The Effect of the Physical Parameterizations and the Land Surface on Rainfall in Poland

Abstract

AbstractHigh-resolution numerical experiments were conducted over two separate months to study the effect of different physical parameterizations and different representations of the land surface on the prediction of rainfall events in Poland. The Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS) was used with 2-km grid spacing. Four sets of forecast experiments were performed. The control experiment used a slab model for the surface energy budget, a Lin-based moist physics parameterization, and a Mellor and Yamada (MY) based turbulent kinetic energy (TKE) parameterization; the Louis experiment used a version of the Louis TKE parameterization in place of MY; the LSM experiment used the Noah land surface model (LSM) in place of the slab model; and the Thompson experiment used the Thompson microphysics parameterization instead of the Lin-based microphysics. The forecasts were validated against surface and upper-air observations, as well as radar reflectivity. The Louis parameterization yielded improvements to the total rainfall and small improvements to the near-surface temperature and moisture. The Noah LSM yielded the largest improvements to the prediction of near-surface temperature and moisture, and while it led to correctly forecasting an increase in precipitation in one month, it erroneously predicted a decrease in precipitation in the other month. The Thompson microphysics produced the most skillful precipitation forecasts for late spring, but produced precipitation forecasts that were less skillful than the control experiment for early fall. The use of higher horizontal resolution (0.5 km) for two rain events led to the overprediction of rainfall, but suggested a better distribution of the rainfall.