Using Geomorphological Indicators to Predict Earthquake Magnitude (MOb-Max): A Case Study from Cao Bang Province and Adjasent Areas (Vietnam)

This paper presents analysis of geomorphological indices for predicting the maximum observed earthquake (MOb-Max) in research region, conducted in three steps: (i) determination of the magnitude of completeness (M-c) of the study area; (ii) assessment of the sensitivity of the geomorphological indicators (10 indices) to earthquake magnitude prediction; (iii) determination of the MOb-Max value. With a multilayer perceptron (MLP) neural network model with one hidden layer (10-14-1) and three methods of data sensitivity assessment for neural networks (Garson's algorithm, Olden's algorithm, Partial Derivatives method) combined with the simple ranking method, we evaluated the importance of 10 geomorphological indices correlated with earthquake magnitude. We find that group-1 is drainage density index (L) and Slope index (S) (average rank 28.5). Group-2 is the mountain front sinuosity index (S-mf), drainage basin shape index (B-s) (average rank 18). The most important group-3 is asymmetry factor index (AF), focal statistics index (Delta GH) and hypsometric integral index (HI) (average rank 13.4). The last group-4 with the lowest earthquake magnitude correlation value includes stream length-gradient index (SL), valley floor width to height ratio (V-f), fractal dimension (FD) (average rank 10.7). Using an MLP neural network with one hidden layer (8-18-1) and eight input geomorphological indices (L, S, S-mf, B-s, AF, Delta GH, HI, SL), the earthquake prediction within the Cao Bang and neighboring provinces becomes MOb-Max = 6.0 and includes the following seven sources: Song (River) Nuoc Den, Song Quay Son, Quang Yen-Song Bang, Cao Bang-Tien Yen, Song Ta Chien, Ha Giang, and Na Hang areas.

Automated summary

This paper presents an analysis of geomorphological indices for predicting the maximum observed earthquake in a research region. The study finds that certain geomorphological indices are correlated with earthquake magnitude, providing insights into earthquake prediction.