Implementation of IES Data Spatial Interpolation Using KNN and Improved Genetic Algorithm

The Inverted Echo Sounder (IES) measures the round-trip propagation time of sound waves from the seafloor to the sea surface to invert marine dynamic environments, making it one of the most important measurement devices in the field of deep-sea acoustic tomography. However, issues such as sparse spatial deployment prevent spatially precise matching in studies based on multi-source data characteristics. This paper proposes an interpolation method combining K-Nearest Neighbors (KNN) and an improved Genetic Algorithm (GA). The method consists of two steps. The first step uses the KNN interpolation method to directly fill missing nodes with sufficient real neighbors by mean imputation. The second step employs the genetically optimized algorithm to interpolate missing nodes without sufficient real neighbors. The GA interpolation constructs a fitness function based on the physical constraints of the sea surface height (SSH) and the standard deviation of the information matrix, achieving the filling of missing nodes without neighbors through statistically constrained initialization and statistically constrained mutation. This paper compares the proposed method with common spatial interpolation algorithms, demonstrating the superior performance of the KNN + improved GA. In addition, comparative experiments are conducted on four different optimization algorithms, showing that the improved GA achieves excellent results with minimal computational cost.