<![CDATA[Full Waveform Inversion (FWI) is a key method in seismic exploration due to its ability to generate high-resolution subsurface velocity models. However, its success is highly dependent on the accuracy of the source wavelet, an incorrect wavelet estimation can lead to unstable inversion results. To overcome this limitation, Source-Independent FWI (SI-FWI) has been developed, which eliminates the need for precise source wavelet information. This study introduces WaveInsight, an application that efficiently implements SI-FWI using the Julia programming language, with Python integration via Devito for wave simulation. The application features a Streamlit-based graphical user interface, enabling users to perform a complete FWI workflow, including data preparation, initial model generation, and inversion. Experiments on synthetic Overthrust data compare three objective functions: Student-t, Mean Squared Error (MSE), and SI-FWI. The results show that SI-FWI outperforms conventional approaches under wavelet uncertainty, Student-t is more robust to noise, while MSE performs well with clean data. Thus, WaveInsight demonstrates its dual role as both an educational platform and a promising tool for advancing geophysical research and industrial applications.]]>
Copyrights © 2025
