Two high resolution radar data assimilation case studies as a prelude to phased array radar (PAR) data assimilation in support of UAS operations

Work has been underway at the University of North Dakota to develop sense-and-avoid technologies for Unmanned Aircraft Systems (UAS) operations based upon a combination of technologies including operation, in concert, of a trio of Phased Array Radars (PAR). Part of the development work includes exploring the potential utilization of the PAR network for observing atmospheric phenomena in the lower troposphere as well as the incorporation of the PAR data, via data assimilation, into the Weather Research and Forecasting (WRF) modeling system for both research and operational purposes. Such purposes include the short-term prediction of phenomena hazardous to UAS operations, including explosive convective development and the development of anomalous electromagnetic propagation conditions (e.g., ducts), which could interfere with the sense-and-avoid operation as well as communications with the UAS.

In this paper we present preliminary results of two case studies where we have attempted to incorporate, via data assimilation, radar data within very high resolution simulations (horizontal grids spacings less than 1 km on the innermost grid within a quadruply-nested grid hierarchy). Our goal is to examine the potential utility of the PAR data in improving short-term (less than 1 hour) predictions of conditions hazardous to UAS operations. In both cases we will assimilate the data at various resolutions on the individual grids via subsampling to the coarser grids and interpolation to the finest sub-1 km grid.

The first case involves explosive convective development associated with the Northwood, ND, tornado outbreak that occurred in late August 2007, and is considered an extreme case of the scenarios we expect to encounter in UAS operations. The second case involves more moderate, spatially extensive convection, but in an area (Saudi Arabia) where conventional observational data that might be assimilated into any modeling system are relatively sparse, thus increasing the potential impact of the radar assimilation.

In our presentation, we examine the impact of various methodologies for data assimilation, including not only variational approaches but older-style relaxation approaches which, due to the lack of fine-scale error covariance information, could be equally effective at the fine scales we are attempting to simulate. We also present results to date of both case studies mentioned above, including both qualitative and quantitative performance metrics (i.e. standard skill scores). These metrics will provide the basis for recommendations for the potential use of PAR data in the sense-and-avoid application.