Evaluation of PBL Parameterizations for Modeling Surface Wind Speed during Storms in the Northeast United States

Abstract

his study identifies conditions that determine errors in numerical simulations of 10-m wind speed over moderately complex terrain, emphasizing winds that lead to overhead power-line damage over a subregion of the northeast United States. Simulations with the Mellor?Yamada?Janji? (MYJ) scheme, the Yonsei University (YSU) scheme, and a subgrid-scale topographic drag correction (Topo) applied to YSU are used to investigate error components. The wind speed distribution is dominated by low speeds, which are well depicted by Topo, but are underestimated by the MYJ and YSU schemes. Conversely, moderate and high speeds are underestimated by Topo, and MYJ and YSU perform better across specific ranges. Verification samples are conditioned by season, diurnal cycle, topography, and spatial patterns obtained with a clustering analysis. The systematic error is characterized by a positive bias in low speeds, and as speed increases the biases become more negative. Quantile comparisons, along with systematic and random errors, indicate that beyond the dependence on wind speed itself, errors also depend on seasonal characteristics, indirectly defined by scheme stability profiles. The positive relationship between absolute bias and speed originates in the friction velocity parameterization, and the correction for drag in the Topo scheme exacerbates the effect. The Topo scheme adjusts the total bias and sharpens the bias spread but penalizes moderate and high winds. Clusters reveal that in Topo the bias is primarily driven by wind direction. Excessive correction occurs on terrain-interacting flows, and oceanic flow modulates the adjustment, enhancing the scheme?s performance.