Influence of the ambient noise level of Raspberry Shake seismographs on the performance of the Ayiti-séismes platform

International audienceThe development of low-cost Raspberry Shake (RS) seismological stations capable of providing scientifically relevantseismological information in near-real time has enabled a Haitian citizen network to be set up as part of a participatoryseismology experiment that Started in July 2019. This network currently operates 12 RS instruments. These are associatedwith broadband stations (BB) of the Bureau of Mines and Energy of Haiti (BME) and the Canadian National SeismographNetwork (CNSN) for the continuous monitoring of seismicity in Haiti. All these stations are combined with other stations inthe region for the detection and automatic location of earthquakes in Haiti, with an open broadcast from the Ayiti-Séismes(https://ayiti.unice.fr/ayiti-seismes/) platform.While the BB sensors are placed in relatively quiet places and benefit from a professional installation, the RS stations areplaced in private homes, sometimes in their living room or garden. Due to the variability of the quality of their internetcommunication and power, the percentage of available RS recordings varies considerably from one site to another. Moreover,these recordings are only available in a reduced range of frequencies and are a priori of lower quality, as they are attenuatedby local ambient noise. We propose here to quantify the effect of this noise and the impact of the environment on the qualityof the recorded seismic signals.To do so, we analyzed a set of data sets, from a period of one month with a minimum availability of 84.8%, from ten RS andone SDDR broadband station located in the Dominican Republic. To quantify the noise, we estimate the Power SpectralDensity (PSD) of the noise at 1/8 octave bands that we accumulate in 1 dB bands. Then, we perform a static analysis of theset of PSD curves computed at each of the study stations to present the final result as a probability density function (PDF).For the final evaluation of the ambient noise variation around each station, we retain the highest probability noise level whichis the statistical mode of the noise levels from the PDFs. The results thus obtained show overall that the environment of thestations is very noisy. However, RS R50D4, located 21 km from the epicenter of the August 14, 2021 earthquake, has anacceptable noise level compared to the SDDR station, taken as the reference station in this study. Based on the appreciablenoise level of this station, we conclude that a favorable path to detect and locate low magnitude and ultimately decrease themagnitude of completeness would be to multiply the RS while following the installation protocol of R50D4. Despite a highambient noise level for the majority of RS stations, a large number of seismic events could be detected and located in Haitiand near its borders.In the near future, we plan to install 30 additional RS stations in Haiti to improve the detection and characterization of eventsof magnitude lower than 2.5