Automatic first-arrival picking method via intelligent Markov optimal decision processes

Abstract Picking the first arrival is an important step in seismic processing. The large volume of data calls for automatic and objective picking. In this paper, we formulate first-arrival picking as intelligent Markov decision process multi-dimensional feature attribute space. By designing a reasonable model, global optimization carried out reward function space to obtain path with largest cumulative value, achieve purpose automatically up arrival. state-value contains distance-related discount factor γ, which enables pick continuity consider lateral avoid bad trace information data. On basis, method paper further introduces optimized model that fuzzy clustering-based structure-based Gaussian stochastic policy, thereby reducing difficulty design, making more accurately automatically. Testing approach field reveals its properties shows it can arrivals has certain quality control ability, especially energy weak (the signal-to-noise ratio low) or there are adjacent complex waveforms shallow layer.