TOWARDS DEVELOPMENT OF A TOOL FOR THE AUTOMATED ASSESSMENT OF THE SPATIAL ACCURACY OF NATURE OBSERVATION DATASETS

Croatia is one of the most species-rich countries in Europe, thanks largely to its geographical position, where several biogeographical regions overlap, each with its own specific climate, geomorphology, and ecology (Radović et al., 2006). With more than 37,000 taxa within the major taxonomic units, the research on distribution, ecology and conservation, population genetics etc. generates a large amount of spatial data distributed across a large number of taxon-specific databases which are constantly increasing.
Majority of nature obervation databases are a by-product of research projects funded by the government or are set up by non-governmental organisations with little funding. Their data are mostly publicly available, with some exceptions, such as data on endemic and endangered taxa. They often use the help of citizen science initiatives to expand their knowledge of target taxa (Virić Gašparić et al., 2022). Against this background, most nature observation databases, especially the smaller ones, do not have quality control for new records, or only partially apply quality control methods, depending on budget constraints. Input errors related to logical consistency, completeness or duplicate entries are usually mitigated in the initial steps of database architecture development in the form of mandatory fields, input formats and cross-checks with existing records (Dalcin et al., 2012).