Abdullah Al Fahad, Ph.D.
Climate Scientist · NASA Goddard Space Flight Center, Global Modeling and Assimilation Office
About
I build forecasting systems where dynamical physics meets generative AI. At NASA GSFC's Global Modeling and Assimilation Office, I co-develop the GEOS-MITgcm coupled atmosphere–ocean model and combine it with PyTorch diffusion and flow-matching models for coupled Earth-system prediction. As Principal Investigator of a NASA-funded project, I pair AI with physics to diagnose, correct, and improve forecast skill from weather, subseasonal-to-seasonal, and decadal timescales.
I earned my Ph.D. in Climate Dynamics from George Mason University in 2020. My research spans coupled model development and adjoint data assimilation, kilometer-scale "Nature Run" simulations for satellite observing-system experiments, and generative machine learning for downscaling, bias correction, and probabilistic ensembles — turning physics into decision-ready products for extreme weather and climate risk.
That work runs end to end, from model development to forecasts people can act on. Weather2Grid is a live experimental forecast that converts official, numerical, and AI weather guidance into calibrated county-level power-outage risk for emergency managers, with the meteorological and impact-model uncertainty reported separately. SURMA-Flow assimilates rain-gauge and satellite observations through a score-guided rectified-flow model to produce BRISHTI-05, a 30-member daily 5-km precipitation reanalysis over Bangladesh. Alongside them sit a global flow-matching ensemble emulator, kilometer-scale hybrid downscaling, and multi-hazard catastrophe-risk layers used by government agencies and commercial insurance and energy partners.
R&D
Dynamical Modeling & Prediction
GEOS-MITgcm coupled model development; kilometer-scale global simulation; S2S-to-decadal prediction.
Data Assimilation & Optimization
Adjoint methods, balanced coupled initialization, Green's-function and Bayesian calibration.
Physics-ML & Generative AI
Diffusion, flow-matching, and super-resolution for downscaling, bias correction, and ensembles.
Climate Forcing & Attribution
Attributing circulation, monsoon, and hydroclimate change to CO₂, aerosol, and SST forcing.
Extreme Events & Climate Risk
Tropical cyclones, floods, and monsoon extremes — probabilistic hazard science.
Earth Action: Applications
Flood mapping, energy-market wind/solar fields, and insurance catastrophe outlooks.