Résumé
The Montpellier research observatory of the environment (OSU OREME) is one of the 27 french observatory of universe sciences. It manages and supports national and local observation services in various scientific fields (hydrogeology, geophysics, ecology, marine biodiversity, astro-physics, etc).One of its missions is to provide access to observation data by applying the FAIR principles (data have to be Findable, Accessible, Interoperable and Reusable).To demonstrate how we have built the information system to reach these goals, we propose to expose our technical support to observation data of the french KARST observatory network. The KARST Observation Service is part of the OZCAR and eLTER research infrastructures. It aims to strengthen knowledge sharing and to promote interdisciplinary research on karst systems in France. It relies on a network of 9 observatories and research teams. The collected data comefrom more than 600 stations, with 150 different measured parameters in meteorology, hydrogeology, chemistry or pedology. Data are collected either manually or by automated sensors.In 2016, we took over the management of the database. We designed a PostgreSQL / PostGIS database to store time series with detailed descriptions: sensors, acquisition methods, parameters, data levels, stations, etc. We used existing repositories to standardize metadata and help for their understanding and reusability, such as environmental thesaurus (GEMET, Theia-OZCAR, GCMD) or french geographical repositories (such as BD Carthage for french rivers).We have developed web interfaces (PHP / JavaScript) to enable data producers to easily describe their devices and time series. Based on this description, sheet templates are automatically generated to manually upload their observation data.Discharge and water level data from the french hydrological bank that are of interest for observatories are automatically queried and inserted into the database several times a day. Data can be searched and accessed thanks to 4 main web interfaces on the OSU OREME data portal (https://data.oreme.org/observation/snokarst) : (i) the standardized OSU OREME catalogue (ISO 19139, GeoNetwork) that is automatically generated from the database, (ii) an interactive map querying standardized map flows (WMS / WFS, GeoServer) , (iii) a multicriteria search interface with data download in CSV, (iv) and an interactive tool to search time series by keywords or geographical criteria and to plot and compare them. Datasets are associated with Digital Object Identifiers that provide citations and improve their accessibility.Finally, more than 600 timeseries are publicly accessible (CC-BY licence), representing more than 14 millions of measurements. All public data and metadata are automatically processed and sent to the french Theia-OZCAR information system.We now plan to implement standardized tools to distribute data, such as Sensor Observation Service or SensorThings standards.Most of the developed tools are shared to manage data of other OSU OREME observation services, such as the Larzac observatory (eLTER site, part of the french network of hydrogeological research sites, SNO H+). We are thus able to query, plot and compare ecological or geophysical observations with the same interactive tool, and therefore help for the discovery of environmental interactions.