OUR PROJECT WILL DELIVER A MACHINE-LEARNING BASED MODEL TO SPECIFY AND FORECAST NEAR-EARTH SOLAR WIND (SW) CONDITIONS BASED ON SPACECRAFT MEASUREMENTS AROUND L1. ALMOST ALL MAGNETOSPHERE-IONOSPHERE RESEARCH AND SPACE WEATHER FORECASTS RELY ON NEAR- EARTH SW CONDITIONS THAT ARE OBTAINED BY PROPAGATING THE L1 MEASUREMENTS TO THE EARTH S BOW SHOCK. THE INDUSTRY-STANDARD AND WIDELY USED OMNIWEB MODEL IS ALMOST 15 YEARS OLD AND RELIES ON A NUMBER OF ASSUMPTIONS INCLUDING IMF FRONTS BEING PLANER AND SW PROPERTIES STAYING THE SAME ALONG THE FRONT AND OVER TIME. ALTHOUGH THESE ASSUMPTIONS ALLOW ONE TO DEDUCE NEAR-EARTH SW VALUES THEY ARE WIDELY KNOWN TO BE OFTEN INCORRECT. OMNIWEB DOES NOT PROVIDE SW UNCERTAINTY EITHER. WE PROPOSE TO OVERCOME THE LIMITATIONS BY DEVELOPING A MACHINE-LEARNING BASED MODEL USING THE STATE-OF-ART ALGORITHM OF ARTIFICIAL RECURRENT NEURAL NETWORK AND THE LONG HISTORY OF MULTI-POINT SW MEASUREMENTS. BY PROVIDING THIS MODEL TO THE COMMUNITY WE WILL IMPROVE THE OUTPUTS OF ALL MODELS THAT USE SW VALUES AS INPUTS. THIS PROJECT IS BASED ON AN EFFECTIVE AND NOVEL USE OF TWO-DECADES OF SW OBSERVATIONS AND THE OUTCOME WILL BENEFIT A WIDE RANGE OF SPACE-WEATHER-RELATED RESEARCH AND OPERATIONS. IT SITS SQUARELY WITHIN THE PROGRAM GOAL IMPROVEMENTS OF OPERATIONAL CAPABILITIES AND ADVANCEMENTS IN RELATED FUNDAMENTAL RESEARCH AND TO IMPROVE NUMERICAL MODELS AND/OR DATA UTILIZATION TECHNIQUES THAT COULD ADVANCE SPECIFICATION AND/OR FORECASTING CAPABILITIES AND THAT COULD ALSO LEAD TO IMPROVED SCIENCE UNDERSTANDING. PARTICULARLY IT WILL SIGNIFICANTLY PROMOTE THE UNDERSTANDING OF THE SUN AND ITS INTERACTIONS WITH EARTH AND THE SOLAR SYSTEM TO ADVANCE SPACE WEATHER MODELING AND PREDICTION CAPABILITIES APPLICABLE TO SPACE WEATHER FORECASTING AND SUPPORT THE TRANSITION OF SPACE WEATHER MODELS AND TECHNOLOGY FROM RESEARCH TO OPERATIONS AND FROM OPERATIONS TO RESEARCH.