KEY SCIENCE&TECH QUESTIONS: OUR INVESTIGATION ADDRESSES THE FIRST OF THE FOLLOWING SCIENCE QUESTIONS THE NEXT SCIENCE QUESTIONS WILL FRAME THE PATH FORWARD FROM THIS PLANNED WORK. (1) IN WHAT MAY BE DISCERNED FROM TOPOGRAPHY AND SATELLITE IMAGING TO WHAT DEGREE ARE MERCURY S CHAOTIC TERRAIN AND HOLLOWS SIMILAR TO EARTH S GLACIAL AND PERMAFROST THERMOKARST AND CHEMICAL KARST? HOW DIFFERENT ARE THEY? (2) WHAT FAMILIAR VOLATILE-RELATED PROCESSES MIGHT HAVE CAUSED SIMILAR LANDSCAPES TO DEVELOP? HOW COULD THESE PROCESSES HAVE FUNCTIONED ON MERCURY? (3) IF THE RESPONSIBLE AGENTS ON MERCURY WERE NOT ICE OR LIQUID WATER AND DID NOT INVOLVE AQUEOUS DISSOLUTION OF SALTS OR AQUEOUS DISSOLUTION OF SOMETHING ELSE THEN WHAT MIGHT THOSE AGENTS HAVE BEEN AND BY WHAT PROCESSES DID THEY CREATE LANDSCAPES THAT ARE SO COMPELLINGLY SIMILAR TO KARST AND THERMOKARST LANDSCAPES ON EARTH? CONCEPTUALLY MIGHT UNEARTHLY PROCESSES (EITHER INVOLVING OR NOT INVOLVING VOLATILES) IF ANY CAN BE IMAGINED CREATE SIMILAR-LOOKING LANDSCAPES? WAS THERE PERHAPS ANOTHER SOLVENT NOT WATER THAT PRODUCED KARST? OR A VOLATILE SOLID NOTICE THAT PRODUCED THERMOKARST? PERHAPS CAN VAPOR DRIVEN PROCESSES PRODUCE SUCH LANDSCAPES? IS IT POSSIBLE THAT NO VOLATILE WHATSOEVER WAS INVOLVED AND SOME OTHER PHYSICAL PROCESSES WERE RESPONSIBLE? WHAT FUTURE OBSERVATIONS PRESUMABLY SPACEBORNE WOULD BE NEEDED TO DIFFERENTIATE THE POSSIBILITIES? TECHNOLOGY QUESTIONS: (1) HOW CAN WE USE MACHINE LEARNING TO PROCESS AND UNDERSTAND MULTIDIMENSIONAL TOPOGRAPHIC PARAMETERS IN TERMS OF LANDSCAPE COMPARISONS AND CHARACTERIZATIONS AND ULTIMATELY MODELS OF LANDSCAPE ORIGINS? (2) CAN WE MAKE EFFECTIVE USE OF MACHINE LEARNING IN INTEGRATED ANALYSIS OF THE DETAILED TOPOGRAPHY FROM MLA AND TRAIN IT WITH SIMILAR LIDAR DATA FROM EARTH (ICESAT-2 ATLAS) AND FOLD INTO IT LOW-RESOLUTION MAPPING OF THE DISTRIBUTIONS OF ELEMENTS ESPECIALLY VOLATILES AND IMAGING OF BRIGHT-BLUISH MATERIALS TO ARRIVE AT A BETTER UNDERSTANDING OF THESE SPECIAL MERCURY TERRAINS? WE PLAN TO MAKE SOME RESTRICTED APPLICATIONS OF MACHINE LEARNING TO ADDRESS MAINLY TECH QUESTION #1 MAKE AS MUCH PROGRESS AS WE CAN ON TECH QUESTION #2 THEN DEVELOP A WHITE PAPER THAT WILL EXPLORE A POSSIBLE ROADMAP FOR FUTURE APPLICATIONS OF MACHINE LEARNING AND OF POSSIBLE FUTURE MERCURY MISSION OBJECTIVES. OUR APPROACH WILL INVOLVE THESE METHODS: (1) USE OF SINGLE LIDAR (MLA) TRACKS OVER THE PHOTOGRAMMETRIC MESSENGER DEM AND MDIS IMAGERY IN AOIS FOR USE IN PROVIDING CONTEXT. THEN DO THE SAME FOR EARTH USING ICESAT-2 LIDAR DATA. (2) USE OF STATISTICAL MEASURES OF LANDSCAPE PROPERTIES WITH COMPARISONS OF SIMILAR-LOOKING TERRAINS ON EARTH AND MERCURY FROM LIDAR DATA AND LANDFORM PLANIMETRIC PROPERTIES FROM IMAGING AND PHOTOGRAMMETRIC DEMS. (3) USE OF MACHINE LEARNING. THIS IS A NOVEL AND EXPLORATORY PART OF OUR RESEARCH DESIGN. TOOLS WE HAVE USED SUCCESSFULLY IN GLACIOLOGY-- BUT APPLIED TO MULTISPECTRAL IMAGES AND DEMS-- WILL BE EXPLORED FOR THEIR POSSIBLE USEFULNESS IN ANALYSIS OF LIDAR DATA FOR MERCURY AND EARTH. THESE MACHINE LEARNING TOOLS INCLUDE SUPPORT VECTOR MACHINES SUPERVISED NEURAL NETWORKS RANDOM FOREST CLASSIFIER AND CONVOLUTIONAL NEURAL NETWORK .