HYPOTHESES FOR THE GEOLOGIC EVOLUTION AND THEIR IMPORTANCE IN SEEKING SIGNS OF LIFE FOR THE 4 REMAINING LANDING SITES FOR MARS 2020 ARE TIGHTLY COUPLED TO THEIR MINERALOGY. FOR EXAMPLE AT THE NE SYRTIS AND JEZERO LANDING SITES A REGIONAL OLIVINE-BEARING UNIT IS VARIABLY ALTERED TO CONTAIN MG-CARBONATE (MAGNESITE). THE ALTERATION PATHWAY OF OLIVINE TO MAGNESITE CAN TAKE SEVERAL FORMS WITH DIFFERENT INTERMEDIATE AND FINAL PRODUCT ASSEMBLAGES DEPENDING ON THE DIFFERENT ENVIRONMENTAL CONDITIONS OF FORMATION. TO DATE THE ONLY PHASES POSITIVELY IDENTIFIED ARE OLIVINE AND MAGNESITE WHILE THE PRESENCE OF HYDRATED SILICA TALC AND SERPENTINE HAS BEEN PROPOSED BUT THE PHASES HAVE NOT BEEN POSITIVELY VERIFIED IN ABUNDANCES NECESSARY TO SUPPORT THE PROPOSED FORMATION HYPOTHESES. THE PROTOLITH ORIGIN CHARACTER IS ALSO IMPORTANT AND IS COUPLED TO ITS ORIGIN. THIS CRYSTALLINE IGNEOUS OLIVINE-BEARING UNIT IS HYPOTHESIZED TO BE FROM IMPACT MELT VOLCANIC FLOWS AIRFALL TEPHRA OR SILICATE RAIN FOLLOWING THE ISIDIS BASIN-FORMING IMPACT. RESOLUTION AMONG THESE HYPOTHESES IN ADVANCE OF THE MARS 2020 CACHING MISSION WOULD HAVE A VALUABLE IMPACT ON SCIENCE RETURN IF THE SAMPLES ARE RETURNED TO EARTH. EACH OF THE HYPOTHESES IS BACKED UP BY DATA ANALYSIS BUT THEY DIFFER IN THEIR TREATMENT OF THE HYPERSPECTRAL DATA FOR MINERALOGIC DETERMINATION. HOWEVER CONVOLUTION OF SPECTRAL SIGNATURES IN MIXTURES CAN OBSCURE MINERAL SIGNATURES LIMITING TRADITIONAL PRESENCE ABSENCE APPROACHES. TO MOVE BEYOND THESE LIMITATIONS OTHERS HAVE USED ADVANCED SPECTRAL MIXING AND CONVOLUTION ALGORITHMS TO PROPOSE RICH MINERAL ASSEMBLAGES. WHILE PROMISING THESE RESULTS AWAIT VALIDATION AND CONFIRMATION THROUGH RIGOROUS TESTING. THE PROPOSED EFFORT HERE WILL RIGOROUSLY EXAMINE HYPOTHESES FOR MINERAL ASSEMBLAGES AND GEOLOGIC EVOLUTION WITH ADVANCED DATA ANALYSIS. WE WILL COMPARE AND CONTRAST 4 ADVANCED MINERAL CHARACTERIZATION ALGORITHMS IN THE CONTEXT OF THE SCIENCE OF THE 3 REMAINING MARS 2020 LANDING SITES. THE ALGORITHMS THAT WILL BE COMPARED ARE (1) RADIATIVE TRANSFER UNMIXING FROM OPTICAL CONSTANTS (2) SPARSE UNMIXING WITH LARGE LIBRARIES OF ENDMEMBERS (3) TARGET TRANSFORMATION FOR MINERAL PRESENCE/ABSENCE DETERMINATIONS IN CLUTTERED SCENES AND FOR MINERALS WITH SUBTLE FEATURES AND (4) MACHINE LEARNING APPROACHES. THE PROJECT WILL FOCUS ON THE BEST MINERAL EXPOSURES IN THESE FIVE SITES. THE SITES THAT HAVE ROVER COVERAGE (OPPORTUNITY COLUMBIA HILLS AND CURIOSITY) HAVE A MEASURE OF GROUND TRUTH TO COMPARE WITH. THE MULTIPLE WORKING HYPOTHESES FOR THE GEOLOGIC EVOLUTION WILL BE EVALUATED AGAINST THESE MORE ROBUST ANALYSES OF THE MINERAL. THE PROPOSED WORK IS BROKEN INTO FOUR TASKS. (1) EVALUATION AND COMPARISON OF THE FOUR TECHNIQUES USING TWO SOURCES OF LABORATORY MIXTURE DATA: 1) 8 SUITES OF MIXTURES THAT HAVE EXISTING LABORATORY REFLECTANCE SPECTRA AND 2) LABORATORY HYPERSPECTRAL DATA OF AQUEOUS MINERALS MIXED WITH A NEW MARS SOIL SIMULANT. THE EVALUATION WILL INCLUDE THE ACCURACY OF THE APPROACHES IN DETERMINING THE CORRECT MINERAL ASSEMBLAGES AND WHERE APPROPRIATE ABUNDANCES. (2) ANALYSIS OF HIGH QUALITY CRISM OBSERVATIONS OF THE 4 MARS 2020 LANDING SITES WITH THE FOUR TECHNIQUES. THIS WILL REQUIRE CAREFUL TAILORING OF THE TECHNIQUES TO THE UNIQUE QUALITIES OF CRISM DATA COMPARED TO LABORATORY DATA. (3) CRITICAL EVALUATION OF THE QUALITY OF THE APPROACHES INDEPENDENTLY USING STATISTICAL MEASURES OF GOODNESS OF FIT AND INTER COMPARISON OF THE RESULTS. (4) SYNTHESIS AND RECOMMENDATION FOR FUTURE APPLICATIONS OF THE METHODS. THIS RESEARCH ADDRESSES SEVERAL ELEMENTS OF THE MARS DATA ANALYSIS CALL FROM THE 2018 ROSES. THIS WORK WILL CONTRIBUTE TO THE OPEN SCIENCE QUESTIONS AT THE MARS 2020 LANDING SITES BY ENHANCING UNDERSTANDING OF MINERAL ASSEMBLAGES AND ALSO DIRECTLY ADDRESS THE HIGH PRIORITY AREA CHARACTERIZATION OF POTENTIAL LANDING SITES FOR FUTURE MARS EXPLORATION MISSIONS OUTLINED IN THE SOLICITATION NNH18ZDA001N-MDAP.