Shear wave velocity and soil type microzonation using neural networks and geographic information system

Frequent casualties and massive infrastructure damages are strong indicators of the need for dynamic site characterization and systematic evaluation of a site's sustainability against hazards. Microzonation is one of the most popular techniques in assessing a site's hazard potential. Improving conventional macrozonation maps and generating detailed microzonation is a crucial step towards preparedness for hazardous events and their mitigation. In most geoscience studies, the direct measurement of parameters imposes a huge cost on projects. On one hand, field tests are expensive, time-consuming, and require specific high-level expertise. Laboratory methods, on the other hand, are faced with difficulties in perfect sampling. These limitations foster the need for the development of new numerical techniques that correlate simple-accessible data with parameters that can be used as inputs for site characterization. In this paper, a microzonation algorithm that combines neural networks (NNs) and geographic information system (GIS) is developed. In the field, standard penetration and downhole tests are conducted. Atterberg limit test and sieve analysis are performed on soil specimens retrieved during field-testing. The field and laboratory data are used as inputs, in the integrated NNs-GIS algorithm, for developing the microzonation of shear wave velocity and soil type of a selected site. The algorithm is equipped with the ability to automatically update the microzonation maps upon addition of new data.