhabited locales show the greatest difference with other (uninhabited) places (Foltête et al. 2002 ; Wharton 1982). The first result of this type of analysis should be information on the determining factors for appropriate spatial scales of analysis, the second expected result is evidence for the development of environment preferences over time. These results will be used in the third research programme as a basis for predictive modelling.
Through the first two approaches, ModeLTER will be able to provide an accurate description of changes regarding land use and settlement pattern for multiple areas and periods within a long term framework. Even if one can observe changes and infer some strategy, the question remains: why did people make particular choices? What problems and responses have they developed according their own contexts (cultural, geographic, etc.)? GIS tools offer an opportunity to include “cognitive” criteria in modelling, by using visibility or pathway algorithms for example (Zakšek et al. 2007). Including these factors may illuminate part of the decision making process. However, these tools are usually formatted for present-day economic applications, and the cognitive criteria are not well integrated. Archaeologists, ethnologists and geodesists , working collaboratively, are better able to formulate, to process and to test inputs to the algorithms and to validate the results. This type of analysis may lead to a better understanding of human perceptions and actions, particularly for past societies, in the production of territories and landscapes. The main goal is to clearly and accurately define assumptions and models of strategies in terms of indicators which can be used in predictive models.
The team focuses on using the anthropogenic environment to characterize models of territorial and landscape development over the long term.
First, it is important to understand the long term development of different kinds of settlement patterns and types of territories. Relationships between settlements can be considered in terms of distances, accessibility or visibility from one settlement to the others. These inter-site variables are mainly derived from settlement location and DEM analysis (produced in the first step, see 1.3 Terrain modelling). A conventional approach based on classical gravity models takes in account these inter-sites variables to build hierarchical networks of settlement (Durand-Dastes et al. 1998; Nuninger 2004; Nuninger and Sanders coord., 2006). This model, however, does not consider temporal factors. The temporal dimension of relationships will be described as working though a 'heritage process' from period to period. Information on connectivity between sites can be summarized in contiguity matrices encapsulating both temporal and spatial links. Then each settlement, described by this matrix, can be treated as a statistical record belonging to general matrix featuring all the connections. This general matrix can be analyzed using multivariate analysis algorithms to produce synthetic typology of site connections, in space on the one hand, and in time on the other. The root of this idea is the use of settlement networks as a basic analytical unit for they description and study of territories.
The second level of analysis aims to study landscape production, in this model treated as secondary, complementary information, enhancing the picture created based on the settlement network to describe territories. The French part of the group has extensively studied techniques and processes of land scape development from prehistory to the present. These studies are based on records of entities such as field systems, shapes of parcels or parcel boundaries. The aim is to understand the functioning of processes of land development around the network of inhabitation, to understand how communities built their environment to define their own space (Di Méo 1998 , Klopatek and Gardner 1999). To achieve this the landscape will be systematically described, using analyses developed in geography, and processed using multivariate analysis (Tourneux 2000, Tolle 2005). The challenge here will be to adapt the geographic analyses and descriptors to historical and archaeological criteria at data, as has been done elsewhere for settlement patterns, to create a broad overview of long term landscape production ,based on the same methodology.
Finally, based on the results of the detection part of our research program (satellite images and aerial photography), we aim to identify out anomalies, such as unusually fast changes or otherwise extraordinary behavior in the models of territorial and landscape development (Nuninger 2004). Combining historical, ethnographic and geographical perspectives on what constitutes anomalies should give a better understanding of the trajectories followed by the communities of each study area. Importantly, to allow comparison across diverse study areas (Nuninger 2009), the interpretation or explanation of phenomena will be based on consistent and carefully defined reference criteria, clearly elaborated at each level of analysis, and uniform across all the study areas. The definition of anomalies and any possible explanations for their appearance can be used to introduce uncertainty into the predictive modelling processes.