Examining the Prediction of Heavy Rainfall Events during Winter in the Western United States and Effects from Meteorological Environments

This investigation is aimed at determining verification scores for Weather Prediction Center’s 24-hour quantitative precipitation forecasts of heavy rainfall events, from a 7-day lead time, in the western United States. An original approach is utilized so that both subjective and objective measures are employed. Ten dates between September 2013 and February 2017 are analyzed. Events are scored and then categorized as ‘well-forecasted’, ‘poorly-forecasted’ or ‘very poorly-forecasted’ based on a threat score threshold of 0.5. Scores generally decrease for all cases as both lead time and precipitation threshold increase. Well-forecasted events tend toward higher score spread as threshold increases, but both poor forecast types exhibit fairly little spread. Categorized events are then composited, based on various meteorological variables and environments, and compared to identify signals that could aid future forecasting of heavy rainfall events. Integrated water vapor transport (IVT) is found to be a strong variable indicator of atmospheric rivers (ARs). The effect of particularly anomalous ARs on the predictability of heavy rainfall events is then examined as is mesoscale orographic lift, which can prove difficult to resolve with models but is known to influence localized precipitation amounts. Signals of highly anomalous IVT (99^th percentile or higher compared to climatology) identified in earlier forecasts, and staying consistent throughout the remainder of the forecast period, lead to better forecasts. Poor (and very poor) forecasts are characterized by a later signal of highly anomalous IVT. Signals that appear suddenly on a forecast, as opposed to those appearing gradually, tend to produce poor forecasts. Most events that verified poorly have slightly anomalous IVT versus highly anomalous. Composites also show the presence of a southwesterly low level jet in well-forecasted events. Poor events include stronger jets with more westerly direction, and very poor events regularly lacked a pronounced jet. Boundary layer winds influencing orographic lift are also composited. While the angle of incidence of wind on the coastline is important, the combination of speed and incidence angle matters more. Well-forecasted events tend to have light-moderate winds blowing parallel to the west coast from the south. Winds in poorly-forecasted events have a higher incidence angle, but similar speeds. Winds in very poorly-forecasted events blow parallel to the shore and are quite weak. Additional cases could reinforce these findings and future research will expand the temporal domain, explore alternative case classification techniques and examine Northern Pacific Jet patterns.