The minimum confidence limit for diameters in crater counts

Abstract

Size-frequency distribution (i.e., SFD) of impact craters are frequently used to study relative and absolute model ages for planetary surfaces. It is well accepted that crater statistics are incomplete and crater recognition is inaccurate at diameter ranges approaching a few pixels of the images used, but there is no benchmark about the minimum confidence limit for diameter (Dmin). Here we resolve this problem by comparing the SFD of each studied crater population using a same image dataset that has been down-sampled by different folds. A systematical study of several different-aged lunar crater populations using various image datasets found that crater statistics can be regarded as completed and crater rim identification can be regarded as accurate at diameters larger than about 10 pixels of the base images. Integrating other potential difficulties in crater counts (e.g., illumination conditions, personal biases, topography roughness, size of counting areas) would indicate that 10 pixels should be considered as a conservative suggestion. Using 10 pixels as Dmin could substantially enhance the reliability of crater counts, especially for prevalent high-resolution regional geological studies that employ populations of small craters.