Sensitivity of AGCM‐simulated regional JJAS precipitation to different convective parameterization schemes

Abstract

ABSTRACTThis study examines the effects of convective parameterization schemes (CPSs) on the simulated June–September (JJAS) precipitation climatology, variability and predictability over southwestern Arabian Peninsula and northeast Africa during 1981–2014 within an atmospheric global climate model framework. The three CPSs used are: the simplified Arakawa–Schubert (SAS) scheme; the SAS scheme with Tokioka modification (STOK) and the Emanuel (EMAN) scheme coupled with a probability distribution function‐based cloud parameterization scheme. SAS and STOK overestimate JJAS total precipitation over the selected domain compared to observations, while EMAN underestimates. EMAN provides the most realistic distribution of precipitation and spatial distribution of convective precipitation despite the tendency for underestimating the total precipitation. The principal patterns of precipitation variability (as measured by the leading empirical orthogonal functions) in EMAN and STOK are clearly related to the observed pattern, while the maximum variability in SAS occurs in a completely different location. The El Niño Southern Oscillation‐related JJAS precipitation teleconnection simulated by the SAS scheme is very weak as compared to observations and other two CPSs. The signal‐to‐noise ratio estimated by SAS and STOK is very low as compared to EMAN. Correlation analysis shows that EMAN performs better than the other two CPSs. Low (high) outgoing longwave radiation (OLR) values are indicative of enhanced (suppressed) convection and hence more (less) cloud coverage. The root mean square error estimated for OLR in EMAN is lower than that of the other two CPSs. The vertical structure of specific humidity and temperature shows less error in EMAN than that of the other two CPSs, which could be a reason for better predictability over the region of interest.