A Look at Winter 2015/2016 Precipitation Forecasts at Eight Locations in the Western U.S

The NOAA/ESRL/Physical Sciences Division has a Water Vapor Flux Tool designed to assist in forecasting heavy precipitation events. This decision-support web-based tool displays observed and modeled hourly-averaged freezing level, wind profiles, upslope flow, integrated water vapor (IWV), IWV flux, and precipitation. All forecasts have a 3-hour verification time. The tool can be accessed at: http://www.esrl.noaa.gov/psd/data/obs/datadisplay.

In this study we evaluate the 3-hr precipitation forecasts from two models that are featured in the tool: (1) the Rapid Refresh (RAP) and (2) the High-Resolution Rapid Refresh (HRRR). The RAP and the HRRR models are run operationally by NOAA and are initialized hourly. The 3-hour forecasts from these models were added to the tool in December 2014. Visual inspection of the tool graphics comparing the 3-hour precipitation forecasts with the observations indicate many discrepancies between the forecast and observed hourly precipitation amounts and in the timing of precipitation events. The goal of our analysis is to characterize and quantify these discrepancies for individual sites. We also assess the upslope wind layer, IWV, and IWV flux forecasts as part of our understanding of the precipitation forecast assessments.

The water vapor flux tool observation sites for this analysis period include five sites along the west coast of the U.S. (between Forks, WA and Bodega Bay, CA) and three inland sites (Chico, CA in the Central Valley, and Troutdale and Wasco, OR, along the Columbia River). This arrangement of sites provides an opportunity to assess forecasts for sites with various kinds of geographic influences and orographic forcing. Our analysis period extends from 1 December 2015 to at least early March 2016 (as of the time of this writing). The 2015/2016 El Niño/Southern Oscillation event brought plentiful rainfall compared to average for this time period for some of the sites, and much more precipitation compared to last year's very dry winter. Various analysis metrics will be presented, including forecast biases and root mean square errors that highlight forecast issues for individual sites.