Geodata analysis at regional level integrates inevitably some datasets from various sources (statistical, geographical, environmental,...), various scale (regional, national, ..) and various quality: While political structures are constantly changing, as in a potentially conflicting region such as Caucasus, these data integration issues increase. Implementation of quality control methods is an initial and essential step in the integration of geodata inside a spatial regional model. This report provides tools for data harmonization that can be applied to other datasets and other region when existing data sources do not evaluate the quality of their information. The goal of this report is to provide a quality assessment of the Caucasian GIS dataset to build the Caucasus geomodel of instability/stability. This report evaluates qualitatively and quantitatively the adequacy of this dataset to the objective in following a structured quality assessment protocol (Johnston et al. 1999) and consolidates a final geodatabase. Integrating data from a multitude of derivative geospatial products produced by different sources pose severe problems. Challenges are also introduced by the GIS technology itself. Various data are introduced in this study but the main source of statistical and spatial information is the acquisition of the geopolitical atlas dataset, the "Caucasian dataset" (Radvanyi, INALCO, 2006). In this report, four data quality elements are identified and described in the specific case of the Caucasian dataset. Lineage information, the three accuracy dimensions (positional, temporal and attribute), logical consistency and completeness evaluations are qualitatively and quantitatively assessed by various metrics. This paper illustrates the use of automatic cartographic and data cleanup techniques of Geographic Information System (GIS) to solve data issues (self overlapping, dangles, pseudonodes and gap in spatial data). This report can further be used as a reference for both the producer and the user to somewhat replace the missing metadata information. Clear statements on dataset quality allow to better communicate in a common goal of understanding the geopolitical Caucasus context. The bulk of this report has aimed to illustrate how spatial data from various sources have been collected and made ready for use within a GIS. The different evaluation tests allow to give an overall estimation of the dataset quality. This type of data cannot be used at a scale higher than approximately 1:500 000. This Caucasian dataset has the objective to provide an overall picture of the regional security complex and not a precise localisation of specific real features. This fact has to be kept in mind in the following processing modelling stages. Based on the results of this report, especially the completeness and fitness of the dataset to represent the scope of the model, the Caucasus study will further explore two distinct modelling approaches: (i) a spatial and continuous muticriteria model of instability integrating in a continuous GIS the geopolitical factors, (ii) defining instability indicators values for subnational spatial entities (district units) throughout the Caucasus region. This report provides an adapted methodology to assess quantitatively the quality of a database when no metadata information is available. The elements of data quality are envisaged in a progressive way in this report and thoroughly studied for the settlement layer. The other layers are evaluated in a less in-depth way but allow the test of different methods associated to the three types of features (point, line, polygon).
