During June-September, 2024, active and passive ground-based remote sensors were deployed at the Bondville Environmental and Atmospheric Research Site, in Illinois, to supplement other meteorological long-term observations. This area has extensive agricultural fields where vegetation and surface characteristics are dominated by varying crops throughout the warm season. For this reason, the surface-atmosphere exchange is of primary importance at this site. The goal of this deployment is to study the boundary layer evolution with a specific focus on the morning and evening transition, for which Numerical Weather Prediction (NWP) models seem to have trouble getting the timing correct. In this study, data collected by a Light Detection and Ranging (LiDAR), an Infrared Spectrometer (IRS), a Ceilometer (CL61) and several near-surface in-situ observations are used to assess the High-Resolution Rapid Refresh version 4 (HRRRv4) operational NWP model (3km horizontal grid spacing). The HRRRv4 model errors of surface variables as well as those on wind speed, direction, and temperature within the boundary layer are evaluated as a function of the day of the deployment, of the daily cycle, and of model initialization time. While the daily cycle and vertical structure of wind speed and temperature in the boundary layer are well captured by the HRRR model, errors on planetary boundary layer height and near-surface variables such as temperature, mixing ratio, and wind speed show a dependence on the time of the year, being larger during the peak summer months, and a distinct diurnal signature, with HRRRv4 being drier and warmer than the observations during the daytime. Similarities with comparable results found for instance in the Southeastern United States will be highlighted. The boundary layer evolution is a result of a combination of multi-scale forcing processes including mesoscale and large-scale processes, radiation, clouds, sensible and latent heat fluxes at the surface. By comparing the observed and simulated boundary layer evolution, atmospheric conditions prone to large model errors can easily be identified. This study will allow the subsequent targeted investigation of the processes responsible for the errors.