Southern Andes Volcano Observatory (OVDAS)

Implementation of "Serving TensorFlow models with TFServing"

For volcanic monitoring, it is essential to have an instrumental network that collects data in real time for follow-up and evaluation of volcanic activity. OVDAS currently monitors 45 volcanoes in real time, which translates into a high volume of data that must be processed in real time, a task that is currently performed mainly manually. In recent years, advances in artificial intelligence have allowed the development of tools that perform automatic analysis of multiple types of data, such as seismic signals, infrasound, and images, which makes AI a necessary tool for implementation. In this sense, OVDAS has advanced the implementation of AI in different areas of volcanic monitoring, and the next step is the installation and configuration of a Tensorflow(1) server that serves as a base for the implementation of the different AI algorithms.
The objective of the work to be developed is to configure the Tensorflow server, install the Phasenet(2) model, a model developed for the identification of P and S waves in tectonic seismicity, which has shown potential for the identification of these waves in volcano-tectonic signals, and finally the creation of the data input-output interface. With the above, Phasenet is expected to be used as a tool for real-time monitoring of volcano-tectonic seismicity in OVDAS.

(1)https://keras.io/examples/keras_recipes/tf_serving/#:~:text=TensorFlow%20Serving%20provides%20out%2Dof,well%20as%20HTTP%20inference%20endpoints
(2)https://github.com/AI4EPS/PhaseNet