Database of bat acoustic activity at onshore wind turbine nacelles

# Presentation of the data This database is the result of data collection and preparation over 5 projects conducted at CESCO (UMR7204 - MNHN) and funded between 2020 and 2025 by France Renouvelables, ADEME, the FRB and the CNRS. The aim of these projects was to train algorithms to predict the collision risk between onshore wind turbines and bats, using data from continuous acoustic monitoring at wind turbine nacelles carried out as part of French ICPE regulatory monitoring. The database compiles the results of the identification of recording files, which have all been processed in a homogeneous manner using the TADARIDA software developed by CESCO (UMR7204 - MNHN). These data consist of a sum of 5-second acoustic detections of bats for each time slot on each date for each site. There are four versions of the database, corresponding to different sizes of time slot in which the sum of detections was made: the whole night, per 10, 5 or 2.5% of the night elapsed. For each of these bat detection sums, the operation of the wind turbine, the weather conditions and other contextual information are documented, making it possible to model the factors influencing bat activity around wind turbines. This database contains a total of more than 130 sites across metropolitan France, totalling more than 29,000 nights of monitoring over 7 years between 2017 and 2023. It is being used to train algorithms to predict bat activity at specific times in the future, which will enable more effective management of wind turbine operation so as to minimise their ecological impact. # Files - P2_5_ER90_Final.csv: dataset with a temporal resolution by time slot of 2.5% of the night elapsed- P5_ER90_Final.csv: dataset with a temporal resolution by time slot of 5% of the night elapsed- P10_ER90_Final.csv: dataset with a temporal resolution by time slot of 10% of the night elapsed- P100_ER90_Final.csv: dataset with a temporal resolution by complete nights.- Definition_variables: presentation of the definition, class, unit, format, missing value presence and meaning, for each column of datasets.- data_2_5_t1.csv: the same than P2_5_ER90_Final.csv but with lines of temporal covariates moved one line before in order to model bat activity at t time in relation to weather conditions and wind turbine operation at t-1 time. In this way, we can use predictions from models to manage wind turbines (e.g. shutting down) at t+1 based on conditions observed at t time. # Related publication Barré, K., Froidevaux, J. S. P., Sotillo, A., Roemer, C., & Kerbiriou, C. (2023). Drivers of bat activity at wind turbines advocate for mitigating bat exposure using multicriteria algorithm-based curtailment. Science of The Total Environment, 866, 161404. https://doi.org/10.1016/j.scitotenv.2023.161404 # Contacts Kévin Barré ([e-mail])Christian Kerbiriou ([e-mail])