Parameter Optimization of the SAR POT Method for NCC Stacking and Yanong Glacier Velocity Extraction

Objective: The surface movement characteristics of mountain glaciers are crucial for identifying potential glacial chain disasters and studying global climate change. The pixel offset tracking (POT) method based on image matching is an effective method of detecting large-gradient deformations on glacier surfaces. However, the accuracy of the POT method is closely related to image resolution. The Sentinel-1 imagery, acquired at regular intervals, has a relatively low resolution (approximately 5m×20 m). In the case of small window matching on narrow mountain glaciers, the method is highly susceptible to noise, which reduces normalized cross-correlation (NCC) peak values and hinder the acquisition of reliable glacier displacement measurements.Method: To improve NCC peak values and reduce noise interference, this study utilizes NCC stacking from time-series image pairs, which enhances the accuracy of the POT method with small window matching, albeit at the expense of some temporal resolution. By comparing the effects of different parameter settings—such as the number of stackings, search distance, and matching window size—on the experimental results, the study offers an optimal strategy for parameter selection when calculating displacement using the POT method. Results: The settings of parameters like search distance and matching window size in POT calculations significantly influence experimental outcomes. Compared to single-image pair calculations with the same parameter settings, NCC stacking significantly improves the spatial continuity and coverage of the velocity field of the Yanong Glacier. Additionally, the mean and variance of velocity residuals in stable regions are reduced to varying degrees, with most areas exhibiting a strong signal-to-noise ratio gain. Conclusion: The settings for the search window and matching window are crucial for ensuring accurate displacement tracking. Larger windows tend to produce better displacement field results but may smooth out the edges and local details of the glacier. Using NCC stacking allows for robust tracking of glacier velocity on smaller windows, achieving coverage comparable to large-template pairwise NCC. This makes NCC stacking the preferred method, especially when the template size is constrained by the area of small-scale glaciers.