A long-term goal of our work is to find an efficient system for probabilistic planetary boundary layer (PBL) very short-range forecasting (nowcasting) that can be employed wherever surface observations are present. Our approach makes use of a single column model (SCM) to predict the state of a PBL column, and an ensemble filter (EF) to assimilate surface observations. Although the system may be run under different levels of complexity it is not immediately clear that the additional complexity will improve the performance of a PBL ensemble system based on a simple model. We address this question in the frame of Alpert–Stein Factor Separation Methodology (ASFSM) analysis with regard to treatment of parameterized radiation, horizontal advection and assimilation of surface observations. Results show that the added complexity often improves the forecasts under most skill metrics, but that assimilation of surface observations is the most important contributor to improved skill.

Introduction

Successful nowcasting and forecasting of the state of the PBL is of value for a wide range of practical forecasting applications, and the aim of this work is to find an efficient method for probabilistic nowcasting (0–3 h forecasting) of the state of the PBL wherever surface observations are available. Convective initiation and forecasted precipitation have shown sensitivity to the PBL structure (Crook 1996; McCaul and Cohen, 2002; Martin and Xue, 2006). Air-quality analysis and plume dispersion can benefit from accurate PBL analysis of stability and mixing depth (e.g., Kumar and Russel, 1996; Shafran et al., 2000).