Nature vs Nurture: Relative Influence of Inherited and Environmental Factors in Convective-Scale Ensemble Spread

A well-known issue with the use of convective-scale ensembles is the lack of spread between members when compared to the eventual verification, especially for precipitation patterns. This lack of spread provides overconfident forecasts and reduces forecaster trust in these computationally expensive outputs. While much work has been done to improve spread in the convective-scale ensemble itself, relatively little work has been done to explore the culpability of the driving ensemble for this lack of spread. Here, we borrow terminology used in psychology to describe these different sources of convective-scale spread: in the ``nature'' scenario, the spread is simply inherited from the driving ensemble, while in the ``nurture'' scenario the environment is the dominant source of spread (e.g., through the upscale growth of small, convective-scale differences). As such, there is the potential for large spread differences to emerge between the two ensembles under the ``nurture'' scenario.

Our work uses operational ensemble forecast data run by the UK Met Office from summer (JJA) 2023 to explore these connections in detail. We use two methods for assessing the similarity of precipitation patterns between the nested and driving ensembles: one based on correlating spread data calculated using the dispersion Fractions Skill Score (dFSS), and the other using a novel metric which directly compares driving-nested member pairs within the FSS framework. We find that the nurture scenario provides a larger than average influence in two situations: firstly, at early leadtimes before the boundary information can transmit across the domain, and secondly, during the prediction of convective events. We also link these results to the spread-skill relationship to provide important context for tackling the underspread problem faced by convective-scale ensembles.