Impacts drive lunar rockfalls over billions of years

This catalog contains all lunar rockfalls mapped in LRO NAC images using a Convolutional Neural Network, as presented and discussed in Bickel et al. (2020): nature.com/articles/s41467-020-16653-3, capped at 80°N and 80°S and 60°N and 60°S (only images with solar incidence angles between 10 and 80°/60° have been used). The 80°-capped catalog (used for Bickel et al., 2020) contains detections with CT values above 0.6 (136,610 detections) and 0.8 (22,108 detections).

The 60°-capped catalog is a re-run performed in 2021 that contains detections with CT values above 0.6 (92,843 detections), 0.7 (44,859 detections), and 0.8 (15,134 detections). Please note that the 80°-capped catalogs erroneously contain a number of duplicate entries. The catalogs contain information about (sequentially): NAC_ID (image id), C_LON (detection box center longitude), C_LAT (detection box center latitude), UL_LON (detection box upper left longitude), UL_LAT (detection box upper left latitude), R_DIAM (estimated rockfall diameter in meter), X_LEN_P (detection box x horizontal length in pixel), Y_LEN_P (detection box y vertical length in pixel), X_LEN_M (detection box x horizontal length in meter), Y_LEN_M (detection box y vertical length in meter), UL_X_IMG (detection box upper left image x horizontal coordinate), UL_Y_IMG (detection box upper left image y vertical coordinate), LR_X_IMG (detection box lower right image x horizontal coordinate), LR_Y_IMG (detection box lower right image y vertical coordinate), C (network confidence [posterior probability] in percent), RES (NAC spatial resolution in m/pixel).