Mesoscale Convective Systems, Rossby Waves, and Error Growth in Global Numerical Weather Prediction

A accurate representation of convection systems remains a critical barrier to advancing earth system prediction across time-scales. On the time-scale of hours, the development of a successful Warn on Forecast convective-permitting ensemble system will take many years (Stensrud et al. 2013). More accurate projections of the future climate will require more improved representation of the competition between shallow and deep convection and the resultant lower-tropospheric mixing (Sherwood et al. 2014). Convection is also an important source of error in global numerical weather prediction (see review in Zhang et al. 2019). Another very specific example that occurs in medium range global prediction systems is forecast busts that occur over Europe in the ECMWF (European Centre for Medium Range Weather Forecasts) modeling system that are often associated with the occurrence of mesoscale convective systems (MCSs) over the United States, especially in the region near the Great Lakes and Mississippi Valley (Grazzini and Isaksen 2002; Rodwell et al. 2018). These forecast busts are associated with specific upper-level flow regimes (e.g., Rodwell et al. 2013; Lillo and Parsons 2017).

Error growth and busts in global numerical weather prediction has been placed in a theoretical framework by numerous studies of predictability that suggest errors grow more rapidly with moist processes at smaller scales (Hohenegger and Schär 2007) and subsequently grow upscale through gravity waves, baroclinic, and/or barotrophic instability (e.g., Zhang et al. 2003, 2007; Bierdel et al. 2017, 2018; Selz and Craig 2015; Judt 2018). Recent work by Baumgart (2018; 2019) has focused on how mesoscale errors near the tropopause grow up-scale through Rossby waves. The tendency for forecast failures over Europe to occur with MCS over specific regions of North America raises the possibility that these forecast busts might be occurring due to systematic short-comings in the model treatment of convection over these regions. These systematic small-scale errors could then subsequently grow up-scale more rapidly in these specific flow regimes. In particular, the upper Mississippi Valley and Great Lakes region falls within the area characterized by a nocturnal maximum in deep convection during the summer (e.g., Wallace 1975). Accurate prediction of nocturnal storms over this region remains a challenge in numerical weather prediction models even on shorter time-scales using high resolution models as the forecasted MCS tend to dissipate too quickly (Pinto et al. 2015). Recent efforts including systematic studies (Haghi et al. 2017; Parsons et al. 2019; Loveless et al. 2019; Haghi et al. 2019) have examined nocturnal convection suggesting that, in contrast to day-time squall lines driven by convectively-generated cold pools, bores and gravity waves are more likely to play a more significant role in maintaining summer-time, nocturnal MCS over the central US.

This study investigates the interactions between MCS, Rossby wave packets, and forecast errors over the central U.S. in an effort to advance knowledge of the causes of forecast busts and to guide model improvements. The time period for the study was June 2015 during the large multi-agency Plains Elevated Convection at Night Experiment (PECAN) field campaign. PECAN was designed to increase our understanding of nocturnal convection and fortuitously contained several forecast busts. Our research shows that the nocturnal, propagating MCS negatively impacting forecasts are indeed characterized by stable layer waves and are associated with strong front-to-rear updrafts with significant, non-symmetric outflows. When these outflows impinge on the jet stream, a variety of Rossby waves packets responses can occur ranging from the amplification of an existing wave packet to the generation of new short-wave length Rossby wave packets that move rapidly downstream. Significant error growth often occurred in associated a strong, well-defined jet stream. The impacts of these MCS and the up-scale growth appear to be challenging to accurately represent in the ECMWF model. The persistent linkage of the MCS to the meridional flow aloft associated with the Rossby wave packet and weak cyclogenesis along frontal boundaries associated with the MCS suggests that the long-lived disturbances have some characteristics of weak diabatically forced Rossby waves.