Orographic Quantitative Precipitation Forecast Sensitivity to the Choice of Microphysics Parameterization during a U.S. West Coast Atmospheric River Event

September 8, 2026

The study “Orographic Quantitative Precipitation Forecast Sensitivity to the Choice of Microphysics Parameterization during a U.S. West Coast Atmospheric River Event” has been published in Weather and Forecasting. CW3E-funded contributors Sawyer Smith (CW3E and North Carolina State University) and Gary M. Lackmann (North Carolina State University) authored the paper along with Jason M. Cordeira (CW3E and North Carolina State University). Funding was provided by CW3E’s California Department of Water Resources AR Program awards.

This work supports the priority in the CW3E 2025–2029 Strategic Plan to better understand the science of atmospheric rivers (ARs) and extreme precipitation, including physical processes, forecasting, and impacts. Forecasting orographic precipitation during landfalling ARs remains a persistent challenge, and one under-tested source of that uncertainty is the bulk microphysics parameterization (BMP), which governs how a model represents cloud and precipitation processes. Traditional BMPs sort ice into fixed categories and convert mass abruptly between them, whereas a newer generation of schemes allows ice-particle properties to evolve continuously. Using the CW3E West-WRF (Weather Research and Forecasting) 34-year reforecast dataset, the authors generated a 14-member, 168-hour re-reforecast ensemble of a high-impact AR (7–10 January 2017), varying only the BMP.

Relative to Stage IV precipitation analyses and gauge observations, the ensemble mean overpredicted precipitation along the windward slopes of the Sierra Nevada by at least 75–100 mm while underpredicting leeward precipitation by 25–50 mm. Several newer schemes, including the Ice-Spheroids Habit Model with Aspect-Ratio Evolution (ISHMAEL), Morrison, and the Stony Brook University–Lin scheme (SBU-YLIN), reduced this bias couplet by redistributing precipitation across the barrier (Fig. 1). Mean Sierra Nevada quantitative precipitation forecast (QPF) correlated strongly with the drying ratio (r = 0.75), which accounted for roughly 56% of the intermember variance (Fig. 2a); upstream water flux varied by less than 5% (Fig. 2b). Schemes producing more graupel exhibited larger drying ratios and greater windward precipitation, while some newer schemes featured up to 5% lower drying ratios, consistent with reduced windward bias and more realistic cold-season orographic precipitation. Further case studies are needed to generalize these findings.

Figure 1. The 72-h QPF difference fields (member minus ensemble mean) for each reforecast (mm; shaded as in legend). Contours enclose departures from the Stage IV quantitative precipitation estimate exceeding the 75th percentiles of the overprediction (magenta) and underprediction (black) error distributions pooled across the ensemble. Because the thresholds are ensemble-relative rather than absolute, the contoured area is directly comparable between schemes. (Fig. 5 from Smith et al. 2026).

Figure 2. (a) Scatterplot mapping ensemble member 72-h mean QPF in a pre-defined Sierra averaging box to its corresponding drying ratio (DR). (b) Scatterplot mapping ensemble member 72-h mean QPF in the Sierra averaging box to its corresponding upstream (i.e. offshore) water flux. The DR correlation is significant (r = 0.75, 95 % CI [0.36, 0.92], p = 0.002); the upstream-flux correlation is not (r = 0.49, 95 % CI [−0.06, 0.81], p = 0.075). Confidence intervals reflect the small ensemble size (n = 14). The Thompson 2mom and Morrison schemes are labeled as T2MOM and MORR, respectively. (Fig. 8 from Smith et al. 2026).

Citation:

Smith, S., Lackmann, G. M., & Cordeira, J. M. (2026). Orographic Quantitative Precipitation Forecast Sensitivity to the Choice of Microphysics Parameterization during a U.S. West Coast Atmospheric River Event. Weather and Forecasting (published online ahead of print 2026), e250198, Article e250198. https://doi.org/10.1175/WAF-D-25-0198.1