Modeling of Uncrewed Aerial Vehicle Uncertainty for Assimilation into Numerical Weather Predication

When considering in-situ measurements of meteorological quantities, a data gap is evident in the atmospheric boundary layer. This gap is formed by the large variability withing boundary layer processes, which are currently under-sampled spatially and temporally by radiosonde measurements, and under sampled vertically by ground-based sensors. One low-cost approach to address this data gap is to utilize uncrewed aerial vehicles (UAVs) equipped with suitable sensors and profiling at regular intervals. In recent years, significant advancements towards this have been made in adapting UAVs for meteorological use including the assimilation of their data into numerical weather prediction. However, most assimilation approaches utilize Kalman-Filter-based assimilation schemes, which rely on a priori knowledge of the measurement uncertainty. Although great strides have been made in reducing this uncertainty in the sensor values, little has been done to address the measurement uncertainty introduced by the non-stationarity and heterogeneity in the wind field which lies below the resolution of the numerical weather prediction and is not captured by the conventional vertical profiles being proposed for operational UAV use. We will describe a series of experiments in which multiple UAVs are flown simultaneously to obtain a horizontal distribution of the heterogeneous wind field. This data is then analyzed to produce a simple model that can be used to estimate the uncertainty associated with horizontal heterogeneity introduced by granular surface processes.