In the last decade, rainfall radar has been deployed at volcanoes like Mt. Merapi in Indonesia, and can even cover a whole country like in Japan, where the X-Rain (eXtended Radar Information network) product has been available for local research. However, the linkage between raingage data and radar spatial data (over a 250 m x 250 m grid) still presents discrepancies, and these challenges are particularly accurate in regions of high local-topographic variations like at Mount Unzen in Japan. As the volcano is over the Shimabara peninsula, it is surrounded by the sea with topography locally rising to 1,483 m.

To improve the forecast and to better understand the triggering mechanisms of lahars (volcanic debris-flows) at Mount Unzen, quantifying the spatial distribution of rainfalls is essential, and as a first step it is important to first understand how data taken locally by rain-gages relate to the radar data. Because empirical models have not been able to show any clear correlation, the present contribution has been developing a neural-network with two hidden layers that takes into account the rainfall per hour, the temperature and the wind speed and direction. The model takes a logistic activation function and the loss function is optimized using the Mean Squared Errors and the Mean Absolute Error. The choice of the activation function and the optimizer is the result of running several combinations of optimization functions with different activation functions. Once the best fit was chosen, the sigmoid with a SGD (Stochastic Gradient Descent) optimizer was chosen, and when training the model for 120 cycles, Shimabara station and the XRain data shows an error < 4 mm rainfall, while at the Unzen summit, even after 300 cycles, the validation error remained at 8 mm while the training loss was < 4mm. This shows that location specific functions might be necessary for each location, not only taking into account the weather data, but also the local topographic variability and the topographic position on slopes. Finally, neither the raingage, nor the Xrain are “true data” and the remaining error can also be the result of an error existing in any of the two dataset. Furthermore, the X-Rain data are spatialized data over an area, whereas the raingage is very limited in terms of spatial representativity.