Improving the Prediction of Nocturnal Convection through the Assimilation of Novel Datasets: Observation Impacts and Error Treatment

Abstract

Numerical weather prediction (NWP) models often fail to correctly forecast both the initiation and evolution of nocturnal convection. To improve our understanding of such events, researchers collected a unique dataset of thermodynamic and kinematic remote sensing profilers as part of the Plains Elevated Convection at Night (PECAN) experiment. This dissertation evaluates the impacts made to forecasts of nocturnal convection when assimilating this network that includes atmospheric emitted radiance interferometers (AERIs), Doppler lidars, radar wind profilers (RWPs), high-frequency rawinsondes, and mobile surface observations. Using an advanced, ensemble-based data assimilation system, we first evaluate the impacts of these datasets for a single nocturnal convection initiation (CI) event. Compared to operational forecasts, assimilating the PECAN dataset improves the timing, location, and orientation of the CI forecast. We show that AERIs, RWPs, and rawinsondes produce the largest benefits by enhancing the moisture advection into the region of CI. The impacts of assimilating these datasets also remains large throughout the growth of the CI event into a mesoscale convective system (MCS). Assimilating Doppler lidar and surface data only slightly improves the CI forecasts by enhancing the convergence along an outflow boundary that partially forces the CI. While this case study suggests positive results from assimilating high-frequency profiling data, one single event cannot fully represent the wide diversity of mechanisms and environments that can lead to nocturnal convection. To address additional types of nocturnal CI, we next expand our work into a systematic evaluation of the impact of assimilating a collocated network of high-frequency, ground-based thermodynamic and kinematic profilers collected during PECAN. For 13 nocturnal CI events, we find small but consistent improvements when assimilating thermodynamic observations collected by AERIs. Through midlevel cooling and moistening, assimilating the AERIs increases the skill for both nocturnal CI and precipitation forecasts. Assimilating composite kinematic datasets collected by Doppler lidars and RWPs instead results in slight degradations to the forecast quality, including decreases in skill and traditional contingency metrics. The impacts from assimilating thermodynamic and kinematic profilers often counteract each other, such that we find little impact on the detection of CI when both are assimilated. However, assimilating both datasets improves various properties of the CI events that are successfully detected (timing, distance, shape, etc.). We hypothesize that a lack of flow-dependent methods to assign observation errors likely contributes to the forecast degradations for some cases. This theory motivates our final study which evaluates the impact of using novel methods for assigning observation errors when assimilating ground-based thermodynamic profilers. We find that a static error inflation method results in forecast degradations compared to a reference experiment where no remote sensing data are assimilated. These issues are partially resolved when adaptively inflating the observation errors or when using a method that computes the full observation error based on observation-space diagnostics. Flow-dependent extensions of each method are shown to further improve forecasts compared to their static counterpart by increasing observation errors for problematic retrievals. Assuming that the observation errors are correctly diagnosed, the results from this dissertation suggest that assimilating a network of ground-based thermodynamic profilers can greatly improve forecasts of nocturnal convection.