TRANSPIRATION IS ONE OF THE MOST IMPORTANT SURFACE WATER FLUXES BUT CURRENT ESTIMATES VARY WIDELY. ACCURATE ESTIMATES OF TRANSPIRATION ARE ESSENTIAL AS THEY CONTROL PLANT WATER USE AND ITS EFFICIENCY AND AS SUCH CONNECT THE WATER CYCLE TO THE CARBON CYCLE BUT ALSO BECAUSE THEY CAN PROVIDE ESTIMATES OF PLANTS' RESILIENCE TO DROUGHTS.WE PROPOSE TO DEVELOP A NEW REMOTE SENSING BASED ALGORITHM TO BETTER ESTIMATE THE PARTITIONING OF EVAPOTRANSPIRATION TOWARDS TRANSPIRATION. AT PRESENT VERY FEW CONSTRAINED ESTIMATES OF TRANSPIRATION EXIST. WE PROPOSE TO BUILD ON A RECENT MACHINE-LEARNING ESTIMATE OF EVAPOTRANSPIRATION BASED ON ONLY REMOTE SENSING DATA NAMED WECANN [ALEMOHAMMAD ET AL. 2016] TO ESTIMATE TRANSPIRATION. TO CONSTRAIN TRANSPIRATION DURING THE LEARNING PROCESS WE WILL USE TWO MAIN PRODUCTS: 1) GOME-2 SOLAR-INDUCED FLUORESCENCE AN INDICATOR OF PLANT GROSS PRIMARY PRODUCTIVITY THUS DIRECTLY TO THE RATE OF PLANT TRASNPIRATION THROUGH LEAF GAS EXCHANGE AND 2) ISOTOPIC OXYGEN ATMOSPHERIC COMPOSITION AN INDICATOR OF SURFACE WATER PARTITIONING. THE RETRIEVAL WILL BE TESTED AND VALIDATED AGAINST 1) EDDY-COVARIANCE FLUX MEASUREMENTS USING THE CONCEPT OF UNDERLYING WATER USE EFFICIENCY (RELATING TRANSPIRATION TO GPP) AND AGAINST A NEW PRODUCT CALLED 2) SAPFLUXNET: A COLLECTION OF SAPFLOW RECORD ACROSS THE GLOBE. THE NEW PRODUCT WILL BE AN ESSENTIAL TOOL FOR LAND-SURFACE MODEL DEVELOPEMENT AND FOR DROUGHT MONITORING.