[S-VC28] International volcanology

[SVC28-P11] An effect of eliminating the influence of snowmelt from tilt data in monitoring volcanic activity

*Kazuhiro Kimura¹, Katsuhisa Kawashima², Takane Matsumoto² (1.Meteorological College, Japan Meteorological Agency, 2.Research Institute for Natural Hazard & Disaster Recovery, Niigata University )

Keywords:tilt, snowmelt, volcanic activity

Tilt observation that can detect the movement of hydrothermal fluid or magma in real time have been proven at many volcanoes such as Kilauea in Hawaii (Fiske, 1969) and Mt. Merapi in Indonesia (Subandriyo et al., 1998). Thus, it is clear that tilt observation is important for predicting volcanic eruptions or monitoring of volcanic activity. And, it is well known that tilt observation is influenced by rainfall and snowmelt (ex. Jahr et al., 2009). Tilt data influenced by rainfall and snowmelt reduce the ability to detect volcanic activity for several days to several months. Of these, many methods have been proposed to eliminate the influence of rainfall from the tilt data by using rainfall data and tank models, and good results had been obtained (ex. Tanaka, 1971). However, no method has been proposed to eliminate the influence of snowmelt from the tilt data.

We have developed a method for eliminating the influence of snowmelt from the tilt data. In Japan, many tiltmeters have been installed near volcanoes in the 21st century for the purpose of monitoring volcanic activity, but most of Japan's active volcanoes are snow-capped volcanoes and are influenced by snowmelt. If the effect on many tiltmeters in different locations is confirmed, it will be versatile and may be applicable to many tiltmeters around the world.

As a method of eliminating the influence of snowmelt from the tilt data, we devised a model in which a snow tank is added to the tank model that has been recognized as effective as a method of eliminating the influence of rainfall from the tilt data. The data required are precipitation including snow, temperature, sunshine duration or solar radiation, all of which are hourly data. Five-year data from 2012 to 2016 were used to estimate the values for each parameter of the model. Using a model that considered that melting snow and moving through the ground and exerting it on the tiltmeter was the same process as rainfall, the effectiveness was confirmed at many tilt data. However, at tiltmeter located near inhomogeneous terrain, the influence of snowmelt tended to differ between the first half and the second half of the snowmelt season, and such a simple model did not work. Therefore, when we used a model to change the effect of snowmelt in the first half and the second half of the snowmelt period, it was incomplete but improved.

In this presentation, we will introduce the effect of tilt data that eliminates the influence of snowmelt on the monitoring of volcanic activity.