The EURODEER database is the repositories where data coming from the different partners are standardized, stored and shared to be analyzed. The first goal of the project is to investigate variation in roe deer behavioural ecology along environmental gradients or population responses to specific conditions, such as habitat changes, impact of human activities, different hunting regime. Therefore, the main set of data stored into the EURODEER database are those generated by biotelemetry sensors to monitor animals movement, particularly:
- GPS data
- VHF data
- Activity data (accellerometer)
While these data sets provide a huge mine of data to be mined, it is when they are couple with additional information like that of the individuals, populations, management regimes and the environment that they can help to fully decipher the mechanisms of animals movement. This is why EURODEER is investing a lot in getting these information from each partner, standardize them and integrate into the analysis. This is a complex process but ultimately it is the biggest added value of the collaboration among scientist in addition to the integration in a single repository of the data coming from sensors. The work of data standardization requires long and iterative discussions but it is also an opportunity to identify good practice that can then be shared with a wider scientific community
The quality of the data is of primary importance for the project because inconsistent or erroneous information can lead to misleading scientific interpretations of the animals movement. In additions to the controls and quality checks performed by each research group, all data are validated before they are included into the database. The experience of the project clearly show how, due to the nature (size and complexity) of the data sets, very rarely data provided by each group are without errors. These can be due to a number of possible reasons, for example to wrong information recorded in the field (e.g. start/end of a sensor deployment), errors in data acquisition/transmission by the sensors, mistakes in data handling when it is not properly automated/supervised.
Database data modeling
Data modeling is an important issue when data structure becomes large and complex, as in the case of animal movement tracking, where a large set of sensors, environmental variables and ancillary data sets (e.g. individuals, captures, population, managment regimes, study areas) can potentially interact and be used by many research groups in a collaborative framework. A data model (or database model) is the link between the elements of the real-world (e.g. moving animals, sensors, the environment) and the physical representation of those elements in a database (in EURODEER case, a relational database). It "translates" the researcher's conceptual model into a scheme inside the information system. A special effort towards a conceptual data model capable of accurately describing the real world was needed to develop the EURODEER database. It took into account the complexity of the real world problems to be solved, the nature of the application of the data, the technical environment, the policies governing the use of the information and the expected performance standard of the application. This modeling approach ensure in the long term that the software platform built around GPS data can efficiently manage data. Developing a database without a proper data model is like building a house without an architectural plan.
In the figure on the left, we present a scheme of EURODEER data model. It gives a picture of just the main elements of the database. It is limited to GPS data, but the same structure is used for VHF and activity data. The two main objects are the GPS devices and the animals monitored. The device provides coordinates of the location at a specified time interval, identifying the position of the animal carrying it. A device can be associated to an animal for a defined time range. On the base of the timestamp of the fix and of the table that associates individuals to GPS devices, a spatial table can be filled with the code of the individual, the point of the location (as a spatial object, not just as a pair of numbers), the timestamp and the code of the GPS device. Once transformed into a spatial feature (i.e., a point), the locations were intersected with other spatial layers enriching the record with all the relevant environmental and socio-economic information (i.e., elevation, slope, aspect, habitat class, temperature, rainfall, administrative boundary, distance from the closest road, distance to the previous location, angle of the movement, season, night/day, vegetation index). Two further objects are Studies and Research groups. Data can come from a specific project managed by one or more research groups, and the same research group can manage many projects. This scheme is specially important to keep the database open for future complex collaborations between EURODEER partners.
Many other objects are present in the database and linked to those mentioned, like home ranges or convex hulls (polygons) of specific selections of locations, trajectories or segment of trajectories (lines), that were also intersected with other spatial layers in the database. More information on individuals, GPS devices and any other aspect of each wildlife project (including the use of other sensors) were added to further extend the core data model.