Publication Date

Summer 2014

Degree Type


Degree Name

Master of Science (MS)


Meteorology and Climate Science


Sen Chiao


atmospheric rivers, data assimilation, mesoscale forecasting, Northern California, orographic precipitation, WRF

Subject Areas

Meteorology; Atmospheric sciences


In this study, data assimilation methods of 3-D variational analysis (3DVAR), observation nudging, and analysis (grid) nudging were evaluated in the Weather Research and Forecasting (WRF) model for a high-impact, multi-episode landfalling atmospheric river (AR) event for Northern California from 28 November to 3 December, 2012. Eight experiments were designed to explore various combinations of the data assimilation methods and different initial conditions. The short-to-medium range quantitative precipitation forecast (QPF) performances were tested for each experiment. Surface observations from the National Oceanic and Atmospheric Administration's (NOAA) Hydrometeorology Network (HMT), National Weather Service (NWS) radiosondes, and GPS Radio Occultation (RO) vertical profiles from the Constellation Observing System for Meteorology Ionosphere and Climate (COSMIC) satellites were used for assimilation. Model results 2.5 days into the forecast showed slower timing of the 2nd AR episode by a few hours and an underestimation in AR strength. For the entire event forecasts, the non-grid-nudging experiments showed the lowest mean absolute error (MAE) for rainfall accumulations, especially those with 3DVAR. Higher-resolution initial conditions showed more realistic coastal QPFs. Also, a 3-h nudging time interval and time window for observation nudging and 3DVAR, respectively, may be too large for this type of event, and it did not show skill until 60-66 h into the forecast.