Satellite observations from GOES are used within the AVHRR Pathfinder Atmospheres-Extended (PATMOS-x) retrieval package to determine cloud properties which are then coupled with GFS wind observations to advect cloud positions forward in time. Advection is performed on a per-pixel basis as well as within self-similar cloud groups identified by a novel cloud-grouping algorithm utilizing thresholds of comparisons with the PATMOS-x input. Advected cloud positions, which account for parallax and shadow locations, are then used to run a single-column radiative transfer model to compute the forecasted surface irradiance value.
Forecasts using the advection technique during calendar year 2012 are presented, along with validation results from the SURFRAD network. Issues and benefits from using satellite advection techniques versus traditional numerical weather prediction model forecasts are also presented.