Effects of H₂S Sensor Sensitivity Drift of Multi-GAS: Quantitative Approaches for Continuous Volcanic Gas Composition Monitoring

<p>Volcanic gas composition provides us a crucial clue to investigate magma plumbing and geothermal systems. Sensor-based instruments named Multi-GAS have been used for monitoring the volcanic gas compositions at volcanoes. A sensitivity of sensors changes with time caused by deterioration, masking volcanic signals especially during long-term monitoring. Frequent calibration of the sensors is desirable for precise monitoring; however, that is pragmatically not easy because a location of a targeted volcano is remote and rural in general. Sophisticated evaluation of the long-term changes in the sensor sensitivity has not been made yet. In this study, we examined the sensitivity change of the chemical sensors within the Multi-GAS during long-term observations by comparing with other methods such as gas detector tubes and gas sampling. The volcanic gas compositions were monitored using Multi-GAS at Kusatsu-Shirane volcano and Kuju volcano, Japan. Intermittent gas composition measurements using gas detector tubes and gas sampling were conducted at fumaroles around where the Multi-GAS stations are installed. Some disagreements of the CO₂/H₂S ratios are observed between those measured using the Multi-GAS from those measured using other methods. In such cases, large decreases of the H₂S sensor sensitivity were found by the sensor calibration after the monitoring. We found a roughly linear behavior of the H₂S sensor sensitivity changes with time based on a long-term sensor sensitivity monitoring in a laboratory and propose a simple linear sensitivity correction of the H₂S sensors using the calibration results obtained before and after the monitoring. The corrected Multi-GAS results agree well with the results of other methods. Our results open up a possibility for extraction of volcanic signals from the long-term volcanic gas data streams monitored using the Multi-GAS that are masked by the changes in the sensitivity of the sensors.</p>