TRANSPIRATION COMPRISES ABOUT 65% OF GLOBAL EVAPOTRANSPIRATION (ET) AND BY INFLUENCING THE PARTITIONING OF WATER AT THE SURFACE HAS A FIRST-ORDER EFFECT ON HOW PRECIPITATION VARIABILITY IS TRANSLATED TO DOWNSTREAM WATER RESOURCES AND ECOSYSTEM SERVICES. MOST LAND SURFACE MODELS TREAT (EVAPO-)TRANSPIRATION AS A FUNCTION OF SOIL MOISTURE. INSTEAD TRANSPIRATION PHYSIOLOGICALLY DEPENDS ON LEAF WATER POTENTIAL AND IS THUS INFLUENCED BY PLANT WATER STORAGE AND THE MOVEMENT OF WATER THROUGH THE PLANTS AS WELL AS BY SOIL MOISTURE. BECAUSE OF THE MYRIAD FEEDBACKS BETWEEN PLANT WATER STRESS STOMATAL CLOSURE LEAF AREA INDEX AND ATMOSPHERIC CO2 AND VPD CONDITIONS CAPTURING THESE PLANT HYDRAULIC EFFECTS IS CRITICAL FOR CORRECTLY CAPTURING THE RESPONSE OF TRANSPIRATION TO WATER STRESS AND PREDICTING FUTURE FLUXES OF ET. HOWEVER WHILE INCLUDING THE EQUATIONS NECESSARY TO ACCOUNT FOR PLANT HYDRAULICS IN LAND SURFACE MODELS IS STRAIGHTFORWARD PARAMETERIZING THEM IS A KEY DIFFICULTY. HYDRAULIC TRAITS ARE GENERALLY MORE VARIABLE WITHIN PLANT FUNCTIONAL TYPES THAN ACROSS THEM SO STANDARD METHODS OF PARAMETERIZATION BASED ON PLANT FUNCTIONAL TYPES ARE NOT RELIABLE NOR INFORMATIVE. HERE WE WILL USE REMOTE SENSING OBSERVATIONS TO INFER SUCH PARAMETERS AT MODEL-INFORMATIVE SCALES BUILDING A STEPPING STONE FOR THE NEXT GENERATION OF EVAPOTRANSPIRATION APPROACHES FOR LAND SURFACE AND EARTH SYSTEM MODELS. IN THIS PROJECT WE WILL BUILD A PLANT HYDRAULIC MODEL-DATA FUSION SYSTEM (PHODDS) TO SIMULATE TRANSPIRATION AND DETERMINE SPATIALLY VARIABLE EFFECTIVE PLANT-SCALE HYDRAULIC TRAITS. THE PHODDS SIMULATES THE LAND SURFACE WATER BALANCE AS CONSTRAINED BY UNCERTAIN REMOTE-SENSING BASED OBSERVATIONS. IT UPDATES MODEL PARAMETERS TO MATCH THESE OBSERVATIONS RATHER THAN MERELY UPDATING THE MODEL STATE AS IN DATA ASSIMILATION. THE PROPOSED SYSTEM IS ENABLED BY THE USE OF TWO (UNCERTAIN) OBSERVATIONAL CONSTRAINTS: TOTAL ET FROM ALEXI AND MICROWAVE VEGETATION OPTICAL DEPTH WHICH MY GROUP HAS RECENTLY SHOWN CARRIES INFORMATION ABOUT LEAF WATER POTENTIAL. WE WILL ALSO ASSIMILATE SURFACE SOIL MOISTURE. PHODDS WILL BE USED TO INVESTIGATE FOR WHAT CLIMATES AND PLANT TRAITS CONVENTIONAL METHODS OF MODELING TRANSPIRATION AS A FUNCTION OF SOIL MOISTURE LEAD TO THE GREATEST ERROR DURING DROUGHT. BY INVESTIGATING THIS QUESTION WITH PARAMETERS THAT ARE LOCALLY OPTIMIZED BY THE MODEL-DATA FUSION SYSTEM WE WILL BE ABLE TO ISOLATE THE ROLE OF MODEL STRUCTURE FROM THE ROLE OF PARAMETERIZATION ERRORS IN DETERMINING TRANSPIRATION MODEL QUALITY. WE WILL FURTHER USE PHODDS TO INVESTIGATE THE CONTROLS ON THE BUFFERING EFFECTS OF TRANSPIRATION WE WILL DETERMINE WHAT HYDRAULIC TRAITS ARE LINKED TO SENSITIVITY TO DIFFERENT TYPES OF DROUGHT (E.G. VAPOR PRESSURE DEFICIT OR SOIL MOISTUREDRIVEN EFFECTS) ENABLING IMPROVED UNDERSTANDING OF HYDROLOGIC RESILIENCE TO FUTURE DROUGHT OCCURRENCE. DIFFERENCES IN MODEL TUNING AND RESOLUTION MAY PREVENT PARAMETERS FROM PHODDS TO BE DIRECTLY USEABLE IN THE NEXT GENERATION OF LAND SURFACE OR EARTH SYSTEM MODELS. HOWEVER BECAUSE OF THE LIMITED INFORMATION OF PLANT FUNCTIONAL TYPES ABOUT HYDRAULIC TRAITS AN ALTERNATIVE WAY OF DETERMINING LIKELY TRAIT VALUES IS NECESSARY FOR MODEL PARAMETERIZATIONS. IN THE LAST COMPONENT OF THIS PROPOSAL WE WILL BUILD A PREDICTIVE SYSTEM THAT CAN CLASSIFY ECOSYSTEMS INTO ALTERNATIVE CLUSTERS WITH SIMILAR TRAITS VALUES. THIS SYSTEM WILL BE OPTIMIZED TO MAXIMIZE PREDICTABILITY OF THE TRAITS AND WILL DEPEND ECOSYSTEM PROPERTIES THAT HAVE PREVIOUSLY BEEN SHOWN TO INFLUENCE HYDRAULIC TRAITS: CANOPY HEIGHT TOPOGRAPHY MEAN CLIMATE ETC. THIS PREDICTIVE SYSTEM SHOULD DRAMATICALLY INCREASE THE USABILITY OF PHODDS RESULTS ACROSS A VARIETY OF MODELS.