On the Importance of Optimal Orography Resolution for Numerical Weather Prediction

Operational numerical weather prediction (NWP) systems are increasingly prioritizing finer spatial resolutions, often converging towards non-hydrostatic scales. The primary goal is to enhance flow representation at the smallest scales, particularly over complex terrain. Consequently, there is a growing research emphasis on accurately capturing the dynamic effects of the finer orographic scales in operational NWP by retaining scales in the resolved orography that approach the Nyquist limit.

The operational NWP systems at Environment and Climate Change Canada (ECCC) are based on the Global Environmental Multiscale (GEM) atmospheric model which has a semi-Lagrangian dynamical core along with a terrain-following vertical coordinate. In the past, similarly to other operational NWP centers, the orography for the various NWP systems was subjected to aggressive filtering. While eliminating variance at the Nyquist limit, these filters concurrently introduced non-negligible smoothing for scales as large as 20 times the grid resolution. However, in accordance with the shifting focus towards high-resolution forecasting, the orography filter for the GEM model has recently been substantially revised to make it sharper, with a reduced cut-off wavelength tailored to the smallest scales.

The most recent implementation of ECCC’s operational 15-km resolution global deterministic prediction system (GDPS) adopts an orography filter that preserves a minimum of 95% variance for orography scales as small as 3 times the grid resolution. However, upon the application of similar high-resolution orography within ECCC’s 10-km resolution continental-scale regional deterministic prediction system (RDPS), a pronounced tendency to underestimate freezing rain accumulations was uncovered within the valleys in and around the Canadian Rockies. Subsequent investigations have unveiled a discernible pattern of anomalous winter-time warming within these valleys. A comprehensive study based on the RDPS configuration – but with a reduced domain focused on the Canadian Rockies – identified the adiabatic and inviscid GEM dynamical core as the source of the spurious valley warming, which is quite pronounced during winter.

To investigate the underlying intricacies, a series of two-dimensional idealized experiments was carried out utilizing the GEM dynamical core. Results from these controlled experiments illustrated that discrepancies in flow representation at scales below an optimal resolution for the resolved orography can lead to secondary erroneous signals akin to winter-time valley warming. Furthermore, the two-dimensional experiments revealed that increased near-surface thermal stability plays a major role in exacerbating the erroneous flow response at the finest scales, thereby amplifying the warming signal. The implications of this investigation underscore the critical importance of appropriately filtering sub-optimal scales from the resolved orography for operational NWP systems to provide acceptable forecast guidance. The details of the study and the pathway for research-to-operation will be presented at the conference.