Application of the Scale Decomposition Technique for Assessing Precipitation Forecasts of a High-Resolution Weather Research and Forecasting Advanced Research WRF (WRF-ARW) Ensemble for a Complex Mixed Precipitation Event near the Washington, DC, Area

An ensemble of short-range weather forecasts was generated using the Advanced Research version of the Weather Research and Forecast model for a challenging winter precipitation forecast event near Washington, DC, on 9 February 2016. An assessment was conducted to quantify the uncertainty of the precipitation nowcasts produced by the 28-member ensemble. The assessment used a combination of tools to quantify uncertainty, which was the ensemble mean and observed accumulated precipitation, time series plots of forecast and observed accumulated precipitation, rank histograms, and 2-D observation ranks. These tools showed 1) uncertainty in the location of a maximum in accumulated precipitation, 2) the presence of inadequate spread in forecast precipitation relative to the spread in the observations, and 3) that there was spatial variation in the bias. The scale-decomposition technique was applied to better quantify the uncertainty by isolating the precipitation maximum and assessing the quality of the forecast structure in terms of the spatial scale of the error. The results showed that this technique provided an assessment of model skill as a function of precipitation threshold value and spatial scale. It enabled the separation of the larger errors, attributable to displacement, from the smaller errors attributable to smaller scale processes.