Surface Change Detection on Venus: An Inter‐Mission Comparison Framework Between Magellan and Future Radar Images

Abstract

Detecting surface changes on Venus by comparing Magellan synthetic aperture radar (SAR) data with future high‐resolution data from Envision and VERITAS will be essential for constraining present‐day volcanism and characterizing planetary surface processes. However, such inter‐mission SAR change detection requires resampling images to comparable pixel spacing, which results in image pairs with different equivalent numbers of looks (ENLs) and this invalidates classical statistical models that assume the two images are identically distributed. To address this issue, we propose an unsupervised change detection method for SAR image pairs with different ENLs, termed GB2KI. This method models the ratio image with the generalized beta prime distribution (GB2) and incorporates it into the minimum‐error thresholding framework. An entropy‐weighted maximum likelihood estimation method is introduced for robust parameter estimation when the changed class occupies only a small fraction of the scene. Additionally, a multiscale post‐processing technique is developed to suppress isolated false detections in the final change map. Simulations using Magellan Cycle 1 and Cycle 3 images of a volcanic plain on Venus, with synthetic lava‐flow changes varying in intensity, extent, and location, show that GB2KI improves overall detection accuracy and reduces false alarms. Further validation is performed using two Earth analog data sets from the 2014–2015 Holuhraun lava flow in Iceland, including Sentinel‐1 images from different imaging modes (1‐year interval) and Radarsat‐1/Sentinel‐1 images (15‐year interval). Experimental results indicate that GB2KI accurately identifies lava‐flow changes in both cases, supporting inter‐mission comparisons between Magellan and future Venus SAR missions.